CN111325001A - Paper identification, identification model training method, device, equipment and storage medium - Google Patents

Abstract

The paper identification, identification model training method, device, equipment and storage medium provided by the present disclosure include obtaining a paper to be identified; determining a paragraph identification corresponding to the to-be-identified paper according to a preset identification model; wherein the preset identification model is It is pre-trained on the training set of the paper. In the solution provided by the present disclosure, a model for representing a paper can be obtained by training a model using a paper training set provided with a paragraph preset identifier, so that the identifier corresponding to each paragraph in the paper can be identified based on the preset recognition model, which solves the problem of the prior art. The problem of needing to rely on manual editing of the format of the paper.

Description

Paper identification, identification model training method, device, equipment and storage medium

technical field

The present disclosure relates to text processing technology, and in particular, to a paper identification, identification model training method, device, equipment and storage medium.

Background technique

At present, many papers are published online or offline, so that more users can read the content of the papers. Before a paper is published, it needs to be edited, formatted, and published.

In the prior art, in order to unify the format of the thesis and facilitate the identification of the content corresponding to each paragraph, the author of the thesis is required to write the thesis in a prescribed format. However, this method cannot guarantee that all authors will write their papers in this format, which requires editing the format of the paper before publication.

Therefore, the publishing methods in the prior art cannot be completely automated, and the format of the paper needs to be edited manually.

SUMMARY OF THE INVENTION

The present disclosure provides a paper identification, an identification model training method, device, equipment and storage medium, so as to solve the problem in the prior art that the format of the paper needs to be edited manually before the paper is published.

A first aspect of the present disclosure is to provide a paper identification method, including:

Obtain the papers to be identified;

Determine the paragraph identifier corresponding to the to-be-recognized paper according to the preset recognition model;

Wherein, the preset recognition model is obtained by training according to the training set of the paper in advance.

A second aspect of the present disclosure is to provide a method for training a preset recognition model, including:

Get the thesis training set;

According to the training model of the paper training set, a preset recognition model is obtained;

Wherein, the paper training set includes a plurality of training papers, and paragraphs of the training papers are preset with identifiers.

A third aspect of the present disclosure is to provide a dissertation identification device, comprising:

The acquisition module is used to acquire the papers to be identified;

A determination module, configured to determine the paragraph identifier corresponding to the to-be-recognized paper according to a preset recognition model;

Wherein, the preset recognition model is obtained by training according to the training set of the paper in advance.

A fourth aspect of the present disclosure is to provide a training device for a preset recognition model, including:

The acquisition module is used to acquire the paper training set;

A training module for training a model according to the thesis training set to obtain a preset recognition model;

Wherein, the paper training set includes a plurality of training papers, and paragraphs of the training papers are preset with identifiers.

A fifth aspect of the present disclosure is to provide a paper identification device, comprising:

memory;

processor; and

Computer program;

Wherein, the computer program is stored in the memory, and is configured to be executed by the processor to implement the method for identifying papers as described in the first aspect above.

A sixth aspect of the present disclosure is to provide a training device for a preset recognition model, including:

memory;

processor; and

Computer program;

Wherein, the computer program is stored in the memory and configured to be executed by the processor to implement the method for training a preset recognition model as described in the second aspect above.

A seventh aspect of the present disclosure is to provide a computer-readable storage medium on which a computer program is stored, the computer program being executed by a processor to implement the paper identification method described in the first aspect above.

An eighth aspect of the present disclosure is to provide a computer-readable storage medium on which a computer program is stored, the computer program being executed by a processor to implement the method for training a preset recognition model as described in the first aspect above.

The technical effects of the paper identification, identification model training method, device, equipment and storage medium provided by the present disclosure are:

The dissertation identification, identification model training method, device, equipment and storage medium provided by the present disclosure include acquiring a dissertation to be identified; determining a paragraph identification corresponding to the dissertation to be identified according to a preset identification model; wherein the preset identification model is It is pre-trained on the training set of the paper. In the solution provided by the present disclosure, a model for representing a paper can be obtained by training a model on a paper training set provided with a paragraph preset identifier, so that the identifier corresponding to each paragraph in the paper can be identified based on the preset recognition model, which solves the problem of the prior art. The problem of needing to rely on manual editing of the format of the paper.

Description of drawings

FIG. 1 is a flowchart of a method for identifying papers according to an exemplary embodiment of the present invention;

2 is a flowchart of a method for identifying a paper according to another exemplary embodiment of the present invention;

3 is a flowchart of a training method for a preset recognition model according to an exemplary embodiment of the present invention;

4 is a flowchart of a method for training a preset recognition model according to another exemplary embodiment of the present invention;

5 is a structural diagram of a dissertation identification device according to an exemplary embodiment of the present invention;

6 is a structural diagram of a dissertation identification device according to another exemplary embodiment of the present invention;

7 is a structural diagram of an apparatus for training a preset recognition model according to an exemplary embodiment of the present invention;

8 is a structural diagram of an apparatus for training a preset recognition model according to another exemplary embodiment of the present invention;

9 is a structural diagram of a dissertation identification device according to an exemplary embodiment of the present invention;

FIG. 10 is a structural diagram of a training device for a preset recognition model according to an exemplary embodiment of the present invention.

Detailed ways

When writing a paper, the author can write according to the preset format, for example, write "title: XXX" first, then write "author: XXX", and "author unit: XXX", etc. If the author writes With the identifiers of "Title:", "Author:" and "Author's Unit:", these identifiers in the paper can be directly identified, so as to determine the corresponding content of these paragraphs, and provide a basis for automatic editing and publishing of the paper. However, some authors may not strictly follow the preset format when writing their papers, for example, instead of writing "title:", they directly write the content of the title. Identify the content of the title in the paper.

Based on this, an embodiment of the present invention provides a method for identifying a paper and a method for training a preset recognition model, and a preset recognition model is obtained by training based on the training set of the paper. The model can identify each paragraph in the paper according to the content of each paragraph in the paper. For the corresponding identification, even if the author does not write the identification in the paper, the identification corresponding to each paragraph can be identified based on the model, thereby solving the problem that the format of the paper needs to be edited manually in the prior art.

FIG. 1 is a flowchart of a method for identifying papers according to an exemplary embodiment of the present invention.

As shown in FIG. 1 , the method for identifying papers provided by this embodiment includes:

Step 101: Acquire the papers to be identified.

The method provided in this embodiment may be executed by an electronic device having a computing function, such as a computer. The electronic device is used to identify the identification of individual paragraphs in the paper.

Specifically, a preset recognition model is stored in the electronic device, and the recognition model may be obtained by training other devices, or may be obtained by executing the method provided in this embodiment, which is not limited in this embodiment.

Further, a database for storing papers may be provided, and the database may be provided in the electronic device performing the method provided in this embodiment, or in other devices. For example, the database may be set in a server, and the server is connected with the electronic device that executes the method provided by this embodiment. Users can upload papers through the client, and the server can receive the papers uploaded by users and store them in the database. The electronic device can access the database, and the database can also actively push papers to the electronic device, so that the electronic device can obtain the papers to be identified. When the electronic device obtains the papers to be identified, the principle of first-in, first-out can be adopted, and the papers uploaded by the user can be processed first.

In practical applications, when a user uploads a paper through a user terminal, it can be directly received by the electronic device executing the method provided in this embodiment, so that the electronic device can obtain the paper to be identified and directly identify the uploaded paper.

Step 102: Determine the paragraph identifier corresponding to the to-be-recognized paper according to the preset recognition model.

Among them, the preset recognition model is pre-trained according to the training set of the paper.

Specifically, a preset recognition model is stored in the electronic device that executes the method provided in this embodiment. The papers to be identified can be input into a preset identification model, and the identification results can be output by the preset identification model.

Further, before training the preset recognition model, a large number of papers can be collected as a training set of papers. These papers may or may not have labels marked by the authors during the writing process. The identifier corresponding to each paragraph in these papers can be predetermined, and the content of the paper and the paragraph identifier corresponding to the paper can be input into the model, so as to train the weight value inside the model.

In practical application, the trained recognition model can be directly applied to output the identification of the paper to be recognized. Specifically, the identifier corresponding to each paragraph in the to-be-recognized paper can be output, for example, the first paragraph is the title, the second paragraph is the author, and the third paragraph is the author's unit, etc.

Among them, if the author writes specific identifiers during the writing process, the model can determine the identifiers corresponding to the paragraphs according to these existing identifiers. If the paragraphs have no identifiers, the identifiers corresponding to each paragraph can be calculated according to the weights in the model.

The method provided by this embodiment is used to identify the to-be-recognized paper, and the method is executed by a device provided with the method provided by this embodiment, and the device is usually implemented in hardware and/or software.

The method for identifying a paper provided by this embodiment includes acquiring a paper to be recognized; determining a paragraph identifier corresponding to the paper to be recognized according to a preset recognition model; wherein, the preset recognition model is pre-trained based on a paper training set. The method provided by this embodiment can identify the identifier corresponding to each paragraph in the paper based on the preset recognition model, thereby solving the problem that the format of the paper needs to be edited manually in the prior art.

FIG. 2 is a flowchart of a method for identifying papers according to another exemplary embodiment of the present invention.

As shown in Figure 2, the method for identifying papers provided by this embodiment includes:

Step 201: Acquire the papers to be identified.

The specific principles and implementation manners of step 201 and step 101 are similar, and are not repeated here.

Step 202: Perform word segmentation on the paragraphs included in the to-be-recognized paper to obtain word segmentations included in each paragraph.

In the method provided in this embodiment, the to-be-recognized paper may not be directly input into the preset recognition model, but the to-be-recognized paper may be subjected to word segmentation processing, and then the to-be-recognized model may determine the identifier corresponding to each paragraph based on the obtained word segmentation.

Wherein, a word segmentation algorithm can be preset, so that each paragraph included in the to-be-recognized paper is subjected to word segmentation processing according to the word segmentation algorithm. Chinese Word Segmentation refers to dividing a sequence of Chinese characters into individual words. The word segmentation algorithm is the process of recombining consecutive word sequences into word sequences according to certain specifications. You can also set a word segmentation dictionary, so as to perform word segmentation processing on the paragraphs in the paper according to the word segmentation dictionary.

Specifically, the paper may include multiple paragraphs, and word segmentation can be performed on each paragraph to obtain the word segmentation corresponding to each paragraph. In the process of word segmentation, the modal particles in the paragraph and the words without actual meaning, such as "ah, de", etc., can also be removed, so that the paragraph is only identified according to the words with actual meaning, which can reduce the number of words in the paragraph recognition process. amount of calculation.

Step 203: Determine the identifier corresponding to the paragraph according to the preset recognition model and word segmentation.

Further, in the method provided in this embodiment, the preset recognition model may determine the identifier corresponding to the paragraph based on the word segmentation in the paragraph. For example, the word segmentation in each paragraph can be used as a word segmentation combination and input into a preset recognition model, so that the recognition model can determine the paragraph identifier corresponding to the word segmentation combination, that is, the paragraph identifier corresponding to the paragraph to which the word segmentation combination belongs.

In another embodiment, a preset thesaurus can also be set, and a plurality of words are set in the preset thesaurus, and it can be considered that the words in the preset sub-library belong to the same type of vocabulary. For example, it can be synonyms or synonyms to form a preset thesaurus.

Wherein, the preset thesaurus to which the word segmentation belongs can be determined, and the frequency of the word segmentation included in each paragraph belonging to each preset sub-library can be determined. For a paragraph, multiple word segments can be determined, and each word segment has a corresponding preset thesaurus. Therefore, it is possible to count the frequency that the segmented words included in the paragraph belong to each preset thesaurus. For example, if the paragraph includes 5 segmented words, which belong to the thesaurus A, A, B, C, and C, respectively, the preset corresponding to the paragraph can be obtained. The frequency of the secondary library is A-0.4, B-0.2, C-0.4.

Specifically, the feature vector corresponding to each paragraph is determined according to the frequency. The frequency can be directly used as an eigenvector, such as (0.4, 0.2, 0.4). Usually, each preset thesaurus contains words, and it is impossible for them to appear in the same paragraph. Therefore, in order to express the meaning of the paragraph more accurately through the feature vector, the probability of non-appearing thesaurus can also be set to 0. For example, if the preset secondary library also includes thesaurus D, but the segmented words in the paragraph do not fall into thesaurus D, the feature vector may be (0.4, 0.2, 0.4, 0).

Further, the data processing amount of the recognition model can be reduced by replacing the paragraphs of the entire paper by the feature vector.

In practical application, the identifier corresponding to the paragraph can be determined according to the feature vector based on the preset recognition model. The feature vector can be input into the preset recognition model, and the feature vector can be calculated according to the weight value inside the preset recognition model, and the paragraph identification corresponding to the feature vector can be output, that is, the identification of the paragraph corresponding to the feature vector. For example, the output paragraph identification may be "thesis content".

Among them, the method provided in this embodiment can perform word segmentation on the papers to be identified, and then determine the feature vector corresponding to each paragraph based on the obtained word segmentation, and then input the feature vector into the preset recognition model, so as to determine according to the preset recognition model. The identification corresponding to the paragraph can reduce the calculation amount of the recognition model.

FIG. 3 is a flowchart of a method for training a preset recognition model according to an exemplary embodiment of the present invention.

As shown in FIG. 3 , the training method of the preset recognition model provided by this embodiment includes:

Step 301, obtaining a training set of papers.

The method provided in this embodiment may be executed by an electronic device having a computing function, such as a computer. The electronic device is used to train a preset recognition model.

Specifically, the electronic device can be used only for training a preset recognition model, and can also recognize papers to be recognized, that is, the electronic device used for training and the electronic device used for identifying papers can be the same or different. This embodiment does not limit this.

Further, a database for storing the training set of the thesis may be provided, and the database may be provided in the electronic device that executes the method provided in this embodiment, or may be provided in other devices. For example, the database may be set in a server, and the server is connected with the electronic device that executes the method provided by this embodiment. These papers can be acquired in advance, and the paragraph identifiers corresponding to each paragraph in the paper can be determined.

Step 302: Train the model according to the thesis training set to obtain a preset recognition model.

Among them, the paper training set includes a plurality of training papers, and the paragraphs of the training papers are preset with identifiers.

Specifically, the model can be trained based on a pre-prepared collection of papers to obtain a preset recognition model.

Further, the weight values inside the model can be initialized. And input the papers included in the thesis into the model, the model can determine the corresponding identification of each paragraph of the paper according to the current weight, and then compare the determined result with the preset paragraph identification, and adjust the weight value inside the model according to the comparison result.

In practical applications, the papers can be input into the model one by one, and each time a paper is input, the weights in the model can be corrected once, and through multiple adjustments, a set of accurate models for identifying papers can be obtained.

Among them, a part of the papers in the training set of papers can also be used as the test set. When the model is trained, the papers in the test set can be input into the model, and the paragraph identifiers corresponding to each paragraph in the paper can be determined and compared with the pre-determined identifiers. If the correct rate is higher than the preset threshold, the The model is considered accurate and can be used to identify user-provided papers.

Specifically, the papers in the test set can be used for training the model, or only for testing. The above-mentioned preset thresholds can be set according to requirements.

The method provided in this embodiment is used to train a preset recognition model, and the method is executed by a device provided with the method provided in this embodiment, and the device is usually implemented in hardware and/or software.

The training method for a preset recognition model provided by this embodiment includes acquiring a training set of papers; training a model according to the training set of papers to obtain a preset recognition model; wherein, the training set of papers includes a plurality of training papers, and the paragraphs of the training papers are preset with logo. In the method provided in this embodiment, a model for representing a thesis can be obtained by training a model using the thesis training set provided with the preset identification of the paragraph, so that the identification corresponding to each paragraph in the thesis can be identified based on the preset identification model, so as to solve the problem of the current situation. There are issues in technology that rely on manual editing of the format of a paper.

FIG. 4 is a flowchart of a method for training a preset recognition model according to another exemplary embodiment of the present invention.

As shown in FIG. 4 , the training method of the preset recognition model provided by this embodiment includes:

Step 401, obtaining a training set of papers.

The specific principles and implementation manners of step 401 and step 301 are similar, and are not repeated here.

Step 402: Perform word segmentation on the paragraphs included in the training paper to obtain the training word segmentation included in each paragraph.

In the method provided by this embodiment, in the process of training the model, instead of directly inputting the training paper into the model, the training paper may be processed for word segmentation, and then the model is trained according to the obtained training word segmentation.

Wherein, a word segmentation algorithm may be preset, so as to perform word segmentation processing on each paragraph included in the training paper according to the word segmentation algorithm. Chinese Word Segmentation refers to dividing a sequence of Chinese characters into individual words. The word segmentation algorithm is the process of recombining consecutive word sequences into word sequences according to certain specifications. You can also set a word segmentation dictionary, so as to perform word segmentation processing on the paragraphs in the paper according to the word segmentation dictionary.

Specifically, the training paper may include multiple paragraphs, and word segmentation processing can be performed on each paragraph to obtain the training word segmentation corresponding to each paragraph. In the process of word segmentation, the modal particles in the paragraph and the words without actual meaning, such as "ah, de", etc., can also be removed, so that the paragraph is only identified according to the words with actual meaning, which can reduce the number of words in the paragraph recognition process. amount of calculation.

Step 403: Train the model according to the training word segmentation to obtain a preset recognition model.

Further, in the method provided in this embodiment, the model may be trained based on the training word segmentation in the paragraph. For example, you can initialize the weight value in the model first, then use the training word segmentation in each paragraph as a word segmentation combination, input the model, and the model can determine the paragraph identifier corresponding to the word segmentation combination according to the current weight value, that is, the corresponding word segmentation combination. For the identification of the paragraph, the determined identification can be compared with the preset identification of the paragraph, and the weight value in the model can be adjusted according to the comparison result. Through multiple rounds of adjustment, an accurate preset recognition model can be obtained.

In another embodiment, a preset thesaurus can also be set, and a plurality of words are set in the preset thesaurus, and it can be considered that the words in the preset sub-library belong to the same type of vocabulary. For example, it can be synonyms or synonyms to form a preset thesaurus.

Wherein, the preset word bank to which the training word segment belongs can be determined, and the frequency at which the training word segment included in each paragraph belongs to each preset sub-bank can be determined. For a paragraph, multiple training word segments can be determined, and each word segment has a corresponding preset thesaurus. Therefore, it is possible to count the frequency that the training word segments included in the paragraph belong to each preset thesaurus. For example, if the paragraph includes 5 training word segments, and the training belongs to the word banks A, A, B, C, and C respectively, then the corresponding paragraphs can be obtained. The preset thesaurus frequencies are A-0.4, B-0.2, C-0.4.

Specifically, the feature vector corresponding to each paragraph is determined according to the frequency. The frequency can be directly used as an eigenvector, such as (0.4, 0.2, 0.4). Under normal circumstances, the words included in each preset thesaurus cannot all appear in the same paragraph. Therefore, in order to express the meaning of the paragraph more accurately through the feature vector, the probability of not appearing in the thesaurus can also be set to 0. For example, if the preset secondary library also includes thesaurus D, but the training word segmentation in the paragraph does not fall into thesaurus D, the feature vector may be (0.4, 0.2, 0.4, 0).

Further, the data processing volume of the model can be reduced by replacing the paragraphs of the entire paper with feature vectors.

In practical application, the preset recognition model can be obtained by training the model according to the training feature vector corresponding to the paragraph and the preset identification. The feature vector can be input into the model, and the feature vector can be calculated according to the current weight value inside the model to determine the paragraph identifier corresponding to the feature vector, that is, the identifier of the paragraph corresponding to the feature vector. For example, the determined paragraph identifier may be "thesis content". The determined identification can be compared with the preset identification, and the weight value in the model can be adjusted according to the comparison result.

Among them, the method provided in this embodiment can perform word segmentation processing on the training paper, and then determine the feature vector corresponding to each paragraph based on the obtained training word segmentation, and then input the feature vector into the model, so as to determine the identifier corresponding to the paragraph according to the model, and then Adjusting the weight value in the model based on the preset identifier can reduce the amount of computation in the model training process.

FIG. 5 is a structural diagram of an apparatus for identifying papers according to an exemplary embodiment of the present invention.

As shown in Figure 5, the paper identification device provided by this embodiment includes:

an acquisition module 51, used to acquire the papers to be identified;

A determination module 52, configured to determine the paragraph identifier corresponding to the to-be-recognized paper according to a preset recognition model;

Wherein, the preset recognition model is obtained by training according to the training set of the paper in advance.

The paper identification device provided by this embodiment includes an acquisition module for acquiring papers to be identified; a determination module for determining paragraph identifications corresponding to papers to be identified according to a preset identification model; wherein the preset identification model is pre-trained according to the papers set of training. The device provided in this embodiment can identify the corresponding identifiers of each paragraph in the paper based on the preset recognition model, thereby solving the problem that the format of the paper needs to be edited manually in the prior art.

The specific principle and implementation manner of the paper identification device provided in this embodiment are similar to the embodiment shown in FIG. 1 , and details are not repeated here.

FIG. 6 is a structural diagram of an apparatus for identifying papers according to another exemplary embodiment of the present invention.

As shown in FIG. 6 , on the basis of the embodiment shown in FIG. 5 , in the paper identification device provided in this embodiment, the determining module 52 includes:

The word segmentation unit 521 is used to perform word segmentation processing on the paragraphs included in the to-be-recognized paper to obtain the word segmentation included in each paragraph;

The determining unit 522 is configured to determine the identifier corresponding to the paragraph according to the preset recognition model and the word segmentation.

The determining unit 522 is specifically used for:

Determine the preset thesaurus to which the segmented word belongs, and determine the frequency at which the segmented word included in each paragraph belongs to each of the preset secondary libraries;

Determine the feature vector corresponding to each of the paragraphs according to the frequency;

The identifier corresponding to the paragraph is determined according to the feature vector based on the preset recognition model.

The specific principle and implementation manner of the paper identification device provided in this embodiment are similar to the embodiment shown in FIG. 2 , and details are not described herein again.

FIG. 7 is a structural diagram of an apparatus for training a preset recognition model according to an exemplary embodiment of the present invention.

As shown in FIG. 7 , the training device for the preset recognition model provided by this embodiment includes:

an acquisition module 71 for acquiring a training set of papers;

A training module 72, configured to train a model according to the thesis training set to obtain a preset recognition model;

Wherein, the paper training set includes a plurality of training papers, and paragraphs of the training papers are preset with identifiers.

The training device for a preset recognition model provided in this embodiment includes an acquisition module for acquiring a training set of papers; a training module for training a model according to the training set of papers to obtain a preset recognition model; wherein the training set of papers includes a plurality of papers The training paper, the paragraphs of the training paper are preset with flags. The device provided in this embodiment can obtain a model for representing a thesis by training a model using the thesis training set provided with the preset identification of the paragraph, so that the identification corresponding to each paragraph in the thesis can be identified based on the preset identification model, so as to solve the problem of the current situation. There are issues in technology that rely on manual editing of the format of a paper.

The specific principles and implementation manners of the training device for the preset recognition model provided in this embodiment are similar to the embodiment shown in FIG. 3 , and details are not described herein again.

FIG. 8 is a structural diagram of an apparatus for training a preset recognition model according to another exemplary embodiment of the present invention.

As shown in FIG. 8 , on the basis of the embodiment shown in FIG. 7 , in the training device for the preset recognition model provided by this embodiment, the training module 72 includes:

The word segmentation unit 721 is used to perform word segmentation processing on the paragraphs included in the training paper to obtain the training word segmentations included in each paragraph;

The training unit 722 is configured to train the model according to the training word segmentation to obtain the preset recognition model.

The training unit 722 is specifically used for:

Determine the preset thesaurus to which the training participle belongs, and determine the frequency at which the training participle included in each paragraph belongs to each of the preset sub-libraries;

Determine the training feature vector corresponding to each of the paragraphs according to the frequency;

The preset recognition model is obtained according to the training feature vector corresponding to the paragraph and the preset identification training model.

FIG. 9 is a structural diagram of a dissertation identification device according to an exemplary embodiment of the present invention.

As shown in Figure 9, the paper identification device provided by this embodiment includes:

memory 91;

processor 92; and

Computer program;

Wherein, the computer program is stored in the memory 91 and is configured to be executed by the processor 92 to implement any of the paper identification methods described in FIGS. 1-2 .

FIG. 10 is a structural diagram of a training device for a preset recognition model according to an exemplary embodiment of the present invention.

As shown in FIG. 10 , the training equipment for the preset recognition model provided by this embodiment includes:

memory 1001;

processor 1002; and

Computer program;

Wherein, the computer program is stored in the memory 1001 and is configured to be executed by the processor 1002 to implement any one of the preset recognition model training methods described in FIGS. 3-4 .

This embodiment also provides a computer-readable storage medium on which a computer program is stored,

The computer program is executed by a processor to implement any of the paper identification methods described in Figures 1-2.

This embodiment also provides a computer-readable storage medium on which a computer program is stored,

The computer program is executed by the processor to implement any one of the preset recognition model training methods described in FIGS. 3-4 .

Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments may be completed by program instructions related to hardware. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps including the above method embodiments are executed; and the foregoing storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: The technical solutions described in the foregoing embodiments can still be modified, or some or all of the technical features thereof can be equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present invention. scope.

Claims (12)

1. A method of paper identification, comprising:

acquiring a paper to be identified;

determining paragraph marks corresponding to the thesis to be recognized according to a preset recognition model;

wherein, the preset recognition model is obtained by training according to a thesis training set in advance.

2. The method according to claim 1, wherein the determining the paragraph identifier corresponding to the paper to be recognized according to the preset recognition model includes:

performing word segmentation processing on the paragraphs included in the thesis to be identified to obtain the word segmentation included in each paragraph;

and determining the mark corresponding to the paragraph according to the preset recognition model and the word segmentation.

3. The method according to claim 2, wherein the determining the corresponding identifier of the paragraph according to the preset recognition model and the segmentation word comprises:

determining a preset word bank to which the participle belongs, and determining the frequency of the participle included in each paragraph belonging to each preset sub-bank;

determining a feature vector corresponding to each paragraph according to the frequency;

and determining the identifier corresponding to the paragraph according to the feature vector based on the preset identification model.

4. A training method for a preset recognition model is characterized by comprising the following steps:

acquiring a thesis training set;

training a model according to the thesis training set to obtain a preset recognition model;

the paper training set comprises a plurality of training papers, and the paragraphs of the training papers are preset with marks.

5. The method of claim 4, wherein training the model according to the thesis training set to obtain a predetermined recognition model comprises:

performing word segmentation processing on paragraphs included in the training paper to obtain training words included in each paragraph;

and training a model according to the training part words to obtain the preset recognition model.

6. The method of claim 5, wherein the training the model according to the training part word to obtain the preset recognition model comprises:

determining a preset word bank to which the training participle belongs, and determining the frequency of the training participle included in each paragraph belonging to each preset sub-bank;

determining training feature vectors corresponding to the sections according to the frequency;

and obtaining the preset recognition model according to the training feature vector corresponding to the paragraph and a preset identification training model.

7. An article identification apparatus, comprising:

the acquisition module is used for acquiring the thesis to be identified;

the determining module is used for determining paragraph identifiers corresponding to the papers to be recognized according to a preset recognition model;

wherein, the preset recognition model is obtained by training according to a thesis training set in advance.

8. A training device for presetting a recognition model is characterized by comprising:

the acquisition module is used for acquiring a thesis training set;

the training module is used for training a model according to the thesis training set to obtain a preset recognition model;

the paper training set comprises a plurality of training papers, and the paragraphs of the training papers are preset with marks.

9. A paper identification device, comprising:

a memory;

a processor; and

a computer program;

wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method of any of claims 1-3.

10. A training device for presetting a recognition model is characterized by comprising:

a memory;

a processor; and

a computer program;

wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method of any of claims 4-6.

11. A computer-readable storage medium, having stored thereon a computer program,

the computer program is executed by a processor to implement the method according to any one of claims 1-3.

12. A computer-readable storage medium, having stored thereon a computer program,

the computer program is executed by a processor to implement the method according to any of claims 4-6.

Images