WO2019114430A1 - Natural language question understanding method and apparatus, and electronic device - Google Patents

Abstract

The present disclosure relates to the technical field of natural language processing. Provided are a natural language question understanding method and apparatus, and an electronic device. The natural language question understanding method comprises: obtaining natural language question information input by a user terminal, the natural language question information being question information related to data query; parsing the natural language question information to obtain a minimum parsing unit; generating, on the basis of the minimum parsing unit and a preset instruction set, a query instruction corresponding to the natural language question information; and retrieving, according to the query instruction, from a preset knowledge base a data result corresponding to the natural language question information, the preset knowledge base being generated according to database data provided by a user, input information data of the user, and/or third-party data. The method can accurately recognize natural language question information.

Description

Method, device and electronic device for understanding natural language questioning

Cross-reference to related applications

The present application claims priority to Chinese Patent Application No. CN2017113616797, entitled "Understanding Method, Apparatus and Electronic Device for Natural Language Questioning", filed on December 15, 2017, the entire contents of which are incorporated by reference. In the present disclosure.

Technical field

The present disclosure relates to the field of natural language processing technologies, and in particular, to a method, device, and electronic device for understanding natural language questions.

Background technique

Natural language processing is a long-term concern and research by human beings. Currently, this technology is mainly used in many fields such as multi-language translation and information query, and has made good progress. However, natural language processing is directly applied to data analysis. There is no precedent on the top.

Natural language processing is divided into many different technical genres. At the beginning, natural language processing methods based on formal languages are in the mainstream, but this technical route cannot handle the expressions that are rich in change, and can only be written mechanically according to certain Templates or rules that translate or generate language are very blunt. Later, there was a way to introduce statistical mathematics theory into language processing. For example, most of the machine translation systems such as Google Translate and Baidu Translation were developed on the basis of such systems. This natural language processing method based on statistical theory can effectively use a large number of corpora to train the model, so as to acquire various forms of language expression. It is currently very good at multilingual translation. However, this technical route still has the defect that the recognition accuracy needs to be improved.

Summary of the invention

In view of this, the present disclosure provides a method, device, and electronic device for understanding natural language questions.

In a first aspect, the present disclosure provides a method for understanding natural language questions, including:

Obtaining natural language question information input by the user end; the natural language question information is question information related to the data query;

Parsing the natural language question information to obtain a minimum parsing unit;

Generating a query instruction corresponding to the natural language question information based on the minimum parsing unit and the preset instruction set;

The data is retrieved from the preset knowledge base according to the query instruction, and the data result corresponding to the natural language question information is obtained; the preset knowledge base is generated according to the database data provided by the user, the input information data of the user, and/or the third party data.

With reference to the first aspect, the present disclosure provides a first possible implementation manner of the first aspect, wherein parsing the natural language question information to obtain a minimum parsing unit includes:

Perform word segmentation on natural language question information to obtain multiple word segmentation segments;

The entity noun recognition is performed on a plurality of segmentation segments to obtain a minimum parsing unit; the minimum parsing unit includes: an attribute minimum parsing unit, a metric minimum parsing unit, and a time-space modified structural word.

With reference to the first aspect, the present disclosure provides a second possible implementation manner of the first aspect, wherein the attribute minimum parsing unit comprises at least one of an attribute item, a calculated operation item, and an attribute logical relationship item; the metric minimum parsing unit includes the metric At least one of the item, the metric logical relationship item, and the calculated modifier.

Optionally, the attribute item in the attribute minimum parsing unit represents a category or entity to which the natural language question information belongs; the calculation operation item in the attribute minimum parsing unit represents classification, grouping, segmentation, partial value, counting, pinyin Ranking and ranking by last name stroke; attribute logical relationship items in attribute minimum resolution unit represent logical relationships of similarity, dissimilarity, inclusion, non-containment, comparison, and comparison.

Optionally, the metric item in the metric minimum analytic unit represents a numerical value; the metric logical relationship item in the metric minimum analytic unit represents a numerical relationship of greater than, less than, equal to, greater than or equal to, less than or equal to, and not equal to; The calculated modifiers in the metric minimum parsing unit represent summation, average, count, standard deviation, variance, correlation, correlation coefficient, daily growth rate, weekly growth rate, monthly growth rate, quarterly growth rate, annual growth rate, and ranking. , maximum, minimum, front N, and reciprocal N.

Optionally, the space-time modified structure word represents at least one of the following: a time period, a time point, a point before, a point after and a point, and a point and a point; a preset distance range somewhere Outside, somewhere within the preset distance range and somewhere set the direction relative position. For example, near a certain place, somewhere within a range of kilometers, miles, meters, kilometers, etc., the relative position of the east, south, west, and north directions.

With reference to the first aspect, the present disclosure provides a third possible implementation manner of the first aspect, wherein the generating the query instruction corresponding to the natural language question information based on the minimum parsing unit and the preset instruction set includes:

Inferring data query logic included in the natural language question information according to the minimum parsing unit;

According to the data query logic, the corresponding instructions are extracted from the preset instruction set and combined to generate a query instruction corresponding to the natural language question information.

Optionally, inferring, according to the minimum parsing unit, the data query logic included in the natural language question information, specifically:

Obtaining a category corresponding to the natural language question information;

According to the obtained category, the query logic corresponding to the natural language question information is obtained.

In conjunction with the first aspect, the present disclosure provides a fourth possible implementation manner of the first aspect, wherein the method is performed after performing a retrieval from a preset knowledge base according to a query instruction to obtain a data result corresponding to the natural language question information Also includes:

Obtaining knowledge base sample data; the knowledge base sample data includes: user-provided database data, user input information data, and/or third-party data;

Generate a preset knowledge base based on the knowledge base sample data.

With reference to the first aspect, the present disclosure provides a fifth possible implementation manner of the first aspect, wherein, after performing a retrieval from a preset knowledge base according to the query instruction, obtaining the data result corresponding to the natural language question information, the method Also includes:

The natural language question information and its corresponding data result are added to the preset knowledge base.

In a second aspect, the present disclosure provides an apparatus for understanding natural language questions, including:

The information acquisition module is configured to obtain natural language question information input by the user end; the natural language question information is question information related to the data query;

An information parsing module, configured to parse natural language question information to obtain a minimum parsing unit;

An instruction generating module, configured to generate a query instruction corresponding to the natural language question information based on the minimum parsing unit and the preset instruction set;

a retrieval module, configured to retrieve from the preset knowledge base according to the query instruction, and obtain a data result corresponding to the natural language question information; the preset knowledge base is generated according to the database data provided by the user, the input information data of the user, and/or the third party data. .

With reference to the second aspect, the present disclosure provides a first possible implementation manner of the second aspect, wherein the information parsing module includes:

a word segmentation module for performing word segmentation on natural language question information to obtain a plurality of word segmentation segments;

The identification module is configured to perform entity noun recognition on the plurality of segmentation segments to obtain a minimum parsing unit; the minimum parsing unit comprises: a minimum parsing unit of the attribute, a minimum parsing unit of the metric, and a structured word in time and space.

Optionally, the attribute minimum parsing unit includes at least one of an attribute item, a calculation operation item, and an attribute logical relationship item; the metric minimum parsing unit includes at least one of a metric item, a metric logical relationship item, and a calculation modification item.

Optionally, the attribute item in the attribute minimum parsing unit represents a category or entity to which the natural language question information belongs; the calculation operation item in the attribute minimum parsing unit represents classification, grouping, segmentation, partial value, counting, pinyin Ranking and ranking by last name stroke; attribute logical relationship items in attribute minimum resolution unit represent logical relationships of similarity, dissimilarity, inclusion, non-containment, comparison, and comparison. Wherein, the segmentation may include cutting the head, ending, and the like. Partial values can include for the string, the nth to mth letters, and so on.

Optionally, the metric item in the metric minimum analytic unit represents a numerical value; the metric logical relationship item in the metric minimum analytic unit represents a numerical relationship of greater than, less than, equal to, greater than or equal to, less than or equal to, and not equal to; The calculated modifiers in the metric minimum parsing unit represent summation, average, count, standard deviation, variance, correlation, correlation coefficient, daily growth rate, weekly growth rate, monthly growth rate, quarterly growth rate, annual growth rate, and ranking. , maximum, minimum, front N, and reciprocal N.

Optionally, the space-time modified structure word represents at least one of the following: a time period, a time point, a point before, a point after and a point, and a point and a point; a preset distance range somewhere Outside, somewhere within the preset distance range and somewhere set the direction relative position.

Optionally, the instruction generating module is specifically configured to generate, according to the minimum parsing unit and the preset instruction set, a query instruction corresponding to the natural language question information by using the following steps:

Deriving data query logic included in the natural language question information according to the minimum parsing unit;

According to the data query logic, corresponding instructions are extracted from the preset instruction set and combined to generate a query instruction corresponding to the natural language question information.

Optionally, the instruction generating module is specifically configured to infer, according to the minimum parsing unit, data query logic included in the natural language question information by:

Obtaining a category corresponding to the natural language question information;

According to the obtained category, the query logic corresponding to the natural language question information is obtained.

In a third aspect, the present disclosure further provides an electronic device comprising a memory and a processor, the memory storing a computer program executable on the processor, the step of implementing the method of the first aspect when the processor executes the computer program.

In a fourth aspect, the present disclosure also provides a computer readable medium having processor-executable non-volatile program code, the program code causing a processor to perform the method of the first aspect.

In the method, device and electronic device for understanding natural language questions provided by the present disclosure, firstly, the natural language question information is parsed to obtain a minimum parsing unit, and then the natural language question is constructed based on the minimum parsing unit and the preset instruction set. The query statement of the information is further retrieved from the pre-established knowledge base according to the query statement, and the data result corresponding to the question information is obtained, wherein the knowledge base is established based on the database data provided by the user, the input information data of the user, and/or The third-party data can provide accurate and calculated data results for the question information, so that it can be applied to professional scenes such as data analysis.

Other features and advantages of the present disclosure will be set forth in the description which follows. The objectives and other advantages of the disclosure are realized and attained by the structure of the invention.

The above described objects, features, and advantages of the present invention will become more apparent from the description of the appended claims.

DRAWINGS

In order to more clearly illustrate the specific embodiments of the present disclosure or the technical solutions in the prior art, the drawings to be used in the specific embodiments or the description of the prior art will be briefly described below, and obviously, the attached in the following description The figures are some embodiments of the present disclosure, and other drawings may be obtained from those of ordinary skill in the art without departing from the drawings.

FIG. 1 is a flowchart of a method for understanding a natural language question provided by the present disclosure;

2 is a flowchart of another method for understanding natural language questions provided by the present disclosure;

FIG. 3 is a flowchart of another method for understanding natural language questions provided by the present disclosure;

4 is a flowchart of another method for understanding natural language questions provided by the present disclosure;

FIG. 5 is a flowchart of another method for understanding natural language questions provided by the present disclosure;

FIG. 6 is a schematic structural diagram of an apparatus for understanding natural language questions provided by the present disclosure;

FIG. 7 is a schematic structural diagram of an electronic device according to the present disclosure.

Detailed ways

The present invention will be clearly and completely described in the following with reference to the accompanying drawings. example. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present disclosure without departing from the inventive scope are the scope of the disclosure.

At present, the existing natural language processing methods have the defects that the recognition accuracy needs to be improved. The research found that the recognition accuracy needs to be improved mainly in the following aspects: (1) If some scenes do not have a large amount of corpus accumulation, their recognition performance will be greatly reduced; (2) the model trained by statistical methods is not accurate. Sexual, difficult to express or resolve the exact meaning. Therefore, it cannot be applied to some very professional scenes, such as the field of data analysis. Based on this, the method, device and electronic device for understanding natural language questions provided by the present disclosure can accurately identify the natural language question information of the user and match the high-accuracy data result, which can be applied to the field of data analysis. And other professional scenes.

In order to facilitate understanding of the present embodiment, a method for understanding natural language questions disclosed in the present disclosure will be described in detail first.

Embodiment 1:

The present disclosure provides a method for understanding a natural language question, which can be applied to a relatively professional scene such as a data analysis field, and can be executed by an electronic device having a data processing function or the like. Referring to Figure 1, the method includes the following steps:

S101: Acquire natural language question information input by the user end; the natural language question information is question information related to the data query.

In the specific implementation, the user can input the natural language question information to the software system by using the input process of voice or typing, and input the natural language question information to the software system, such as "the Tibetan male who is older than 60 (years old) before the age of 2016 ( The average age of the body weight is less than 70kg". The above natural language question information is finally obtained by the server in the form of text, and the natural language question information is question information related to the data query. The electronic device may also pre-store a plurality of natural language question information and hierarchically store a plurality of natural language question information. For example, the first data analysis scenario, the second data analysis scenario, the M data may be set. Analyze the scene, M is an integer greater than 3. Each data analysis scenario can correspond to different sub-scenarios, and each sub-scenario corresponds to different natural language question information. The electrical device displays each data analysis scenario to the user, and the user selects the data analysis scenarios according to the required data to select the desired natural language question information for input.

S102: Parse the natural language question information to obtain a minimum parsing unit.

In order to obtain the final accurate data result corresponding to the natural language question information, the natural language question information needs to be accurately analyzed and understood. Therefore, after obtaining the natural language question information input by the user, the natural language needs to be first The question information is parsed to obtain the minimum parsing unit. The specific parsing process includes the following steps, as shown in Figure 2:

S201: Perform word segmentation on the natural language question information to obtain a plurality of word segmentation segments.

In the specific implementation, the word processing of the natural language question information is firstly processed, that is, the entity boundary recognition, and the natural language question information is segmented into a plurality of word segmentation, for example, the above question "the age of more than 60 (years old) before 2016) The Tibetan male surnamed Wang is less than the average age of 70kg. After the word segmentation, you can get "before 2016", "age", "greater than 60 (years old), "China", "surname", "king" Tibetan ""male"" "weight" "less than 70kg" "average" "age" multiple segmentation.

There are various ways to perform word segmentation processing on natural language question information. For example, word segmentation processing can be performed on natural language question information based on dictionary word segmentation algorithms such as forward maximum matching method, inverse maximum matching method and two-way matching word segmentation method. For another example, the natural language question information can be segmented based on a word segmentation tool such as a Stanford word segmentation tool or a Hanlp word segmentation tool. For another example, a natural language question can be asked based on a Hidden Markov Model (HMM), a Conditional Random Field Algorithm (CRF), a Support Vector Machine (SVM), and a deep learning. Information is processed in word segmentation. It is also possible to perform word segmentation processing on natural language question information based on a combination of one or more of the above.

S202: Perform entity noun recognition on a plurality of segmentation segments to obtain a minimum resolution unit.

After obtaining a plurality of segmentation segments, the entity nouns are identified for each segmentation segment, and the minimum parsing unit corresponding to the natural language question information is determined. The minimum parsing unit includes: an attribute minimum parsing unit, a metric minimum parsing unit, and a time-space modified structural word. The attribute minimum parsing unit includes at least one of an attribute item, a calculation operation item, and an attribute logical relationship item; the metric minimum parsing unit includes at least one of a metric item, a metric logical relationship item, and a calculation modifier; the structural word includes a space-time modifier.

Optionally, the attribute item in the attribute minimum parsing unit indicates the category or entity to which the natural language question information belongs; the calculation operation item in the attribute minimum parsing unit indicates classification, grouping, segmentation, partial value, counting, and pinyin ranking. Attributes in the minimum parsing unit of the attribute represent logical relationships of similarity, dissimilarity, inclusion, and non-containment; metrics in the metric minimum parsing unit represent numerical values; metrics in the smallest parsing unit The logical relationship item represents a numerical relationship of greater than, less than, equal to, greater than or equal to, less than or equal to. The calculated modifiers in the metric minimum parsing unit represent summation, averaging (average usage can be: average xxx, not using the average of xxx), count, standard deviation, variance, correlation, correlation coefficient, daily growth rate, week Growth rate, monthly growth rate, quarterly growth rate, annual growth rate, ranking, maximum value, minimum value, pre-N, reciprocal number N, etc., N is an integer greater than or equal to 1; space-time modified structure words indicate time period, time point, A word describing time before, after, after, at a certain point, and some point, and somewhere within a preset distance range, somewhere within a preset distance range, and somewhere in a certain direction. For example, near a certain place, somewhere within a range of kilometers, miles, meters, kilometers, etc., the relative position of the east, south, west, and north directions. The space-time modified structure word may also be a combination of the above-mentioned modifiers indicating time and space, such as within a certain distance of a certain time period.

It should be noted that the core of the judgment of the attribute minimum parsing unit, the metric minimum parsing unit and the time-space modified structure word is whether the word recognized by the entity noun can be calculated, for example, the above representation is greater than, less than, included, not included, standard deviation, Words such as growth rate are words that can be calculated. Subsequent steps based on these computable words can improve the accuracy of understanding natural language question information and obtain more accurate data results for data analysis. The field provides better data reference information.

By performing word segmentation processing and entity noun recognition on the natural language question information input by the user, the minimum parsing unit corresponding to the natural language question information can be determined. For example, the question information is “box office income of action movies in North China in August 2017”. After the word segmentation, the extracted word segmentation is: time (August 2017), region (North China region), action movie, box office income . Then, the above participle segment is subjected to entity noun recognition, and the minimum parsing unit and structural word of the question information are determined. Among them, time (August 2017) is a space-time modified structure word; the region (North China region) belongs to the attribute minimum parsing unit, the action movie belongs to the attribute minimum parsing unit; the income belongs to the metric minimum parsing unit.

S103: Generate a query instruction corresponding to the natural language question information based on the minimum parsing unit and the preset instruction set.

Specifically, the process of generating the query instruction includes the following steps, as shown in FIG. 3:

S301: Infer the data query logic included in the natural language question information according to the minimum parsing unit.

For example, after analyzing the natural language question information “box office income of action movies in North China in August 2017”, the minimum parsing unit obtained includes: North China (attribute minimum parsing unit), action movie (attribute minimum parsing unit), income (Metric minimum parsing unit); space-time modified structure words: August 2017; then further based on these words to determine the data query logic contained in the question information. Specifically, the combination logic in the instruction set is determined according to the meaning and order of the minimum parsing unit.

In the present disclosure, the category or entity to which the natural language question information belongs may be obtained according to the attribute item in the attribute minimum parsing unit, thereby obtaining the category of the natural language question information. In order to accurately query different types of natural language question information, the query logic corresponding to each category of natural language question information may be preset.

The data query logic corresponding to the natural language question information with different problem categories is different. The types of questions in natural language questioning information are: question quantity class, question ratio class, question ranking class, question relationship class, and so on. The quantity class refers to the metric or attribute value in a certain time period or at a certain time point; the question ratio class refers to the ratio of the count of a certain metric or an attribute in different time periods, and the different metrics are in the same time period. The ratio of the rankings; the ranking class refers to sorting the columns (metrics or attributes) according to a certain dimension (the dimension is finally parsed into a filter condition); the relationship in the relational class is predefined or the program is learned through data training. For example, the most dangerous, most valuable, most relevant, nearby, similar, and so on.

There are different query logics for each problem category. For example, the logic corresponding to the quantity class is: time period or time point | logical modification 1 | value 1 | logical modification n | value n |, the combination of the ratio class The mode is: time period | logical modification 1 | numerical value 1 | logical modification n | numerical value n | classification A | classification B | logical modification B | classification N | numerical value 2 | logical modification | calculation modification | (calculation modification) and the like, by obtaining the category corresponding to the natural language question information, based on the query logic corresponding to the preset natural language question information of various categories, the natural language question information corresponding to the obtained category can be obtained. Query logic.

S302: According to the data query logic, extract corresponding instructions from the preset instruction set to combine, and generate a query instruction corresponding to the natural language question information.

For example, based on the above-mentioned natural language question information "the box office revenue of the action movie in North China in August 2017", and the above-mentioned data query logic, the corresponding instruction is extracted from the preset instruction set, and according to the data query logic Combine and construct the query instruction of the question information, such as: sending a query request through a structured query language such as SQL (or Cyper): select sum (box office) from the movie box office table by region, movie category where area = ' North China'and movie category = 'action' and time>=2017-08-01and time <2018-09-01, the query request is composed of a predefined instruction fragment or instruction set, and the database is composed by parsing these instruction sets. Request, filter, calculate, and return data to answer user questions.

S104: Perform a retrieval from a preset knowledge base according to the query instruction, and obtain a data result corresponding to the natural language question information.

Before performing the retrieval from the preset knowledge base according to the query instruction to obtain the data result corresponding to the natural language question information, the following steps are also included, as shown in FIG. 4:

S401: Acquire knowledge base sample data; the knowledge base sample data includes: user-provided database data, user input information data, and/or third-party data.

S402: Generate a preset knowledge base according to the knowledge base sample data.

The specific knowledge base generation process is not specifically limited. For example, a deep learning model based on a convolutional neural network may be used for establishment, or other methods may be employed.

In order to improve the accuracy of the query result, the method provided by the present disclosure further includes the following steps, as shown in FIG. 5:

S501: Add the natural language question information and the corresponding data result to the preset knowledge base.

Through the above steps, the data in the preset knowledge base can be continuously updated, so that the content in the preset knowledge base is more and more rich, so that the query accuracy of the question information can be continuously improved. The resulting data results will change according to the query time and the knowledge base, and it is not a pre-stored answer. For example, if the preset knowledge base stores different data results corresponding to the same natural language question information, the latest time for obtaining the data result for the same natural language question information may be recorded in the preset knowledge base, and the user input is obtained. After the natural language question information, all data results corresponding to the natural language question information are searched from the preset knowledge base, and a data result closest to the obtained time is selected as the final data result corresponding to the natural language question information.

For example, if the preset knowledge base stores different data results corresponding to the same natural language question information, after obtaining the natural language question information input by the user end, the natural language may be found from the preset knowledge base. All the data results corresponding to the question information are output to the user, and a data result selected by the user from all the data results is obtained, and the number of times the data results are selected in all the data results corresponding to the natural language question information is counted in the presence After the data result is selected more than the set highest threshold, the data result is used as the final data result corresponding to the natural language question information.

In the understanding method of the natural language question provided by the present disclosure, the natural language question information is first parsed to obtain a minimum parsing unit, and then the query statement of the natural language question information is constructed based on the minimum parsing unit and the preset instruction set. Then, the data is retrieved from the pre-established knowledge base according to the query statement, and the data result corresponding to the question information is obtained. In the method, the knowledge base is established based on user-provided database data, user input information data, and/or third-party data. It can provide accurate and calculated data results for the question information, so that the method can be applied to professional scenes such as data analysis.

Embodiment 2:

The present disclosure provides an apparatus for understanding natural language questions. Referring to FIG. 6, the apparatus includes an information acquisition module 61, an information analysis module 62, an instruction generation module 63, and a retrieval module 64.

The information obtaining module 61 is configured to obtain the natural language question information input by the user end; the natural language question information is the question information related to the data query; the information parsing module 62 is configured to parse the natural language question information to obtain a minimum parsing unit; The instruction generating module 63 is configured to generate a query instruction corresponding to the natural language question information based on the minimum parsing unit and the preset instruction set; the retrieving module 64 is configured to perform the retrieving from the preset knowledge base according to the query instruction to obtain the natural language question information. Corresponding data results; the preset knowledge base is generated based on user-provided database data, user input information data, and/or third-party data.

Specifically, the information parsing module 62 further includes:

The word segmentation module 621 is configured to perform word segmentation processing on the natural language question information to obtain a plurality of word segment segments; the recognition module 622 is configured to perform entity noun recognition on the plurality of segment word segments to obtain a minimum parsing unit; and the minimum parsing unit includes: attribute minimum parsing The unit, the metric minimum parsing unit, and the structural word are modified in time.

In the apparatus for understanding natural language questions provided by the present disclosure, each module has the same technical features as the above-described method of understanding natural language questions, and therefore, the above functions can also be realized. For the specific working process of each module in the device, refer to the foregoing method embodiment, and details are not described herein again.

Optionally, the attribute minimum parsing unit includes at least one of an attribute item, a calculation operation item, and an attribute logical relationship item; the metric minimum parsing unit includes at least one of a metric item, a metric logical relationship item, and a calculation modification item.

Optionally, the attribute item in the attribute minimum parsing unit represents a category or entity to which the natural language question information belongs; the calculation operation item in the attribute minimum parsing unit represents classification, grouping, segmentation, partial value, counting, pinyin Ranking and ranking by last name stroke; attribute logical relationship items in attribute minimum resolution unit represent logical relationships of similarity, dissimilarity, inclusion, non-containment, comparison, and comparison.

Optionally, the metric item in the metric minimum analytic unit represents a numerical value; the metric logical relationship item in the metric minimum analytic unit represents a numerical relationship of greater than, less than, equal to, greater than or equal to, less than or equal to, and not equal to; The calculated modifiers in the metric minimum parsing unit represent summation, average, count, standard deviation, variance, correlation, correlation coefficient, daily growth rate, weekly growth rate, monthly growth rate, quarterly growth rate, annual growth rate, and ranking. , maximum, minimum, front N, and reciprocal N.

Optionally, the space-time modified structure word represents at least one of the following: a time period, a time point, a point before, a point after, and a point and a point.

Optionally, the instruction generating module 63 is configured to generate, according to the minimum parsing unit and the preset instruction set, a query instruction corresponding to the natural language question information by: inferring the natural according to the minimum parsing unit The data query logic included in the language question information; according to the data query logic, extracting corresponding instructions from the preset instruction set to be combined, and generating a query instruction corresponding to the natural language question information.

Optionally, the instruction generating module 63 is specifically configured to infer, according to the minimum parsing unit, data query logic included in the natural language question information by: obtaining a category corresponding to the natural language question information; The category obtains query logic corresponding to the natural language question information.

Embodiment 3:

The present disclosure also provides an electronic device. As shown in FIG. 7, the electronic device includes a processor 70, a memory 71, a bus 72, and a communication interface 73. The processor 70, the communication interface 73, and the memory 71 are connected by a bus 72. The processor 70 is operative to execute executable modules, such as computer programs, stored in the memory 71. The steps of the method as described in the method embodiments are implemented when the processor executes a computer program. That is, each method step in Embodiment 1 can be performed by the processor 70.

The memory 71 may include a high speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is implemented by at least one communication interface 73 (which may be wired or wireless), and may use an Internet, a wide area network, a local network, a metropolitan area network, or the like.

The bus 72 may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one double-headed arrow is shown in Figure 7, but it does not mean that there is only one bus or one type of bus.

The memory 71 is used to store a program, and the processor 70 executes the program after receiving the execution instruction, and the method executed by the device defined by the flow process disclosed in any of the foregoing embodiments may be applied to the processor 70. Medium or implemented by processor 70.

Processor 70 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method may be completed by an integrated logic circuit of hardware in the processor 70 or an instruction in the form of software. The processor 70 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP processor, etc.), or a digital signal processor (DSP). ), Application Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware component. The methods, steps, and logic blocks in this disclosure may be implemented or performed. The general purpose processor may be a microprocessor or the processor or any conventional processor or the like. The steps of the method provided in connection with the present disclosure may be directly embodied by the completion of the hardware decoding processor or by a combination of hardware and software modules in the decoding processor. The software modules can be located in a conventional storage medium such as random access memory, flash memory, read only memory, programmable read only memory or electrically erasable programmable memory, registers, and the like. The storage medium is located in the memory 71, and the processor 70 reads the information in the memory 71 and performs the steps of the above method in combination with its hardware.

The present disclosure also provides a computer program product for a method of understanding natural language questions, comprising a computer readable storage medium storing non-volatile program code executable by a processor, the program code comprising instructions operable to perform the previous method embodiments For the specific implementation, refer to the method embodiment, and details are not described herein again.

A person skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the device and the electronic device described above can refer to the corresponding process in the foregoing method embodiments, and details are not described herein again.

The flowchart and block diagrams in the Figures illustrate the architecture, functionality and operation of possible implementations of methods and computer program products according to various embodiments of the present disclosure. In this regard, each block of the flowchart or block diagram can represent a module, a program segment, or a portion of code that includes one or more of the Executable instructions. It should also be noted that in some alternative implementations, the functions noted in the blocks may also occur in a different order than that illustrated in the drawings. For example, two consecutive blocks may be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams and/or flowcharts, and combinations of blocks in the block diagrams and/or flowcharts, can be implemented in a dedicated hardware-based system that performs the specified function or function. Or it can be implemented by a combination of dedicated hardware and computer instructions.

In the description of the present disclosure, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. The orientation or positional relationship indicated is based on the orientation or positional relationship shown in the drawings, and is merely for the convenience of describing the present disclosure and the simplified description, and does not indicate or imply that the device or component referred to has a specific orientation, in a specific orientation. The construction and operation are therefore not to be construed as limiting the disclosure. Moreover, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

In the several embodiments provided by the present application, it should be understood that the disclosed systems, devices, and methods may be implemented in other manners. The device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner. For example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some communication interface, device or unit, and may be electrical, mechanical or otherwise.

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

In addition, each functional unit in various embodiments of the present disclosure 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 functions, if implemented in the form of software functional units and sold or used as separate products, may be stored in a non-transitory computer readable storage medium executable by a processor. Based on such understanding, the portion of the technical solution of the present disclosure that contributes in essence or to the prior art or the portion of the technical solution may be embodied in the form of a software product stored in a storage medium, including The instructions are used to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present disclosure. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

It should be noted that the above-mentioned embodiments are merely specific embodiments of the present disclosure, and are used to explain the technical solutions of the present disclosure, and are not limited thereto. The scope of protection of the present disclosure is not limited thereto, although reference is made to the foregoing. The embodiments are described in detail, and those skilled in the art should understand that any one skilled in the art can still modify the technical solutions described in the foregoing embodiments within the technical scope of the disclosure. The changes may be easily conceived, or equivalents may be substituted for some of the technical features; and the modifications, variations, or substitutions of the present invention are not intended to depart from the spirit and scope of the technical solutions of the present disclosure, and should be covered in the protection of the present disclosure. Within the scope. Therefore, the scope of protection of the present disclosure should be determined by the scope of the claims.

Industrial applicability

The method, device and electronic device for understanding natural language questions provided by the present disclosure can provide accurate and statistically obtained data results for question information, and realize accurate identification of natural language question information, and can be applied to data analysis fields, etc. Professional scenes.

Claims (20)

A method for understanding natural language questions, which is characterized by comprising: Obtaining natural language question information input by the user end; the natural language question information is question information related to the data query; Parsing the natural language question information to obtain a minimum parsing unit; Generating, according to the minimum parsing unit and the preset instruction set, a query instruction corresponding to the natural language question information; Performing a search from the preset knowledge base according to the query instruction to obtain a data result corresponding to the natural language question information; the preset knowledge base is based on database data provided by the user, user input information data, and/or third party data. generate. The method according to claim 1, wherein the parsing the natural language question information to obtain a minimum parsing unit comprises: Performing word segmentation processing on the natural language question information to obtain a plurality of word segmentation segments; Performing entity noun recognition on the plurality of segmentation segments to obtain a minimum parsing unit; the minimum parsing unit includes: an attribute minimum parsing unit, a metric minimum parsing unit, and a space-time modifying structural word. The method according to claim 2, wherein the attribute minimum parsing unit comprises at least one of an attribute item, a calculation operation item, and an attribute logical relationship item; the metric minimum parsing unit comprises a metric item, a metric logical relationship item And calculate at least one of the modifiers. The method according to claim 3, wherein the attribute item in the attribute minimum parsing unit represents a category or entity to which the natural language question information belongs; the calculation operation item in the attribute minimum parsing unit represents classification, grouping, Segmentation, partial value, count, ranking by pinyin and ranking by last name stroke; attribute logical relationship terms in the minimum parsing unit of the attribute represent logical relationships of similarity, dissimilarity, inclusion, non-containment, comparison and comparison. The method according to claim 3 or 4, wherein the metric item in the metric minimum analytic unit represents a numerical value; the metric logical relationship item in the metric minimum analytic unit represents greater than, less than, equal to, greater than or equal to, a numerical relationship of less than or equal to and not equal to; the calculated modifier in the minimum parsing unit of the metric represents summation, average, count, standard deviation, variance, correlation, correlation coefficient, daily growth rate, weekly growth rate, monthly growth Rate, quarterly growth rate, annual growth rate, sort, maximum, minimum, pre-N and reciprocal N. The method according to any one of claims 2 to 5, wherein the space-time modified structure word represents at least one of the following: a time period, a time point, a point before, after a point, and a point Between a certain point and a preset position within a certain distance, and a relative position within a certain distance. The method according to any one of claims 1 to 6, wherein the generating the query instruction corresponding to the natural language question information based on the minimum parsing unit and the preset instruction set comprises: Deriving data query logic included in the natural language question information according to the minimum parsing unit; According to the data query logic, corresponding instructions are extracted from the preset instruction set and combined to generate a query instruction corresponding to the natural language question information. The method according to claim 7, wherein the inferring the data query logic included in the natural language question information according to the minimum parsing unit comprises: Obtaining a category corresponding to the natural language question information; According to the obtained category, the query logic corresponding to the natural language question information is obtained. The method according to any one of claims 1 to 8, wherein before the retrieving from the preset knowledge base according to the query instruction to obtain the data result corresponding to the natural language question information, The method also includes: Obtaining knowledge base sample data; the knowledge base sample data includes: user-provided database data, user input information data, and/or third-party data; A preset knowledge base is generated according to the knowledge base sample data. The method according to any one of claims 1 to 9, wherein after the searching according to the query instruction is performed from a preset knowledge base to obtain a data result corresponding to the natural language question information, The method also includes: The natural language question information and its corresponding data result are added to the preset knowledge base. An apparatus for understanding natural language questions, comprising: The information obtaining module is configured to obtain natural language question information input by the user end; the natural language question information is question information related to the data query; An information parsing module configured to parse the natural language question information to obtain a minimum parsing unit; An instruction generating module, configured to generate a query instruction corresponding to the natural language question information based on the minimum parsing unit and a preset instruction set; a retrieval module, configured to retrieve from the preset knowledge base according to the query instruction, to obtain a data result corresponding to the natural language question information; the preset knowledge base is based on database data provided by the user, input information data of the user, and / or third party data generation. The device according to claim 11, wherein the information parsing module comprises: a word segmentation module configured to perform word segmentation processing on the natural language question information to obtain a plurality of word segmentation segments; The identification module is configured to perform entity noun recognition on the plurality of segmentation segments to obtain a minimum parsing unit; the minimum parsing unit includes: an attribute minimum parsing unit, a metric minimum parsing unit, and a space-modifying structural word. The apparatus according to claim 12, wherein the attribute minimum parsing unit comprises at least one of an attribute item, a calculation operation item, and an attribute logical relationship item; the metric minimum parsing unit comprises a metric item, a metric logical relationship item And calculate at least one of the modifiers. The apparatus according to claim 13, wherein the attribute item in the attribute minimum parsing unit represents a category or entity to which the natural language question information belongs; the calculation operation item in the attribute minimum parsing unit represents classification, grouping, Segmentation, partial value, count, ranking by pinyin and ranking by last name stroke; attribute logical relationship terms in the minimum parsing unit of the attribute represent logical relationships of similarity, dissimilarity, inclusion, non-containment, comparison and comparison. The apparatus according to claim 13 or 14, wherein the metric item in the metric minimum parsing unit represents a numerical value; and the metric logical relationship item in the metric minimum parsing unit represents greater than, less than, equal to, greater than or equal to, a numerical relationship of less than or equal to and not equal to; the calculated modifier in the minimum parsing unit of the metric represents summation, average, count, standard deviation, variance, correlation, correlation coefficient, daily growth rate, weekly growth rate, monthly growth Rate, quarterly growth rate, annual growth rate, sort, maximum, minimum, pre-N and reciprocal N. The apparatus according to any one of claims 12 to 15, wherein the space-time modified structure word represents at least one of the following: a time period, a time point, a point before, a point after, and a certain Between the point and a point; somewhere outside the preset distance range, somewhere within the preset distance range and somewhere set the direction relative position. The device according to any one of claims 11 to 16, wherein the instruction generating module is configured to generate, according to the minimum parsing unit and the preset instruction set, corresponding to the natural language question information Query instructions: Deriving data query logic included in the natural language question information according to the minimum parsing unit; According to the data query logic, corresponding instructions are extracted from the preset instruction set and combined to generate a query instruction corresponding to the natural language question information. The apparatus according to claim 17, wherein the instruction generating module is specifically configured to infer, according to the minimum parsing unit, data query logic included in the natural language question information by: Obtaining a category corresponding to the natural language question information; According to the obtained category, the query logic corresponding to the natural language question information is obtained. An electronic device comprising a memory and a processor having stored thereon a computer program executable on the processor, wherein the processor executes the computer program to implement the above claims 1 to 10 The steps of any of the methods described. A computer readable medium having a processor-executable non-volatile program code, wherein the program code causes the processor to perform the method of any one of claims 1 to 10.

Images