CN118797032A - A processing method, device and electronic equipment for intelligent simulated interview - Google Patents

Abstract

A processing method, device and electronic device for intelligent simulated interviews, relating to the field of data processing technology. In this method, a first interview question input by a first user is obtained; according to a preset knowledge network, multiple first pre-stored solutions for the first interview question are searched in a preset database; the selection habits of the second user pre-stored in the preset database are retrieved; according to the selection habits, if it is determined that the selection rate is equal to the preset selection rate, the total selection rate of the multiple first pre-stored solutions is screened to obtain multiple first processing solutions; the second processing solution is split and processed to obtain at least two processing steps; the processing steps are replaced by pre-stored associated steps to obtain a first interview answer for the first interview question, and the first interview answer is displayed to the first user so that the first user can simulate according to the first interview answer. The technical solution provided by the present application is implemented so that the output interview answer can better meet the needs of students to answer interview questions.

Description

A processing method, device and electronic equipment for intelligent simulated interview

Technical Field

The present application relates to the field of data processing technology, and specifically to a method, device and electronic equipment for processing intelligent simulated interviews.

Background Art

With the development of technology and the continuous changes in the market, companies are also constantly increasing their requirements for job seekers. Especially for college students who are new to the workplace, they generally lack practical experience, especially in the interview stage. Intelligent simulation interview technology came into being as an innovative means of education and training to gain an in-depth understanding of market dynamics, that is, the company's specific signature preferences and employment standards. Given that the intelligent simulation interview system covers a wide range of interview question banks across industries and multiple positions, students can take this opportunity to deeply search for the specific requirements of different companies and positions, and then accurately locate the direction of their personal career planning.

In the process of preparing to participate in intelligent simulation interviews, students often collect and organize potential interview question sets in advance based on the characteristics of the target position. However, faced with these questions, students may be confused due to lack of practical experience and find it difficult to conceive answers that both meet the job requirements and show their personal strengths. Existing intelligent platforms allow students to enter specific interview questions and automatically generate corresponding interview answers based on the questions, but the answers currently generated are relatively simple and vague, and cannot give detailed and specific responses, which does not meet the needs of students to answer interview questions.

Therefore, there is an urgent need for an intelligent simulated interview processing method, device and electronic equipment that can solve the above-mentioned technical problems.

Summary of the invention

The present application provides a method, device and electronic device for processing an intelligent simulated interview. The method searches for multiple pre-stored solutions from a preset database based on a first interview question sent by a first user, and screens, splits and replaces the pre-stored solutions, so that the output interview answers can better meet the students' needs for answering interview questions and improve the efficiency of answering.

In a first aspect, the present application provides a method for processing an intelligent simulated interview, the method comprising: obtaining a first interview question input by a first user; searching a preset database for multiple first pre-stored solutions for the first interview question according to a preset knowledge network; retrieving the selection habits of a second user pre-stored in the preset database, the selection habits including the selection rate of the second user for the pre-stored solutions, the selection rate being the ratio of the number of times the second user selects the pre-stored solutions to the number of displayed pre-stored solutions; if it is determined that the selection rate is lower than the preset selection rate according to the selection habits, then screening the multiple first pre-stored solutions for the total selection rate to obtain multiple first processing solutions, the first processing solution being the first pre-stored solution with a total selection rate higher than the preset total selection rate among the multiple first pre-stored solutions, the total selection rate being the ratio of the total number of selected solutions corresponding to the first pre-stored solutions to the total number of displayed solutions; splitting the second processing solution to obtain at least two processing steps, the second processing solution being any one of the multiple first processing solutions; replacing the processing steps with the pre-stored associated steps to obtain a first interview answer for the first interview question, and displaying the first interview answer to the first user so that the first user can simulate according to the first interview answer.

By adopting the above technical solution, after searching for multiple pre-stored solutions in the preset database according to the first interview question of the first user, the selection habits of the second user are further analyzed. According to the selection rate of the second user for the pre-stored solutions, it is judged that the second user rarely chooses the pre-stored solution and thus adopts the pre-stored solution. Then, the pre-stored solutions are directly screened, split and replaced in sequence, so that the final interview answer can better meet the needs of the first user. Pre-stored solutions with a higher total selection rate are screened from multiple pre-stored solutions, that is, pre-stored solutions that are popular among the user group and have a higher click rate are screened, thereby avoiding the need to recommend unpopular or irrelevant content, improving the quality and relevance of the answer, and then the first processing solution obtained by screening is split into multiple steps, and finally the original processing step is replaced by the associated step. Because the preset database performs multiple processing on the given answer according to the selection habits of the second user, the first user does not need to adjust and limit the answer again and again, thereby improving the efficiency of the answer.

Optionally, before replacing the processing step with a pre-stored association step to obtain a first interview answer to the first interview question, the method also includes: determining a second interview question corresponding to the first processing step based on a preset knowledge network, the first processing step being any one of multiple processing steps; searching a preset database for multiple second pre-stored solutions for the second interview question based on the preset knowledge network; sorting the total selection rates of the multiple second pre-stored solutions in descending order to obtain a sorting result; obtaining a target selection rate corresponding to the last position from the sorting result, and determining a second interview answer corresponding to the target selection rate; and determining the second interview answer to be a second processing step, the second processing step being an association step corresponding to the first processing step among multiple association steps.

By adopting the above technical solution, it is improved to determine that the first processing step corresponds to the second interview question according to the preset knowledge network, and then search out multiple second pre-stored solutions for the second interview question. Because they are all solutions to the same problem, the second pre-stored solutions found have a higher correlation with the second processing solution. The target selection rate corresponding to the last position in the sorting result can be screened out to obtain the second predicted solution, and the second pre-stored solution can be split for subsequent replacement, so that the reply results can have a high correlation with the original reply and are also some relatively novel solutions.

Optionally, the second processing scheme is split and processed to obtain at least two processing steps, specifically including: extracting key features of the second processing scheme to obtain multiple noun phrase key features and multiple verb key features; identifying the sequential logical structure of the second processing scheme to determine multiple sequential logical connectives included in the second processing scheme, and the sequential logical connectives are used to establish sequential logical relationships between sentences or paragraphs in the interview answers; determining a first key feature and a second key feature between the first logical connective and the second logical connective, the first logical connective and the second logical connective being any two adjacent sequential logical connectives among multiple sequential connectives, the first key feature being any one of multiple noun key features, and the second key feature being any one of multiple verb key features; combining the first key feature and the second key feature to obtain the first processing step.

By adopting the above technical scheme, by extracting multiple noun phrase key features and multiple verb key features, it is possible to capture important concepts and key actions in the interview answers, which is helpful to extract key information from large texts. By identifying the sequential logical connectives contained in the second processing scheme, the logical relationship between sentences or paragraphs is determined, which makes the structure of the interview answers clearer and makes it easier to understand the order and correlation between the steps. Finally, through the combination of key features between logical connectives, the generation of processing steps is realized.

Optionally, before replacing the processing step with the pre-stored association step to obtain the first interview answer to the first interview question, the method further includes: determining the second interview question corresponding to the first processing step according to the preset knowledge network, the first processing step being any one of the multiple processing steps; searching for multiple second pre-stored solutions for the second interview question in the preset database according to the preset knowledge network; performing feature extraction on the first processing step to obtain multiple first features, performing feature extraction on the target pre-stored solution to obtain multiple second features, the target pre-stored solution being any one of the multiple second pre-stored solutions;

According to the multiple first features and the multiple second features, the correlation between the first processing step and the target pre-stored solution is calculated. The specific calculation formula is as follows: ; Wherein D(X, Y) is the correlation between the first processing step X and the target pre-stored solution Y, df(u) is the characteristic frequency of the u-th first feature, df(v) is the characteristic frequency of the v-th second feature, argmin(d(u,v)×P _uv ) is the traffic allocation matrix, P _uv is the traffic allocation from the u-th first feature to the v-th second feature, that is, the amount of traffic allocated by moving the u-th first feature to the v-th second feature, and d(u,v) is the distance between the characteristic vector of the first feature and the characteristic vector of the second feature; if it is determined that the correlation is greater than or equal to the preset threshold, the target pre-stored solution is determined to be an association step.

By adopting the above technical solution, by determining the second interview question corresponding to the first processing step in the preset knowledge network, it is ensured that the association step is relevant to the original processing step in the knowledge field, ensuring that the replaced solution is more in line with the interview needs of the first user. By calculating the formula for the correlation, the frequency of the feature, the flow distribution and the distance between the feature vectors are comprehensively considered, making the correlation calculation more comprehensive and accurate. If it is determined that the correlation is greater than or equal to the preset threshold, the target pre-stored solution is determined to be the association step. This mechanism ensures that the selection of the replacement solution is more reliable, and replacement is only performed when the correlation is high, avoiding irrelevant or low-correlation replacements, thereby improving the accuracy of the response.

Optionally, the characteristic frequency of the first characteristic is calculated by the following formula: ; Wherein, df(u) is the characteristic frequency of the u-th first feature, _mu is the number of pre-stored solutions containing the first feature, and _Mu is the total number of pre-stored solutions;

The characteristic frequency of the second characteristic is calculated by the following formula: ; Where df(v) is the characteristic frequency of the vth second feature, m _v is the number of pre-stored solutions containing the second feature, and M _v is the total number of pre-stored solutions; the distance between the characteristic vector of the first feature and the characteristic vector of the second feature is calculated by the following formula: ; Wherein, d(u,v) is the distance between the eigenvector of the first feature and the eigenvector of the second feature, _uk is the eigenvector of the first feature in the kth dimension, and _vk is the eigenvector of the second feature in the kth dimension.

Optionally, before retrieving the selection habits of the second user pre-stored in a preset database, the method also includes: recording a first number of pre-stored solutions for the interview question retrieved after the first user inputs the interview question; recording a second number of pre-stored solutions selected by the second user without any processing; and obtaining a selection rate based on a ratio of the second number to the first number.

By adopting the above technical solution, by calculating the selection rate, the first user's attention to the pre-stored solution can be evaluated. A higher selection rate indicates that the second user is more likely to be interested in the provided solution, while a lower selection rate may indicate that the second user is less interested in the solution. Recording the selection rate of the second user helps to more intelligently adjust the weight of the pre-stored solution. In the subsequent processing steps, relevant content can be recommended more preferentially based on the solution with a high selection rate, thereby improving the first user's satisfaction.

Optionally, searching a preset database for multiple first pre-stored solutions for a first interview question specifically includes: determining a similarity between the first interview question and multiple pre-stored interview questions in the preset database; determining, among the multiple pre-stored interview questions, pre-stored interview questions whose similarity is greater than or equal to a preset similarity threshold; and determining multiple first pre-stored solutions corresponding to the pre-stored interview questions whose respective similarities are greater than or equal to the preset similarity threshold.

By adopting the above technical solution, by calculating the similarity between the first interview question and the pre-stored interview questions, it is possible to accurately match questions similar to those asked by the first user. This helps to improve the relevance of the searched pre-stored solutions to the actual questions of the first user. Setting a similarity threshold can filter out pre-stored interview questions with low similarity, ensuring that only questions that are highly relevant to the first user's questions are retained. This can effectively reduce noise interference and improve the quality of retrieval results. After determining the pre-stored interview questions whose similarity is greater than or equal to the preset similarity threshold, multiple first pre-stored solutions corresponding to these questions can be obtained, thereby providing the first user with a variety of solution options and increasing the comprehensiveness of the answers.

In a second aspect of the present application, a processing device for an intelligent simulated interview is provided, the device comprising an acquisition unit, a processing unit and a splitting unit; the acquisition unit acquires a first interview question input by a first user; the processing unit searches for a plurality of first pre-stored solutions for the first interview question in a preset database according to a preset knowledge network; retrieves the selection habits of the second user pre-stored in the preset database, the selection habits including the selection rate of the second user for the pre-stored solutions, the selection rate being the ratio of the number of times the second user selects the pre-stored solutions to the number of displayed pre-stored solutions; if it is determined that the selection rate is lower than the preset selection rate according to the selection habits, the total selection rate of the plurality of first pre-stored solutions is screened to obtain a plurality of first processing solutions, the first processing solution being the first pre-stored solution with a total selection rate higher than the preset total selection rate among the plurality of first pre-stored solutions, the total selection rate being the ratio of the total number of selected solutions corresponding to the first pre-stored solutions to the total number of displayed solutions; the splitting unit splits the second processing solution to obtain at least two processing steps, the second processing solution being any one of the plurality of first processing solutions; replaces the processing step with the pre-stored associated step to obtain a first interview answer for the first interview question, and displays the first interview answer to the first user so that the first user can simulate according to the first interview answer.

Optionally, the processing unit is used to determine, based on a preset knowledge network, a second interview question corresponding to the first processing step, where the first processing step is any one of multiple processing steps; based on the preset knowledge network, search a preset database for multiple second pre-stored solutions for the second interview question; sort the total selection rates of the multiple second pre-stored solutions in descending order to obtain a sorting result; the acquisition unit is used to obtain the target selection rate corresponding to the last position from the sorting result, and determine the second interview answer corresponding to the target selection rate; the processing unit is used to determine that the second interview answer is a second processing step, where the second processing step is an associated step corresponding to the first processing step among multiple associated steps.

Optionally, the processing unit is used to extract key features of the second processing scheme to obtain multiple noun phrase key features and multiple verb key features; perform sequential logical structure recognition on the second processing scheme to determine multiple sequential logical connectives contained in the second processing scheme, and the sequential logical connectives are used to establish sequential logical relationships between sentences or paragraphs in the interview answers; determine a first key feature and a second key feature between the first logical connective and the second logical connective, the first logical connective and the second logical connective are any two adjacent sequential logical connectives among multiple sequential connectives, the first key feature is any one of the multiple noun key features, and the second key feature is any one of the multiple verb key features; combine the first key feature and the second key feature to obtain a first processing step.

Optionally, the processing unit is used to determine the second interview question corresponding to the first processing step according to a preset knowledge network, where the first processing step is any one of the multiple processing steps; search for multiple second pre-stored solutions for the second interview question in a preset database according to the preset knowledge network; perform feature extraction on the first processing step to obtain multiple first features, perform feature extraction on the target pre-stored solution to obtain multiple second features, where the target pre-stored solution is any one of the multiple second pre-stored solutions; calculate the correlation between the first processing step and the target pre-stored solution according to the multiple first features and the multiple second features, and the specific calculation formula is as follows: ; Wherein D(X, Y) is the correlation between the first processing step X and the target pre-stored solution Y, df(u) is the characteristic frequency of the u-th first feature, df(v) is the characteristic frequency of the v-th second feature, argmin(d(u,v)×P _uv ) is the traffic allocation matrix, P _uv is the traffic allocation from the u-th first feature to the v-th second feature, that is, the amount of traffic allocated by moving the u-th first feature to the v-th second feature, and d(u,v) is the distance between the characteristic vector of the first feature and the characteristic vector of the second feature; if it is determined that the correlation is greater than or equal to the preset threshold, the target pre-stored solution is determined to be an association step.

Optionally, the processing unit is used to calculate the characteristic frequency of the first characteristic by the following formula: ; Where df(u) is the characteristic frequency of the u-th first feature, _mu is the number of pre-stored solutions containing the first feature, and _Mu is the total number of pre-stored solutions; the characteristic frequency of the second feature is calculated by the following formula: ; Where df(v) is the characteristic frequency of the vth second feature, m _v is the number of pre-stored solutions containing the second feature, and M _v is the total number of pre-stored solutions; the distance between the characteristic vector of the first feature and the characteristic vector of the second feature is calculated by the following formula: ; Wherein, d(u,v) is the distance between the eigenvector of the first feature and the eigenvector of the second feature, _uk is the eigenvector of the first feature in the kth dimension, and _vk is the eigenvector of the second feature in the kth dimension.

Optionally, the processing unit is used to record a first number of pre-stored solutions for the interview question retrieved after the first user inputs the interview question; record a second number of pre-stored solutions selected by the second user without any processing; and obtain a selection rate based on a ratio of the second number to the first number.

Optionally, the processing unit is used to determine the similarity between the first interview question and multiple pre-stored interview questions in a preset database; determine the pre-stored interview questions among the multiple pre-stored interview questions whose similarity is greater than or equal to a preset similarity threshold; and determine multiple first pre-stored solutions corresponding to the pre-stored interview questions whose respective similarities are greater than or equal to the preset similarity threshold.

In a third aspect of the present application, an electronic device is provided, which includes a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory, so that an electronic device executes any one of the methods described above in the present application.

In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and when the instructions are executed, any one of the above methods of the present application is executed.

In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

1. After searching for multiple pre-stored solutions in the preset database according to the first interview question of the first user, further analyze the selection habits of the second user. According to the selection rate of the second user for the pre-stored solutions, it is judged that the first user rarely chooses the pre-stored solution and thus adopts the pre-stored solution. Then, the pre-stored solutions are directly screened, split and replaced in sequence, so that the final interview answer can better meet the needs of the first user. Pre-stored solutions with a high total selection rate are screened out from multiple pre-stored solutions, that is, pre-stored solutions that are popular among the user group and have a high click rate are screened out, thereby avoiding the recommendation of unpopular or irrelevant content, improving the quality and relevance of the answer, and then the first processing solution obtained by screening is split into multiple steps, and finally the original processing step is replaced by the associated step. Since the preset database performs multiple processing on the given answer according to the selection habits of the second user, the first user does not need to adjust and limit the answer again and again, thereby improving the efficiency of the answer.

2. Improve the ability to determine that the first processing step corresponds to the second interview question based on the preset knowledge network, and then search out multiple second pre-stored solutions for the second interview question. Because they are all solutions to the same problem, the second pre-stored solutions found have a higher correlation with the second processing solution. The target selection rate corresponding to the last position in the sorting result can be screened out to obtain the second predicted solution, and the second pre-stored solution can be split for subsequent replacement, so that the response results can have a high correlation with the original response and are also some relatively novel solutions.

3. By extracting multiple noun phrase key features and multiple verb key features, it is possible to capture important concepts and key actions in interview answers, which helps to extract key information from large texts. By identifying the sequential logical connectives contained in the second processing scheme, the logical relationship between sentences or paragraphs is determined, which makes the structure of the interview answers clearer and makes it easier to understand the order and correlation between the various steps. Finally, through the combination of key features between logical connectives, the generation of processing steps is achieved.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a flow chart of a method for processing an intelligent simulated interview provided in an embodiment of the present application;

FIG2 is a schematic diagram of the structure of a processing device for an intelligent simulated interview provided in an embodiment of the present application;

FIG. 3 is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present application.

Explanation of the reference numerals: 201, acquisition unit; 202, processing unit; 203, splitting unit; 300, electronic device; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory.

DETAILED DESCRIPTION

In order to enable technicians in this field to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

In the description of the embodiments of the present application, words such as "for example" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "for example" or "for example" is intended to present related concepts in a specific way.

In the description of the embodiments of the present application, the meaning of the term "multiple" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprise", "have" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

With the development of technology and the continuous changes in the market, companies are also constantly increasing their requirements for job seekers. Especially for college students who are new to the workplace, they generally lack practical experience, especially in the interview stage. Intelligent simulation interview technology came into being as an innovative means of education and training to gain an in-depth understanding of market dynamics, that is, the company's specific signature preferences and employment standards. Given that the intelligent simulation interview system covers a wide range of interview question banks across industries and multiple positions, students can take this opportunity to deeply search for the specific requirements of different companies and positions, and then accurately locate the direction of their personal career planning.

In the process of preparing to participate in intelligent simulation interviews, students often collect and organize potential interview question sets in advance based on the characteristics of the target position. However, faced with these questions, students may be confused due to lack of practical experience and find it difficult to conceive answers that both meet the job requirements and show their personal strengths. Existing intelligent platforms allow students to enter specific interview questions and automatically generate corresponding interview answers based on the questions, but the answers currently generated are relatively simple and vague, and cannot give detailed and specific responses, which does not meet the needs of students to answer interview questions.

Therefore, how to change the answer content generated by the existing intelligent platform to not meet the needs of students' interview questions is a problem that needs to be solved urgently. A method for processing an intelligent simulated interview provided in an embodiment of the present application is applied to a server. The server of the present application can be a platform that provides interview question answering services for job seekers. Figure 1 is a flow chart of a method for processing an intelligent simulated interview provided in an embodiment of the present application. Referring to Figure 1, the method includes the following steps S101-step S106.

S101: Obtain a first interview question input by a first user.

In the above S101, the server receives a first interview question input by a first user, where the first user refers to a college student seeking a job, and can also be understood as a job seeker who sends the first interview question to the server. The server includes, but is not limited to, electronic devices such as mobile phones, tablet computers, wearable devices, PCs (Personal Computers), and can also be a background server running a language model, where a language model refers to a model specifically for answering or responding to interview questions input by job seekers. The server can be implemented as an independent server or a server cluster consisting of multiple servers.

First, the first user needs to input his/her interview question to the server, for example, it can be: How to complete XX task? According to the input of the first user, the server obtains the interview question that the first user needs to get an answer to, that is, the first interview question.

S102: Searching a preset database for a plurality of first pre-stored solutions for a first interview question according to a preset knowledge network.

In the above S102, during the training phase of the language model, a large-scale text corpus is usually collected as a training set and a validation set. The training set is used for model training, and the validation set is used to evaluate model performance. When the language model is applied to the question-answering system, it is necessary to use data such as FAQ data sets, large knowledge bases, and manually constructed question-answering data to train the language model. In the training set, each sample usually includes a question and a corresponding answer. The task of the model is to learn to infer the correct answer from the question. These data sets usually annotate the correct answer to each question so that the model loss can be calculated and the parameters can be updated during training. Using such a training set, the language model can learn the ability to understand and generate natural language questions and answers.

Further, the data related to the language model is stored in a preset database connected to the server, and the questions and corresponding answers are used to construct a preset knowledge network. After the first user uses the language model to ask and answer questions and inputs the first interview question, the server uses an appropriate text similarity calculation method, such as cosine similarity, Jaccard similarity, edit distance, etc., for each pre-stored interview question in the preset database to calculate their similarity with the first interview question. These methods can convert text into vector space representation and then compare the similarity between vectors. A preset similarity threshold is set in advance, and the preset similarity threshold can be adjusted based on actual scenario requirements. The selection of the preset similarity threshold depends on the extent to which the two interview questions are expected to be similar. For the calculated similarity of each pre-stored interview question, the questions with a similarity greater than or equal to the preset threshold are screened out, so that a group of pre-stored interview questions with a high similarity to the first interview question are obtained. For each pre-stored interview question screened by similarity, the corresponding one or more first pre-stored solutions are obtained. This can be achieved by retrieving the corresponding interview question record in the preset database to ensure that the correct pre-stored solution is associated.

By calculating the similarity between the first interview question and the pre-stored interview questions, questions similar to those asked by the first user can be accurately matched. This helps to improve the relevance of the searched pre-stored solutions to the first user's actual questions. Setting a similarity threshold can filter out pre-stored interview questions with low similarity, ensuring that only questions that are highly relevant to the first user's questions are retained. This can effectively reduce noise interference and improve the quality of search results. After determining the pre-stored interview questions with a similarity greater than or equal to the preset similarity threshold, multiple first pre-stored solutions corresponding to these questions can be obtained, thereby providing the first user with a variety of solution options and increasing the comprehensiveness of the answers.

S103: Retrieve the selection habits of the second user pre-stored in the preset database.

In the above S103, the selection habit includes the selection rate of the second user for the pre-stored solutions, and the selection rate is the ratio of the number of times the second user selects the pre-stored solution to the number of displayed pre-stored solutions. The public selection habit is preferably the selection rate of the second user for the pre-stored solutions obtained by the server. At this time, the second user refers to other users who send the same interview questions as the first user, and other users can represent other job-seeking college students. That is, when the second user enters the first interview question, the server will provide multiple pre-stored solutions in multiple preset databases based on the preset knowledge network. It can be understood here that these pre-stored solutions are basically answers with a high accuracy rate but are relatively conventional. In theory, the answers can be directly retrieved from the search engine based on the interview questions. For example, for the first interview question "How to complete XX task?" The corresponding first pre-stored solution may be "First, the work arrangement and time planning should be carried out for the XX task. Then, start executing the corresponding task according to the work plan. Finally, after completing the task, check the task and send it to the corresponding person."

However, for some second users, they need the language model to give more complex solutions. After the server gives multiple first pre-stored solutions, they will not directly click on the first pre-stored solution to use it, but will ask further questions. In the above example, for the interview question "How to complete XX task?", the second user may not be satisfied with the several solutions given directly. He will further ask, "Please give a more detailed introduction on how to arrange time reasonably to quickly complete XX task?" This leads to a lower selection rate of pre-stored solutions. The selection rate is the ratio of the number of times the second user selects a pre-stored solution to the number of pre-stored solutions displayed. Therefore, the selection rate of the pre-stored solution can be used to judge the second user's satisfaction with the pre-stored solution. When the selection rate is low, it indicates that the second user is usually not satisfied with the pre-stored solution and needs to further improve the pre-stored solution.

Therefore, a certain user behavior tracking and statistical mechanism is first required to record the selection behavior of the second user and analyze the selection habits of the second user. When the second user enters the interview question and submits it, the server records the number of pre-stored solutions retrieved for the interview question. This can be a simple counting operation, recording the number of pre-stored solutions retrieved by the server and marking it as the first number. When the second user selects a pre-stored solution without any processing, the server records the number of pre-stored solutions selected by the second user and marks it as the second number. This can be achieved by monitoring the selection events of the second user, and each selection event is recorded as a count. Using the recorded first number and second number, the selection rate is calculated, and the selection rate is usually obtained by dividing the second number by the first number.

By calculating the selection rate, the second user's interest in the pre-stored solutions can be evaluated. A higher selection rate indicates that the second user is more likely to be interested in the provided solutions, while a lower selection rate may indicate that the second user is less interested in the solutions. Recording the second user's selection rate helps to more intelligently adjust the weights of the pre-stored solutions. In subsequent processing steps, relevant content can be recommended with higher priority based on solutions with high selection rates, thereby improving user satisfaction.

A selection rate threshold, i.e., a preset selection rate, can be pre-set to determine whether the second user's selection behavior meets expectations. The choice of the preset selection rate depends on the specific application scenario. For example, the preset selection rate can be set to 0.1, indicating that when the second user's selection rate is higher than 10%, the server believes that the second user is more satisfied with the pre-stored solution. Based on the second user's selection habit, if the first user subsequently inputs the same interview question as the second user, the server can directly provide the pre-stored solution in the preset database based on the preset knowledge network. If the first user is not a first-time user, the first user's selection habit can be determined based on the first user's historical selection rate. When the first user subsequently inputs a new interview question, the server directly provides the pre-stored solution in the preset database based on the preset knowledge network.

S104: If it is determined according to the selection habit that the selection rate is equal to the preset selection rate, a total selection rate screening is performed on a plurality of first pre-stored solutions to obtain a plurality of first processing solutions.

In the above S104, if the selection rate of the second user for the pre-stored solutions is lower than the preset selection rate according to the pre-stored selection habits of the second user, that is, the second user may have made further inquiries after obtaining the first pre-stored solutions, resulting in the failure to select these pre-stored solutions. Therefore, the server needs to further process the first pre-stored solutions so that the final answer can better meet the needs of the first user.

First, the server determines the total selection rate of each first pre-stored solution, which is the ratio of the total number of selections of the first pre-stored solution to the total number of displays. It should be noted that for a certain first pre-stored solution, the total number of displays mentioned here can be for different questions of different users. Each time the first pre-stored solution is displayed, it will be counted once, and the total number of displays will be obtained. Similarly, any first pre-stored solution will be recorded once when it is selected by any user, and the total number of selections will be obtained. Finally, the ratio of the total number of selections to the total number of displays is the total selection rate. For example, for the first pre-stored solution A, according to the preset knowledge network, after multiple users raised different questions, the first pre-stored solution A was displayed a total of 200 times. And different users selected it a total of 85 times, so the total selection rate of the first pre-stored solution A is 0.425.

Then, according to the total selection rate, multiple first pre-stored solutions are screened, and the first pre-stored solutions with lower total selection rates are filtered out, and the first pre-stored solutions with higher total selection rates than the preset total selection rates are retained, thereby obtaining multiple first processing solutions. In this embodiment, the preset total selection rate can be adjusted according to actual conditions, and this application does not make specific limitations. When the total selection rate of the first pre-stored solution is higher than the preset total selection rate, it can be understood that the adoption rate of the first pre-stored solution is higher and has more real-time application significance.

S105: Split the second processing solution into two steps.

In the above S105, after obtaining multiple first processing solutions, the server needs to further process each first processing solution in order to make it more novel and meet the needs of the first user. Take any one of the multiple first processing solutions and the second processing solution as an example for illustration.

The second processing scheme is subjected to natural language processing to extract noun phrase key features and verb key features. This can be achieved through technologies such as part-of-speech tagging and entity recognition. Noun phrase key features are usually important noun phrases involved in the scheme, while verb key features are usually verbs related to key actions or operations in the scheme. For example, for the second processing scheme, "First, install and configure a web server. Then, set up the database and create the corresponding table structure. Next, write server-side code to handle user requests and database interactions. Finally, create a front-end page to present blog content." The noun key features it contains include web server, database, table structure, server-side code, user request, database, front-end page, and blog content. The verb key features it contains include installation, configuration, setting, creation, writing, processing, interaction, and presentation.

By using natural language processing technology, multiple sequential logical connectives included in the second processing scheme are identified. Logical connectives are usually used to establish sequential logical relationships between sentences or paragraphs. Sequential logical connectives help guide the finding of sequential relationships in texts, such as first, second, next, then, and finally. The identification of sequential logical connectives can be achieved through a pre-defined connective list or by using a language model.

Then, the second processing scheme is split according to multiple sequential logical connectives, and any two adjacent sequential logical connectives, the first logical connective and the second logical connective, are used as examples for explanation. The server determines the first key feature and the second key feature between the first logical connective and the second logical connective, the first key feature being any one of multiple noun key features, and the second key feature being any one of multiple verb key features. In the above example second processing scheme "First, work arrangements and time planning should be carried out for the xx task. Then, the corresponding tasks should be executed according to the work plan. Finally, after the tasks are completed, the tasks should be checked and sent to the corresponding personnel.", if the first logical connective is then and the second logical connective is finally, then the first key feature is work planning and execution, and the second key feature is checking and sending.

Combine the noun phrase key features and verb key features between the first logical connective and the second logical connective. The combined key features obtained in this way can be regarded as a description of a logical relationship, that is, the processing steps are obtained. According to the above method, for the second processing scheme "First, the work arrangement and time planning should be carried out for the xx task. Then, the corresponding task should be executed according to the work plan. Finally, after the task is completed, the task should be checked and sent to the corresponding personnel." will be divided into the following multiple processing steps: work arrangement, time planning, work plan, task execution, check of completed tasks, and sending to the corresponding personnel.

Next, the server determines the associated steps of each processing step, and replaces the processing step with the associated steps. The associated steps are interview answers pre-stored in the preset database. Therefore, it is necessary to select appropriate associated steps from the database to replace the processing steps. As mentioned earlier, the preset database stores a preset knowledge network of questions and corresponding answers. Each processing step, as an answer, will also include a corresponding question. For example, for any one of the multiple processing steps, the first processing step is "work arrangements and time planning should be carried out for task xx", and the corresponding question may be "How to make a detailed plan for task xx?" Then, the second interview question corresponding to the first processing step can be determined based on the preset knowledge network.

By extracting multiple noun phrase key features and multiple verb key features, it is possible to capture important concepts and key actions in the interview answers. This helps to extract key information from large texts. By identifying the sequential logical connectives included in the second processing scheme, the logical relationship between sentences or paragraphs is determined. This makes the structure of the target answer clearer and the model can more easily understand the order and correlation between the various steps. Finally, the generation of processing steps is achieved by combining the key features between the logical connectives.

Similarly, according to the preset knowledge network, according to the determined second interview question, multiple second pre-stored solutions related to it are retrieved from the preset database. Because there may be multiple different answers to the same question, and thus, for the second interview question, the preset database may have multiple second pre-stored solutions to answer the second interview question. For example, for the second interview question "How to make a detailed plan for the xx task?", the second pre-stored solution may be "split the xx task into work tasks, and then reasonably arrange the time plan for each work task", or it may be "determine the delivery cycle of the xx task, split the xx task according to the delivery deadline, so as to complete the xx task within the delivery cycle", or it may be "determine the number of people who will complete the xx task, split the xx task according to the number of people, and after splitting, reasonably arrange different work tasks according to the individual's expertise."

For each second pre-stored solution, calculate its total selection rate, that is, the ratio of the total number of selections to the total number of displays. In step S130, the total selection rate calculation method is mentioned, which will not be further described here. The server then sorts the multiple second pre-stored solutions in order of total selection rate to obtain a sorting result; obtains the target selection rate corresponding to the last position from the sorting result, and determines the second interview answer corresponding to the target selection rate; the second interview answer is the solution with the smallest total selection rate in the sorting result. The smallest total selection rate may indicate relatively few choices. The server can judge that the novelty of the solution is higher based on this metric, and thus choose to proceed to the next step of processing. Finally, the server determines the second target answer as the second processing step, and the second processing step is an associated step corresponding to the first processing step among multiple associated steps.

By determining the second interview question corresponding to the first processing step according to the preset knowledge network, and then retrieving multiple second pre-stored solutions for the second interview question, since they are all solutions for the same question, the second pre-stored solution found has a higher correlation with the second processing solution. Furthermore, the second pre-stored solution with the smallest total selection rate is screened out, and the second pre-stored solution is split for subsequent replacement. In this way, the reply result can have a high correlation with the original reply and also be some relatively novel solutions.

Furthermore, it is also possible to select a suitable pre-stored solution to replace the processing step according to the correlation between the processing step and the pre-stored solution. Similarly, according to the preset knowledge network, the server retrieves multiple second pre-stored solutions related to the second interview question from the preset database. Then the server extracts features from the first processing step and the second pre-stored solution. Feature extraction is performed on the first processing step to obtain multiple first features, and the first features are keywords and phrases of the content of the first processing step. Feature extraction is performed on the target pre-stored solution to obtain multiple second features, and the second features are keywords and phrases of the content of the target pre-stored solution. The target pre-stored solution is any second pre-stored solution among the multiple second pre-stored solutions.

Then, according to the multiple first features and the multiple second features, the correlation between the first processing step and the target pre-stored solution is calculated. The specific calculation formula is as follows: ; Where D(X, Y) is the correlation between the first processing step X and the target pre-stored solution Y, df(u) is the characteristic frequency of the u-th first feature, df(v) is the characteristic frequency of the v-th second feature, argmin(d(u,v)×P _uv ) is the traffic allocation matrix, P _uv is the traffic allocation from the u-th first feature to the v-th second feature, that is, the amount of traffic allocated by moving the u-th first feature to the v-th second feature, and d(u,v) is the distance between the characteristic vector of the first feature and the characteristic vector of the second feature;

By determining the second interview question corresponding to the first processing step in the preset knowledge network, it is ensured that the associated step is relevant to the original processing step in the knowledge field, ensuring that the replaced solution is more in line with user needs. The formula for calculating the correlation degree comprehensively considers the frequency of the feature, the flow distribution, and the distance between the feature vectors, making the correlation calculation more comprehensive and accurate. If it is determined that the correlation degree is greater than or equal to the preset threshold, the target pre-stored solution is determined to be the associated step. This mechanism ensures that the selection of replacement solutions is more reliable, and replacement is only performed when the correlation is high, avoiding irrelevant or low-correlation replacements, thereby improving the accuracy of the response.

In the above formula, feature frequency refers to the ratio of the number of documents containing a specific word to the total number of documents. Feature frequency is used to measure the prevalence or rarity of a feature. If a feature appears in most documents, its feature frequency will be high, indicating that this feature may be a relatively common and generalized feature. On the contrary, if a feature only appears in a few documents, its feature frequency will be low, indicating that this feature may be a relatively specific or rare feature. Using feature frequency as a weight can reduce the importance of features that appear in most documents, thereby better reflecting semantic information. Specifically, using feature frequency as a weight can reduce the discriminability of common features that appear in most documents to the text, thereby better reflecting semantic information. This is related to the general goal of information retrieval and text analysis, that is, to more accurately locate and weigh those words that appear in fewer documents but have higher information content in a document collection. For example, in a document collection, some common features such as "是", "的" or "作为" may appear in most documents. Because of their prevalence, their feature frequencies are high, but they usually contribute less to the discriminability and semantic information of the text.

The characteristic frequency of the first feature is the ratio of the number of pre-stored solutions containing the first feature to the total number of pre-stored solutions in the preset database. The characteristic frequency of the first feature is calculated by the following formula: ; Wherein, df(u) is the characteristic frequency of the u-th first feature, _mu is the number of pre-stored solutions containing the first feature, and _Mu is the total number of pre-stored solutions.

The characteristic frequency of the second feature is the ratio of the number of pre-stored solutions containing the second feature to the total number of pre-stored solutions in the preset database. The characteristic frequency of the second feature is calculated by the following formula: ; Wherein, df(v) is the characteristic frequency of the vth second feature, m _v is the number of pre-stored solutions containing the second feature, and M _v is the total number of pre-stored solutions;

In the correlation calculation formula, argmin(d(u,v)×P _uv ) is the flow distribution matrix, and P _uv is the flow distribution from the uth first feature to the vth second feature, that is, the amount of flow allocated from the uth first feature to the vth second feature.

This assigned quantity can be seen as a weight representing the semantic distance between words. The flow distribution matrix is obtained by solving the minimum cost maximum flow problem. The solution process of the flow distribution matrix ensures that the minimum cost flow path is found while satisfying the constraints. This path gives the optimal way to match each feature of the first processing step X with the corresponding step of the target pre-stored solution Y, and each P _uv value represents the corresponding weight. This weight can be understood as the semantic similarity from the uth first feature to the vth second feature. The smaller the cost, the higher the similarity.

The constraints are as follows:

The total flow of each first feature in the first processing step X cannot exceed the distribution of the first feature, that is: ; Wherein, P _uv is the flow distribution from the u-th first feature to the v-th second feature, and _eu is the distribution vector of the first feature in the first processing step X.

The total flow of each second feature in the target pre-stored step Y cannot exceed the distribution of the second feature, that is: ; Wherein, P _uv is the flow distribution from the u-th first feature to the v-th second feature, and f _v is the distribution vector of the second feature in the second processing step Y.

For every pair of the first and second features, the flow distribution must be non-negative, that is:

P _uv ≥ 0;

For each first and second characteristic, the flow distribution must satisfy the total outflow equal to the total inflow, that is: ; Finally, in the correlation calculation formula, the distance between feature vectors is used to measure the semantic similarity between features, because the above correlation formula is to find the best feature movement scheme by minimizing the total cost, where the cost is determined by the distance between each pair of features. In the optimization process of the minimum cost maximum flow problem, the distance between feature vectors is used to calculate the cost of each pair of features in the traffic allocation matrix. The goal of optimization is to minimize the total cost by reasonably allocating traffic. In this way, the model can capture the relative position of the features in the semantic space, and then measure the similarity between the first processing step and the target pre-stored scheme. The distance between the feature vector of the first feature and the feature vector of the second feature is calculated by the following formula: ; Wherein, d(u,v) is the distance between the eigenvector of the first feature and the eigenvector of the second feature, _uk is the eigenvector of the first feature in the kth dimension, and _vk is the eigenvector of the second feature in the kth dimension.

Finally, the correlation between the first processing step and the target pre-stored solution is compared with a preset threshold set in advance. If the correlation is less than the preset threshold, it indicates that the first processing step has a low similarity with the target pre-stored solution, and the target pre-stored solution cannot be used to replace the first processing step. On the contrary, if the correlation is greater than or equal to the preset threshold, it indicates that the first processing step has a low similarity with the target pre-stored solution, and the target pre-stored solution can be used to replace the first processing step, and then the target pre-stored solution is determined to be the associated step.

S106: Using a pre-stored association step to replace the processing step, obtaining a first interview answer for the first interview question, and displaying the first interview answer to the first user so that the first user can perform simulation according to the first interview answer.

In the above S106, after splitting the second processing scheme into multiple processing steps, the server then searches for the replacement pre-stored scheme for each processing step to obtain the associated steps. Finally, the associated steps corresponding to each processing step are used to replace the processing steps, and a more complex but complete scheme can be obtained, that is, the first interview answer to the first interview question. For example, how to make a detailed plan for the first interview question? The first pre-stored scheme obtained may be that the work arrangement and time planning should be carried out for the xx task. If the first pre-stored scheme is used as the second processing scheme in the future, it will be split into two processing steps of "arranging the work for the xx task" and "planning the time according to the work arrangement". The associated step for "arranging the work for the xx task" may be "splitting the work for the xx task", and the associated step for "planning the time according to the work arrangement" may be "reasonably arranging the time according to the split work tasks". After the replacement, the first interview answer obtained will be "splitting the work for the XX task, and then reasonably arranging the time planning according to the split work tasks". This scheme is used to solve the interview question that the work arrangement and time planning should be carried out for the xx task. The first interview answer is then displayed to the first user so that the first user can simulate according to the first interview answer, so as to improve the professionalism of the first user's response to the interview question, thereby making the final first interview answer meet the needs of the first user.

By adopting the above technical solution, after retrieving multiple pre-stored solutions in the preset database according to the interview questions of the first user, the selection habits of the second user are further analyzed. If it is judged that the second user rarely chooses the pre-stored solution and adopts the pre-stored solution based on the selection rate of the second user for the pre-stored solution, the pre-stored solution is directly screened, split and replaced in sequence, so that the final interview answer can better meet the needs of the first user. Pre-stored solutions with a higher total selection rate are screened from multiple pre-stored solutions, that is, pre-stored solutions with a higher selection rate that are popular among the second user group are screened, thereby avoiding the recommendation of relatively unpopular or irrelevant content, and improving the quality and relevance of the answer. Then, the first processing solution obtained by screening is split into multiple steps, and finally the original processing step is replaced by the associated step. Since the model directly performs multiple processing on the given answer according to the user's selection habits, without the user having to condition the answer again and again, the efficiency of the language model answer is improved.

An embodiment of the present application also provides a processing device for an intelligent simulated interview. FIG2 is a structural diagram of a processing device for an intelligent simulated interview provided in an embodiment of the present application. Referring to FIG2 , the device includes an acquisition unit 201, a processing unit 202, and a splitting unit 203.

An acquiring unit 201 acquires a first interview question input by a first user.

The processing unit 202 searches for multiple first pre-stored solutions for the first interview question in a preset database according to a preset knowledge network; retrieves the selection habits of the second user pre-stored in the preset database, the selection habits including the selection rate of the second user for the pre-stored solutions, and the selection rate is the ratio of the number of times the second user selects the pre-stored solutions to the number of displayed pre-stored solutions; if it is determined that the selection rate is lower than the preset selection rate according to the selection habits, the multiple first pre-stored solutions are screened for the total selection rate to obtain multiple first processing solutions, the first processing solution being a first pre-stored solution with a total selection rate higher than the preset total selection rate among the multiple first pre-stored solutions, and the total selection rate is the ratio of the total number of selected corresponding first pre-stored solutions to the total number of displayed solutions.

The splitting unit 203 splits the second processing scheme to obtain at least two processing steps, where the second processing scheme is any one of the multiple first processing schemes; the processing steps are replaced by pre-stored associated steps to obtain a first interview answer to the first interview question, and the first interview answer is displayed to the first user so that the first user can simulate according to the first interview answer.

In a possible implementation, the processing unit 202 is used to determine the second interview question corresponding to the first processing step according to a preset knowledge network, where the first processing step is any one of the multiple processing steps; based on the preset knowledge network, search the preset database for multiple second pre-stored solutions for the second interview question; sort the total selection rates of the multiple second pre-stored solutions in descending order to obtain a sorting result; the acquisition unit 201 is used to obtain the target selection rate corresponding to the last position from the sorting result, and determine the second interview answer corresponding to the target selection rate; the processing unit 202 is used to determine that the second interview answer is the second processing step, and the second processing step is an associated step corresponding to the first processing step among multiple associated steps.

In a possible implementation, the processing unit 202 is used to extract key features of the second processing scheme to obtain multiple noun phrase key features and multiple verb key features; perform sequential logical structure recognition on the second processing scheme to determine multiple sequential logical connectives included in the second processing scheme, and the sequential logical connectives are used to establish sequential logical relationships between sentences or paragraphs in the interview answers; determine a first key feature and a second key feature between a first logical connective and a second logical connective, the first logical connective and the second logical connective are any two adjacent sequential logical connectives among multiple sequential connectives, the first key feature is any one of multiple noun key features, and the second key feature is any one of multiple verb key features; combine the first key feature and the second key feature to obtain a first processing step.

In a possible implementation, the processing unit 202 is used to determine the second interview question corresponding to the first processing step according to a preset knowledge network, where the first processing step is any one of the multiple processing steps; search for multiple second pre-stored solutions for the second interview question in a preset database according to the preset knowledge network; perform feature extraction on the first processing step to obtain multiple first features, perform feature extraction on the target pre-stored solution to obtain multiple second features, where the target pre-stored solution is any one of the multiple second pre-stored solutions; calculate the correlation between the first processing step and the target pre-stored solution according to the multiple first features and the multiple second features, and the specific calculation formula is as follows: ; Wherein D(X, Y) is the correlation between the first processing step X and the target pre-stored solution Y, df(u) is the characteristic frequency of the u-th first feature, df(v) is the characteristic frequency of the v-th second feature, argmin(d(u,v)×P _uv ) is the traffic allocation matrix, P _uv is the traffic allocation from the u-th first feature to the v-th second feature, that is, the amount of traffic allocated by moving the u-th first feature to the v-th second feature, and d(u,v) is the distance between the characteristic vector of the first feature and the characteristic vector of the second feature; if it is determined that the correlation is greater than or equal to the preset threshold, the target pre-stored solution is determined to be an association step.

In a possible implementation, the processing unit 202 is configured to calculate the characteristic frequency of the first characteristic by using the following formula: ; Where df(u) is the characteristic frequency of the u-th first feature, _mu is the number of pre-stored solutions containing the first feature, and _Mu is the total number of pre-stored solutions; the characteristic frequency of the second feature is calculated by the following formula: ; Where df(v) is the characteristic frequency of the vth second feature, m _v is the number of pre-stored solutions containing the second feature, and M _v is the total number of pre-stored solutions; the distance between the characteristic vector of the first feature and the characteristic vector of the second feature is calculated by the following formula: ; Wherein, d(u,v) is the distance between the eigenvector of the first feature and the eigenvector of the second feature, _uk is the eigenvector of the first feature in the kth dimension, and _vk is the eigenvector of the second feature in the kth dimension.

In one possible implementation, the processing unit 202 is used to record a first number of pre-stored solutions for the interview question retrieved after the first user inputs the interview question; record a second number of pre-stored solutions selected by the second user without any processing; and obtain a selection rate based on a ratio of the second number to the first number.

In one possible implementation, the processing unit 202 is used to determine the similarity between the first interview question and multiple pre-stored interview questions in a preset database; determine the pre-stored interview questions among the multiple pre-stored interview questions whose similarity is greater than or equal to a preset similarity threshold; and determine multiple first pre-stored solutions corresponding to the pre-stored interview questions whose respective similarities are greater than or equal to the preset similarity threshold.

It should be noted that: when the device provided in the above embodiment realizes its function, only the division of the above functional modules is used as an example. In actual application, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

The present application also discloses an electronic device. Referring to FIG3 , FIG3 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device 300 may include: at least one processor 301 , at least one network interface 304 , a user interface 303 , a memory 305 , and at least one communication bus 302 .

The communication bus 302 is used to realize the connection and communication between these components.

The user interface 303 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

Among them, the processor 301 may include one or more processing cores. The processor 301 uses various interfaces and lines to connect various parts in the entire server, and executes various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Optionally, the processor 301 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 301 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application requests; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 301, and it can be implemented separately through a chip.

Among them, the memory 305 may include a random access memory (RAM) or a read-only memory (Read-Only Memory). Optionally, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area. Instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc. can be stored; the data storage area can store data involved in the above-mentioned various method embodiments, etc. The memory 305 can also be optionally at least one storage device located away from the aforementioned processor 301.

As shown in FIG. 3 , the memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module, and an application program for processing intelligent simulated interviews.

In the electronic device 300 shown in Figure 3, the user interface 303 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 301 can be used to call the application program for processing the intelligent simulation interview stored in the memory 305. When executed by one or more processors, the electronic device executes one or more methods described in the above embodiments.

It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for the present application.

In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

In the several embodiments provided in the present application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: various media that can store program codes, such as USB flash drives, mobile hard drives, magnetic disks or optical disks.

The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, it will be easy for those skilled in the art to think of other embodiments of the present disclosure. This application is intended to cover any variation, use or adaptive change of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field that are not recorded in the present disclosure.

Claims (10)

1. A method for processing intelligent simulation interviews, the method comprising:

Acquiring a first interview question input by a first user;

searching a plurality of first pre-stored schemes aiming at the first interview problem in a preset database according to a preset knowledge network;

Invoking a selection habit of a pre-stored second user in the preset database, wherein the selection habit comprises the selection rate of the second user on a pre-stored scheme, and the selection rate is the ratio of the number of times of selecting the pre-stored scheme by the second user to the number of displayed pre-stored schemes;

If the selection habit is lower than the preset selection rate, screening the total selection rate of the first pre-stored schemes to obtain a plurality of first processing schemes, wherein the first processing schemes are first pre-stored schemes with the total selection rate higher than the preset total selection rate, and the total selection rate is the ratio of the total selected number to the total displayed number of the corresponding first pre-stored schemes;

Splitting a second processing scheme to obtain at least two processing steps, wherein the second processing scheme is any one of a plurality of first processing schemes;

and replacing the processing step by adopting a pre-stored association step to obtain a first face test answer aiming at the first face test question, and displaying the first face test answer to the first user so that the first user can simulate according to the first face test answer.

2. The method of claim 1, wherein prior to said replacing said processing step with a pre-stored association step to obtain a first interview answer to said first interview question, said method further comprises:

determining a second interview question corresponding to a first processing step according to the preset knowledge network, wherein the first processing step is any one of a plurality of processing steps;

Searching a plurality of second pre-stored schemes aiming at the second interview problem in the preset database according to the preset knowledge network;

Sorting the total selectivity of the second pre-stored schemes according to the sequence from big to small to obtain a sorting result;

Obtaining a target selection rate corresponding to the last position from the sorting result, and determining a second test answer corresponding to the target selection rate;

And determining the second test answer as a second processing step, wherein the second processing step is a corresponding association step of the first processing step in a plurality of association steps.

3. The method according to claim 2, wherein the splitting of the second processing scheme results in at least two processing steps, specifically comprising:

Extracting key features from the second processing scheme to obtain a plurality of noun phrase key features and a plurality of verb phrase key features;

Performing sequential logic structure identification on the second processing scheme, and determining a plurality of sequential logic connection words contained in the second processing scheme, wherein the sequential logic connection words are used for establishing sequential logic relations between sentences or between sections in the interview answers;

Determining a first key feature and a second key feature between a first logic connection word and a second logic connection word, wherein the first logic connection word and the second logic connection word are any two adjacent sequential logic connection words in a plurality of sequential connection words, the first key feature is any one noun key feature in a plurality of noun key features, and the second key feature is any one verb key feature in a plurality of verb key features;

and combining the first key feature and the second key feature to obtain the first processing step.

4. The method of claim 1, wherein prior to said replacing said processing step with a pre-stored association step to obtain a first interview answer to said first interview question, said method further comprises:

determining a second interview question corresponding to a first processing step according to the preset knowledge network, wherein the first processing step is any one of a plurality of processing steps;

Searching a plurality of second pre-stored schemes aiming at the second interview problem in the preset database according to the preset knowledge network;

Performing feature extraction on the first processing step to obtain a plurality of first features, and performing feature extraction on a target pre-stored scheme to obtain a plurality of second features, wherein the target pre-stored scheme is any one of the plurality of second pre-stored schemes;

According to the first features and the second features, the association degree between the first processing step and the target pre-stored scheme is calculated, and a specific calculation formula is as follows: ; wherein D (X, Y) is a degree of association between the first processing step X and the target pre-storing scheme Y, df (u) is a feature frequency of the u-th first feature, df (v) is a feature frequency of the v-th second feature, argmin (D (u, v) X P _uv) is a flow distribution matrix, P _uv is a flow distribution from the u-th first feature to the v-th second feature, that is, a quantity of flows distributed by moving the u-th first feature to the v-th second feature, and D (u, v) is a distance between a feature vector of the first feature and a feature vector of the second feature;

And if the association degree is determined to be greater than or equal to a preset threshold value, determining the target pre-stored scheme as the association step.

5. The method according to claim 4, wherein the method further comprises:

Calculating the characteristic frequency of the first characteristic by the following formula: ; wherein df (u) is a characteristic frequency of the u-th first characteristic, M _u is the number of pre-stored schemes including the first characteristic, and M _u is the total number of pre-stored schemes;

calculating the characteristic frequency of the second characteristic by the following formula: ; wherein df (v) is the characteristic frequency of the v-th second characteristic, M _v is the number of pre-stored schemes including the second characteristic, and M _v is the total number of pre-stored schemes;

the distance between the feature vector of the first feature and the feature vector of the second feature is calculated by the following formula: ; where d (u, v) is a distance between the feature vector of the first feature and the feature vector of the second feature, u _k is a feature vector of the first feature in the kth dimension, and v _k is a feature vector of the second feature in the kth dimension.

6. The method of claim 1, wherein prior to said retrieving the pre-stored selection habits of the second user in the pre-set database, the method further comprises:

after the first user inputs an interview question, recording a first number of pre-stored schemes retrieved for the interview question;

recording the second number of pre-stored schemes selected by the second user without any processing;

and obtaining the selectivity according to the ratio of the second quantity to the first quantity.

7. The method according to claim 1, wherein searching a preset database for a plurality of first pre-stored solutions to the first interview question according to a preset knowledge network, specifically comprises:

Determining the similarity between the first interview question and a plurality of pre-stored interview questions in the preset database;

Determining a pre-stored interview question of which the similarity is greater than or equal to a preset similarity threshold value in a plurality of pre-stored interview questions;

And determining a plurality of first pre-stored schemes corresponding to pre-stored interview questions with the similarity being greater than or equal to the preset similarity threshold.

8. A processing device for intelligent simulation interviews, characterized in that the device comprises an acquisition unit (201), a processing unit (202) and a splitting unit (203);

the acquisition unit (201) acquires a first interview question input by a first user;

The processing unit (202) searches a preset database for a plurality of first pre-stored schemes aiming at the first interview question according to a preset knowledge network; invoking a selection habit of a pre-stored second user in the preset database, wherein the selection habit comprises the selection rate of the second user on a pre-stored scheme, and the selection rate is the ratio of the number of times of selecting the pre-stored scheme by the second user to the number of displayed pre-stored schemes; if the selection habit is lower than the preset selection rate, screening the total selection rate of the first pre-stored schemes to obtain a plurality of first processing schemes, wherein the first processing schemes are first pre-stored schemes with the total selection rate higher than the preset total selection rate, and the total selection rate is the ratio of the total selected number to the total displayed number of the corresponding first pre-stored schemes;

The splitting unit (203) performs splitting processing on a second processing scheme to obtain at least two processing steps, wherein the second processing scheme is any one of a plurality of first processing schemes; and replacing the processing step by adopting a pre-stored association step to obtain a first face test answer aiming at the first face test question, and displaying the first face test answer to the first user so that the first user can simulate according to the first face test answer.

9. An electronic device comprising a processor (301), a memory (305), a user interface (303) and a network interface (304), the memory (305) being for storing instructions, the user interface (303) and the network interface (304) being for communicating with other devices, the processor (301) being for executing the instructions stored in the memory (305) to cause the electronic device (300) to perform the method according to any one of claims 1-7.

10. A computer readable storage medium storing instructions which, when executed, perform the method of any one of claims 1-7.