Application question solving knowledge base construction method and device, and application question solving robot

Abstract

The application provides a construction method and device of an application problem solving knowledge base and an application problem solving robot, and the method comprises the following steps: constructing a mathematical knowledge sub-library and constructing a common sense knowledge sub-library; combining the mathematical knowledge sub-library and the common sense knowledge sub-library to form an application problem solving knowledge base, and configuring corresponding query interfaces for the mathematical knowledge sub-library and the common sense knowledge sub-library respectively. The application fully considers that the knowledge content related to mathematical application problem solving not only contains mathematical knowledge content, but also contains common sense knowledge content, and the two parts of knowledge content have great differences, so the application problem solving knowledge base is designed as two sub-libraries, different methods are used to acquire and construct the two sub-libraries according to different characteristics of the two sub-libraries, the acquired application problem solving knowledge base can effectively and accurately assist the development of an application problem automatic solving system, so that the problem solving accuracy of the application problem automatic solving system can be greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a construction method and device of application problem solving knowledge base and application problem solving robot. BACKGROUND

[0002] Automatic problem solving technology as an application of artificial intelligence has attracted great interest of the academic and industrial circles in recent years. On the one hand, automatic problem solving technology has a high degree of demonstration, which can enable ordinary people to easily directly appreciate the specific effect of the related technology, thereby intuitively demonstrating the progress of artificial intelligence technology. On the other hand, automatic problem solving technology has great practical value in the fields of education and various teaching aids and other specific applications.

[0003] In the automatic problem solving task, the automatic problem solving technology for elementary mathematics application problems (mainly involving primary school, junior high school and high school levels) is an important application point, and there are many academic researches and practical products for this task.

[0004] In the automatic solving system of mathematical application problems, the application problem solving knowledge base is an important component module. It is usually used to provide the knowledge required to understand the problem. Since there is a difference between the instructions related to mathematical application problems and ordinary knowledge, the construction of the application problem solving knowledge base and the construction of the ordinary knowledge base must be significantly different, so when constructing the application problem solving knowledge base, a targeted construction method needs to be taken according to the knowledge characteristics involved in the mathematical application problems. SUMMARY

[0005] The present application provides a construction method and device of application problem solving knowledge base and application problem solving robot, which solves the problem that the prior art uses the method of constructing ordinary knowledge base to construct the application problem solving knowledge base, and fully considers the difference of the knowledge content of mathematical application problems, which can effectively develop the application problem solving system.

[0006] In a first aspect, the present application provides a construction method of application problem solving knowledge base, comprising:

[0007] constructing a mathematical knowledge sub-base and a common sense knowledge sub-base;

[0008] combining the mathematical knowledge sub-base and the common sense knowledge sub-base to form an application problem solving knowledge base, and configuring a corresponding query interface for the mathematical knowledge sub-base and the common sense knowledge sub-base respectively.

[0009] According to the construction method of application problem solving knowledge base provided by the present application, the construction of the mathematical knowledge sub-base specifically comprises:

[0010] The authority material is mined, and a mathematical knowledge entity set related to three types of mathematical knowledge entities is acquired, and a skeleton of the mathematical knowledge sub-library is built;

[0011] The three types of mathematical knowledge entities include mathematical knowledge topics, mathematical knowledge points and mathematical knowledge concepts; the authority material includes documents, books and test papers related to teaching materials and examination syllabus;

[0012] The authority material and various teaching aid materials are mined, and a mathematical knowledge content set related to the three types of mathematical knowledge entities is acquired, so that the skeleton of the mathematical knowledge sub-library is filled by using the mathematical knowledge content set, and the mathematical knowledge sub-library is created.

[0013] According to the application, a construction method of an application question solving knowledge library is provided, the authority material is mined, and a mathematical knowledge entity set related to three types of mathematical knowledge entities is acquired, and a skeleton of the mathematical knowledge sub-library is built, including:

[0014] Based on the OCR technology, the paper version of the authority material is converted into an electronic document;

[0015] The meaningless data part in the electronic document is removed, and data cleaning of the electronic document is realized;

[0016] Based on the man-machine mixed mode, the mathematical knowledge entity set is extracted from the electronic document after data cleaning, so as to build the skeleton of the mathematical knowledge sub-library.

[0017] According to the application, a construction method of an application question solving knowledge library is provided, and the method further includes:

[0018] A new version of the authority material is acquired, and the mathematical knowledge sub-library is expanded according to new mathematical knowledge content extracted from the new version of the authority material;

[0019] And / or, unprocessed teaching aid materials are acquired, and the mathematical knowledge sub-library is expanded according to new mathematical knowledge content extracted from the unprocessed teaching aid materials;

[0020] And / or, missing mathematical knowledge content in the mathematical knowledge sub-library is acquired, so that the missing mathematical knowledge content is expanded to the mathematical knowledge sub-library.

[0021] According to the application, a construction method of an application question solving knowledge library is provided, and the method further includes:

[0022] The concept type common sense content set and other common sense content sets are constructed, so that the common sense knowledge sub-library is formed by integrating the concept type common sense content set and the other common sense content sets.

[0023] The other common sense content set includes a numerical common sense content set, a state common sense content set and a homonym common sense content set.

[0024] According to the application, a construction method of an application problem solving knowledge base is provided.

[0025] Based on the open knowledge source architecture and in combination with the candidate concept set extracted from a large number of application problem examples, a concept hierarchical framework is built.

[0026] Candidate target attributes related to each candidate concept in the concept hierarchical framework are obtained, and a candidate target attribute set is constructed.

[0027] According to the candidate target attribute set, a target attribute value is given to the target attribute of each candidate concept in the concept hierarchical framework, so as to obtain a plurality of triples composed of a candidate concept, a target attribute and a target attribute value.

[0028] A triple set containing all the triples is composed of the concept common sense content set.

[0029] According to the application, the open knowledge source includes at least one of Schema.org, Wikipedia, HowNet and network encyclopedias.

[0030] According to the application, a construction method of an application problem solving knowledge base is provided.

[0031] Numerical common sense content, state common sense content and homonym common sense content are extracted from open data sources through heuristic rules, and the numerical common sense content set, the state common sense content set and the homonym common sense content set are respectively constructed.

[0032] And / or, numerical common sense content, state common sense content and homonym common sense content are obtained from search engines through template retrieval, and the numerical common sense content set, the state common sense content set and the homonym common sense content set are respectively constructed.

[0033] And / or, based on Chinese word segmentation, named entity recognition and noun phrase recognition, key words are extracted from a large number of mathematical application problem examples, and numerical common sense content, state common sense content and homonym common sense content are obtained, and the numerical common sense content set, the state common sense content set and the homonym common sense content set are respectively constructed.

[0034] In a second aspect, the application further provides a construction device of an application problem solving knowledge base, comprising:

[0035] a data collection processing module, configured to collect data to construct a mathematical knowledge sub-database and a common sense knowledge sub-database respectively;

[0036] a query interface configuration module, configured to combine the mathematical knowledge sub-database and the common sense knowledge sub-database to form an application question solving knowledge database, and to configure corresponding query interfaces for the mathematical knowledge sub-database and the common sense knowledge sub-database respectively.

[0037] In a third aspect, the present application provides an application question solving robot, comprising an application question solving robot body, wherein a reading unit and an output unit are arranged in the application question solving robot body;

[0038] Further comprising a memory and a program or instruction stored on the memory and executable on the processor;

[0039] The reading unit reads an application question to be solved, so that the program or instruction is executed by the processor to access the application question solving knowledge database, match a query to an answer result, and output the answer result via the output unit.

[0040] The application question solving knowledge database is pre-created based on the steps of the construction method of the application question solving knowledge database.

[0041] In a fourth aspect, the present application provides an electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the construction method of the application question solving knowledge database according to any of the above aspects when executing the program.

[0042] In a fifth aspect, the present application further provides a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executable on a processor to implement the steps of the construction method of the application question solving knowledge database according to any of the above aspects.

[0043] The construction method, device and robot of the application question solving knowledge database provided by the present application fully consider that the knowledge content related to the solving of mathematical application questions not only contains mathematical knowledge content, but also contains common sense knowledge content, and the two parts of knowledge content have great differences, so the application question solving knowledge database is designed as two sub-databases, different methods are used to acquire and construct the contents of the two sub-databases according to their different characteristics, and the acquired application question solving knowledge database can effectively and accurately assist the development of an application question automatic solving system, thereby greatly improving the solving accuracy of the application question automatic solving system. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings described below are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0045] Figure 1 is a flowchart of the method for constructing the application problem solving knowledge base provided by the present application;

[0046] Figure 2 is one of the flowcharts of the method for constructing the mathematical knowledge base provided by the present application;

[0047] Figure 3 is the second flowchart of the method for constructing the mathematical knowledge base provided by the present application;

[0048] Figure 4 is a structural schematic diagram of the conceptual upper-lower frame provided by the present application;

[0049] Figure 5 is a schematic diagram of the information box of the corresponding article in Baidu Encyclopedia for the candidate concept provided by the present application;

[0050] Figure 6 is a schematic diagram of the three types of common sense content obtained by the template retrieval provided by the present application;

[0051] Figure 7 is a structural schematic diagram of the construction device of the application problem solving knowledge base provided by the present application;

[0052] Figure 8 is a structural schematic diagram of the electronic device provided by the present application. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of the present application more clear, the technical solutions of the present application will be described clearly and completely in the following with reference to the drawings in the present application. Obviously, the described embodiments are some embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by those skilled in the art without creative labor belong to the protection scope of the present application.

[0054] It should be noted that in the description of the embodiments of the present application, the terms "comprising", "containing" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or device. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of other identical elements in the process, method, article or device comprising the element. The terms "upper", "lower" and the like indicate the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the present application and simplify the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. Unless otherwise specified and limited, the terms "mounting", "connection", "connection" should be broadly understood, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, or it can be connected inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0055] The terms "first", "second", and the like in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are usually a class, and do not limit the number of objects, for example, the first object can be one or more. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / ", generally indicates that the front and rear associated objects are in an "or" relationship.

[0056] The following will be described in conjunction with Figures 1-8 The application problem solving knowledge base construction method, device and application problem solving robot provided by the embodiments of the present application are described.

[0057] Figure 1 The flowchart of the application problem solving knowledge base construction method provided by the present application is shown in Figure 1 The following steps are included, but are not limited to:

[0058] Step 101: Constructing a mathematical knowledge base and constructing a common sense knowledge base.

[0059] It needs to be noted that the application problems appearing in the subsequent embodiments of the present application mainly refer to mathematical application problems. The solving of mathematical application problems embodies both strong field and rich background knowledge, and requires special mathematical knowledge and common sense knowledge. The following will be described through an application problem example:

[0060] Application problem title: What is the probability that the sum of two points obtained by throwing two uniform dice is 8?

[0061] Answer: There are 36 different results when throwing two dice, and there are 5 cases in which the sum is 8, so the probability is 5 / 36.

[0062] The application problem title itself is not difficult, but to solve it, both mathematical knowledge and common sense knowledge are required.

[0063] Firstly, the mathematical knowledge about the application problem mainly involves the following two knowledge contents (hereinafter referred to as knowledge):

[0064] Knowledge 1, definition of probability: probability is the possibility of occurrence of a random event;

[0065] Knowledge 2, classical probability model calculation formula: probability (A) is equal to the number of basic events contained in A / the total number of basic events n.

[0066] In addition, the common sense knowledge about the application problem mainly involves the following two knowledge:

[0067] Knowledge 1: a die has 6 faces;

[0068] Knowledge 2: the point numbers of each face of the die are 1, 2, 3, 4, 5 and 6.

[0069] Since different specific knowledge is required to solve different application problems, the application problem title in the above application problem example is a probability problem, and other application problem titles can be trigonometric function problems or solid geometry problems, etc., so in order to support the solving of application problems involving different knowledge points and different types, a unified and complete application problem solving knowledge base needs to be constructed. The application problem solving knowledge base should preferably include the mathematical knowledge specified in the mathematical syllabus, and in addition, it should also include rich common sense knowledge related thereto.

[0070] In view of the above, the application provides a method for constructing an application problem solving knowledge base, and the constructed application problem solving knowledge base is mainly composed of a mathematical knowledge sub-base and a common sense knowledge sub-base, so different methods need to be used to acquire knowledge content according to the different knowledge contained in the mathematical knowledge sub-base and the common sense knowledge sub-base, in order to independently construct the mathematical knowledge sub-base and the common sense knowledge sub-base.

[0071] Step 102: The mathematical knowledge sub-library and the common sense knowledge sub-library are combined to form an application question solving knowledge library, and the corresponding query interfaces are configured for the mathematical knowledge sub-library and the common sense knowledge sub-library respectively.

[0072] After the mathematical knowledge and the common sense knowledge related to application question solving are collected to form the mathematical knowledge sub-library and the common sense knowledge sub-library respectively, the two are combined to form the application question solving knowledge library. Since the mathematical knowledge sub-library and the common sense knowledge sub-library are two independent parts, the corresponding query interfaces can be developed for the two sub-libraries according to actual operation requirements, and the user can query all the mathematical knowledge and common sense knowledge related to the application question to be solved through the corresponding query interfaces. Therefore, the application question solving knowledge library generated by the application can be used to support the development of the mathematical application question solving system.

[0073] The application provides the application question solving knowledge library construction method, which fully considers that the knowledge content related to mathematical application question solving not only includes mathematical knowledge content, but also includes common sense knowledge content, and the two parts of knowledge content have great differences. Therefore, the application question solving knowledge library is designed as two sub-libraries, different methods are used to acquire and construct the content of the two sub-libraries according to different characteristics of the two sub-libraries, and the acquired application question solving knowledge library can effectively and accurately assist the development of the application question automatic solving system, so that the problem solving accuracy of the application question automatic solving system can be greatly improved.

[0074] Figure 2 is one of the flowcharts for constructing a mathematical knowledge sub-library provided by the application, as shown in Figure 2 The method for constructing a mathematical knowledge sub-library specifically includes but is not limited to the following steps:

[0075] Step 201: The authoritative materials are mined to acquire a mathematical knowledge entity set related to three types of mathematical knowledge entities, and a skeleton of the mathematical knowledge sub-library is built.

[0076] The three types of mathematical knowledge entities include mathematical knowledge themes, mathematical knowledge points and mathematical knowledge concepts; and the authoritative materials include documents, books and test papers related to textbooks and examination outlines.

[0077] Step 202: The authoritative materials and various teaching aid materials are mined to acquire a mathematical knowledge content set related to the three types of mathematical knowledge entities, so as to fill the skeleton of the mathematical knowledge sub-library with the mathematical knowledge content set and create the mathematical knowledge sub-library.

[0078] Specifically, the mathematical knowledge contained in the mathematical knowledge sub-library provided by the present application mainly involves three types of mathematical knowledge entities: mathematical knowledge topics (hereinafter referred to as knowledge topics), mathematical knowledge points (hereinafter referred to as knowledge points), and mathematical knowledge concepts (hereinafter referred to as knowledge concepts).

[0079] Overall, the three types of mathematical knowledge entities are from general to specific, among which the knowledge topic is the most macroscopic and contains the richest content, and the knowledge concept is the most microscopic and contains the most specific content.

[0080] The first type is the definition of the knowledge topic:

[0081] The present application organizes the knowledge content involved in elementary mathematics through multiple different knowledge topics, such as "set", "basic elementary function", and "spatial geometric body", etc.

[0082] It should be noted that this organization method adopted by the present application is similar to the organization of content through different chapters in professional books, but there are also differences, mainly in that the division of knowledge topics in professional books is carried out from a logical rather than a formal perspective. A knowledge topic containing more content may be divided into multiple chapters, but in the present application they are all classified as the same knowledge topic.

[0083] The second type is the definition of the knowledge point:

[0084] Generally speaking, a knowledge topic includes multiple knowledge points, and a knowledge point is a summary of a part of the content in the knowledge topic. Multiple knowledge points of the same knowledge topic have certain relevance (because they all belong to the same knowledge topic), but also have obvious differences (otherwise they can be classified as the same knowledge point).

[0085] Table 1 is an example table of a knowledge point provided by the present application, as shown in Table 1, the examples of three knowledge points contained in the knowledge topic "set" are as follows:

[0086] Table 1 Example table of knowledge points

[0087] Knowledge point 1: Understand the meaning of set, element and the belonging relationship between set Knowledge point 2: Can use natural language, graphical language, set language to describe different specific problems Knowledge point 3: Understand the meaning of inclusion and equality between sets, and can identify the subset of given set

[0088] The third type is the definition of the knowledge concept:

[0089] A knowledge concept is a description of a specific knowledge content under a certain knowledge topic. For example, "empty set" is a knowledge concept, which belongs to the knowledge topic "set", and it has relevance with the knowledge point "understand the meaning of the universal set and the empty set in specific contexts".

[0090] Further, the mathematical indicator sub-library constructed by the present application adds three kinds of subordination to the above three kinds of mathematical knowledge entities. The first subordination is "knowledge concept - subordinate to - knowledge theme", the second subordination is "knowledge concept - subordinate to - knowledge point", and the third subordination is "knowledge point - subordinate to - knowledge theme". The setting of these subordinations is to be able to find related knowledge concepts through knowledge points or knowledge themes as media when querying the knowledge base, so as to provide more abundant knowledge for problem solving.

[0091] It should be emphasized that the above knowledge concept has specific attributes. For any knowledge concept, the present application designs two attributes to describe it. The two attributes are "definition" and "type".

[0092] Table 2 is an example table of attributes of a knowledge concept provided by the present application, as shown in Table 2, the "definition" attribute is the definition of the corresponding knowledge concept, and the "type" attribute describes the category of the knowledge concept.

[0093] knowledge concept <-> attribute name <-> attribute value empty set <-> a set with no elements is called the empty set A empty set <-> type <-> noun definition

[0094] Table 2 is an example table of attributes of a knowledge concept provided by the present application, as shown in Table 2, the "definition" attribute is the definition of the corresponding knowledge concept, and the "type" attribute describes the category of the knowledge concept.

[0095] Table 3 is an example table of types of a knowledge concept provided by the present application, as shown in Table 3, in the mathematical knowledge sub-library constructed by the present application, any knowledge concept can have four types, such as: noun definition, verb definition, fact and rule.

[0096]

[0097] Table 3 is an example table of types of a knowledge concept provided by the present application, as shown in Table 3, in the mathematical knowledge sub-library constructed by the present application, any knowledge concept can have four types, such as: noun definition, verb definition, fact and rule.

[0098] Figure 3 is a second flowchart of constructing a mathematical knowledge sub-library provided by the present application, as shown in Figure 3 Based on the above definition, the present application provides a method for acquiring knowledge content related to the mathematical knowledge sub-library, which mainly includes the following steps:

[0099] Step 301, mining the authoritative materials to construct a mathematical knowledge entity set composed of three kinds of mathematical knowledge entities such as knowledge subject, knowledge point and knowledge concept.

[0100] Among them, the authoritative materials mainly include documents, books and test papers published by official education institutions such as high school entrance examination outline and general teaching materials. The goal of this step is to mine the mathematical knowledge entity set such as knowledge theme, knowledge point and knowledge concept from it, so as to build the skeleton of the mathematical knowledge sub-library.

[0101] As an optional embodiment, the authority material is mined to obtain a mathematical knowledge entity set related to the three types of mathematical knowledge entities, and a skeleton of the mathematical knowledge sub-library is built, which mainly includes, but is not limited to, the following steps:

[0102] Based on the optical character recognition (OCR) technology, the paper version of the authority material is converted into an electronic document;

[0103] The meaningless data part in the electronic document is removed to realize data cleaning of the electronic document;

[0104] Based on the human-computer hybrid mode, the mathematical knowledge entity set is extracted from the data cleaned electronic document to build the skeleton of the mathematical knowledge sub-library.

[0105] As an optional embodiment, the specific process of mining the authority material includes:

[0106] Step 1.1, authority material electronicization: if the authority material is a paper version, it can be converted into an electronic document through OCR and other technologies.

[0107] Step 1.2, data cleaning: the cleaning operation includes removing random codes and meaningless parts in the data, such as covers.

[0108] Step 1.3, text content extraction: the valuable content in the authority material is extracted, and the main method is the human-computer hybrid method. First, the automatic method (such as writing extraction rules) is used to extract the three types of mathematical knowledge entities from the electronic document, and then the manual verification is performed.

[0109] Further, it further includes step 302 of extracting and mining specific mathematical knowledge content from various teaching aids.

[0110] In a popular way, this step is to extract and mine specific mathematical knowledge content from authority materials and various auxiliary materials (such as teaching aids), and the goal is to fill related specific knowledge content (which can be understood as flesh and blood) around the skeleton of the mathematical knowledge sub-library built in step 301.

[0111] As an optional embodiment, the specific process of step 302 includes:

[0112] Step 2.1, original data acquisition: including material electronicization, related web page data crawling.

[0113] Step 2.2, data cleaning: removing meaningless content, such as web page format, etc.

[0114] Step 2.3, knowledge content extraction and filling: the main method adopted is a man-machine mixed method. First, the relevant knowledge is extracted from the text by using an automatic method (such as writing extraction rules), and then it is manually checked.

[0115] Further, step 303 can be further included to supplement the knowledge content contained in the mathematical knowledge base to realize expansion of the mathematical knowledge base, mainly including but not limited to the following steps:

[0116] Obtaining a new version of authoritative material, expanding the mathematical knowledge base according to new mathematical knowledge content extracted from the new version of the authoritative material;

[0117] And / or, obtaining unprocessed teaching aid material, expanding the mathematical knowledge base according to new mathematical knowledge content extracted from the unprocessed teaching aid material;

[0118] And / or, obtaining missing mathematical knowledge content in the mathematical knowledge base to expand the missing mathematical knowledge content to the mathematical knowledge base.

[0119] Specifically, the above method of expanding the mathematical knowledge base includes the following three methods:

[0120] Method 1, adjusting the knowledge content in the mathematical knowledge base according to the new version of the authoritative material.

[0121] Method 2, processing the unprocessed auxiliary material and extracting new knowledge content;

[0122] Method 3, supplementing the originally missing content in the mathematical knowledge base according to the actual application situation.

[0123] Through the above method, the present application can construct a rich mathematical knowledge base. In order to ensure the correctness of the knowledge content in the mathematical knowledge base, the content is manually checked when necessary.

[0124] Based on the content of the above embodiment, as an optional embodiment, the present application provides a method for constructing a common knowledge base, which specifically includes but is not limited to:

[0125] Constructing a concept type common content set and constructing other common content sets to constitute the common knowledge base by integrating the concept type common content set and the other common content sets.

[0126] The other common content sets include a numerical type common content set, a state type common content set and a same position type common content set.

[0127] Currently, there is much research on common sense knowledge in academia and industry, but no definitive conclusions have been reached. The common sense knowledge sub-base provided by this invention has a clear domain (mathematics) and purpose (applied to solving word problems), rather than being an all-encompassing encyclopedic common sense base. This determines the characteristics of this invention: the construction of a common sense knowledge sub-base targeted at a specific domain and specific goal.

[0128] Specifically, the construction of the common sense knowledge sub-base provided by this invention mainly follows the following two rules:

[0129] (1) The organizational structure is based on the hierarchical relationship of concepts, and all the concepts involved can be categorized into a hierarchical system.

[0130] (2) Concepts and related knowledge should all revolve around the task of solving word problems. Within this scope, try to expand the coverage as much as possible, but do not try to be too comprehensive.

[0131] Table 4. Examples of common sense content included in the common sense knowledge sub-base

[0132]

[0133] Table 4 is an example table of common sense content contained in a common sense knowledge sub-base provided by the present invention. As shown in Table 4, the common sense content contained in the common sense knowledge sub-base provided by the present invention is mainly divided into the following two categories: a set of conceptual common sense content composed of conceptual common sense content; and a set of other common sense content composed of numerical common sense content set, state common sense content set, and parity common sense content set, etc.

[0134] It is particularly important to clarify the difference between state-based common sense content and equivalent common sense content. "State" represents all possible values ​​that a conceptual common sense content can take from a certain perspective. For example, "traffic light" can only be one of "red, yellow, or green." "Equivalent," on the other hand, refers to some possible values ​​that a conceptual common sense content can take from a certain perspective, but other possible values ​​exist that are not included in the equivalent set for some reason. For example, "product" in Table 4, from the perspective of product grade, could also be "fourth-grade product," "fifth-grade product," etc., in addition to the examples given.

[0135] The following describes a method for constructing a conceptual common sense content set provided by this invention, which includes, but is not limited to, the following steps:

[0136] Based on the architecture of open knowledge sources, and combined with the set of candidate concepts extracted from a large number of application problem examples, a hierarchical framework of concepts is built.

[0137] Obtain candidate target attributes related to each candidate concept in the hierarchical framework of the aforementioned concept, and construct a candidate target attribute set;

[0138] According to the candidate target attribute set, a target attribute value is assigned to the target attribute of each candidate concept in the concept hierarchical framework, respectively, to obtain a plurality of triples composed of a candidate concept, a target attribute and a target attribute value;

[0139] A triple set containing all the triples is formed into the concept type common sense content set.

[0140] Specifically, the above steps are as follows:

[0141] Step 1: Construct a concept hierarchical framework.

[0142] Figure 4 is a structural schematic diagram of a concept hierarchical framework provided by the present application, as Figure 4 shown, the concept hierarchical framework is a structure similar to a tree, the top of which is the topmost concept (for example, "entity"), the second layer is the child concept of the top layer (for example, "living beings", "non-living beings"), the third layer is the child concept of the second layer (for example, the child concepts of "living beings" include "animals", "plants", "humans", etc.), and so on.

[0143] Among them, the basic strategy for constructing the concept hierarchical framework is to make full use of the existing public knowledge system, combined with the actual scene of solving mathematical application problems, to obtain the final concept hierarchical framework.

[0144] Specifically, in the method provided by the present application, two sub-steps are taken to construct the concept hierarchical framework:

[0145] Step 1.1, referring to the framework of public knowledge sources such as Schema.org, Wikipedia, HowNet, network encyclopedias, etc., to construct all layers in the hierarchical framework except the bottom layer.

[0146] Step 1.2, using Chinese word segmentation, named entity recognition and noun phrase recognition methods to process large-scale mathematical application problems, and the processing results are used as a candidate concept set, and then the candidate concepts are expanded to obtain the bottom layer of the concept hierarchical framework.

[0147] After the construction of the concept hierarchical framework is realized, the construction of the concept type common sense content set also includes step 2: obtaining the special attribute value class knowledge of the concept, and the above step 2 can be specifically divided into two sub-steps.

[0148] Step 2.1, construct a target attribute set, which is specifically divided into two steps:

[0149] The first step is to obtain a candidate target attribute set according to each candidate concept in the concept hierarchical framework, from the information box (infobox) of the corresponding article in the network encyclopedia China.Figure 5 is a schematic diagram of the information box of the corresponding article of the candidate concept provided by the present application, as shown in Figure 5 is an infobox obtained by inputting the candidate concept "Beijing" into Baidu Encyclopedia, wherein Figure 5 the first column and the third column can be used as the candidate target attributes.

[0150] The second step is to manually supplement and delete the above results to obtain the final attribute set.

[0151] Step 2.2, constructing a knowledge triple (hereinafter referred to as triple). The basic form of the triple is "concept-attribute name-attribute value".

[0152] In the present application, the concept set is constructed by the common sense knowledge base construction step 1, and the target attribute set is constructed by step 2.1. In this step 2.2, the target attribute value is assigned to the target attribute of the concept, so as to obtain the "concept-target attribute-target attribute value" triple.

[0153] Further, the specific implementation method of the above step 2.2 is also divided into two steps:

[0154] The first step is to directly find and obtain the corresponding triple from the public data source by matching;

[0155] The second step is to supplement and verify the above results to obtain the final triple set.

[0156] As an optional embodiment, the present application also provides a method step for constructing other common sense content sets, which specifically includes but is not limited to:

[0157] Extracting numerical common sense content, state common sense content and same position common sense content from open data sources by heuristic rules to form the numerical common sense content set, the state common sense content set and the same position common sense content set, respectively;

[0158] And / or, obtaining numerical common sense content, state common sense content and same position common sense content from search engines by template retrieval to form the numerical common sense content set, the state common sense content set and the same position common sense content set, respectively;

[0159] And / or, based on Chinese word segmentation, named entity recognition and noun phrase recognition, extracting keywords from a large number of mathematical application examples to obtain numerical common sense content, state common sense content and same position common sense content to form the numerical common sense content set, the state common sense content set and the same position common sense content set, respectively.

[0160] With the examples of the numerical common sense content, the state common sense content and the co-located common sense content shown in Table 4, it can be found that the three types of common sense content have strong domain, and thus the corresponding acquisition methods have corresponding characteristics, mainly including the following three types:

[0161] The first type is to extract the three types of common sense content from the existing large-scale open data source through heuristic rules.

[0162] At present, the scale of the open data source is large, and the three types of common sense content in Table 4 are included, which can be extracted by writing and using heuristic rule sets. For example, the numerical common sense content is extracted by using regular expressions.

[0163] The second type is to acquire the three types of common sense content through the template + retrieval method.

[0164] The specific way of this method is: first, construct a plurality of templates, then send the "concept + template" into the search engine, and finally process the obtained required results.

[0165] Figure 6 is a schematic diagram of acquiring the three types of common sense content through the template retrieval method provided by the application, as shown in Figure 6 , which shows how to acquire the co-located common sense content and the state common sense content through the template under the "Baidu" search engine.

[0166] The third type is to use Chinese word segmentation, named entity recognition and noun phrase recognition methods to process large-scale mathematical application examples to obtain a candidate concept set.

[0167] Each application example can be taken as a starting point to extract and expand the three types of common sense content involved therein.

[0168] Through the above three methods, the application constructs a common sense knowledge sub-library. In order to ensure the correctness of the content of the common sense knowledge sub-library, all the contents can be manually checked when necessary.

[0169] At this point, the application completes the construction of the application problem solving knowledge base, which can be used to support the development of the application problem solving system.

[0170] In specific use, the corresponding query interface can be developed according to the actual situation for different sub-libraries.

[0171] Figure 7 is a structural schematic diagram of the construction device of the application problem solving knowledge base provided by the application, as shown in Figure 7 , mainly including a data acquisition and processing module 71 and a query interface configuration module 72, wherein:

[0172] The data acquisition and processing module 71 is mainly used for acquiring data to construct the mathematical knowledge sub-library and the common sense knowledge sub-library respectively.

[0173] The query interface configuration module 72 is mainly used for combining the mathematical knowledge sub-library and the common sense knowledge sub-library to form an application question solving knowledge library, and configuring corresponding query interfaces for the mathematical knowledge sub-library and the common sense knowledge sub-library respectively.

[0174] It should be noted that the application question solving knowledge library construction device provided by the embodiment of the application can execute the application question solving knowledge library construction method of any one of the above-mentioned embodiments in the specific operation, and the embodiment will not be repeated here.

[0175] The application question solving knowledge library construction device provided by the application can fully consider that the knowledge content related to the application question solving not only contains mathematical knowledge content, but also includes common sense knowledge content, and the two parts of knowledge content have great differences, so the application question solving knowledge library is designed as two sub-libraries, different methods are used to acquire and construct the contents of the two sub-libraries according to the different characteristics of the two sub-libraries, and the acquired application question solving knowledge library can effectively and accurately assist the development of the application question automatic solving system, so as to greatly improve the solving accuracy of the application question automatic solving system.

[0176] Further, the application also provides an application question solving robot, which mainly comprises an application question solving robot body, a reading unit and an output unit are arranged in the application question solving robot body.

[0177] Further, the application also provides an application question solving robot, which mainly comprises an application question solving robot body, a reading unit and an output unit are arranged in the application question solving robot body.

[0178] The reading unit reads the application question to be solved, so that the program or instruction is executed by the processor to access the application question solving knowledge library, match the query to the answer result, and output the answer result through the output unit.

[0179] The application question solving knowledge library is created in advance based on the steps of the application question solving knowledge library construction method provided by any one of the above-mentioned embodiments.

[0180] Figure 8 It is a structural schematic diagram of the electronic equipment provided by the application, as shown in Figure 8As shown, the electronic device can include a processor 810, a communications interface 820, a memory 830, and a communications bus 840, wherein the processor 810, the communications interface 820, and the memory 830 complete mutual communication through the communications bus 840. The processor 810 can invoke a logical instruction in the memory 830 to execute a construction method of an application problem solving knowledge base, the method including: constructing a mathematical knowledge sub-base and constructing a common sense knowledge sub-base; combining the mathematical knowledge sub-base and the common sense knowledge sub-base to form an application problem solving knowledge base, and configuring a corresponding query interface for the mathematical knowledge sub-base and the common sense knowledge sub-base, respectively.

[0181] In addition, the logical instruction in the memory 830 described above can be implemented in the form of a software functional unit and sold or used as an independent product, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0182] On the other hand, the present application also provides a computer program product, the computer program product includes a computer program stored on a non-transitory computer readable storage medium, the computer program includes program instructions, when the program instructions are executed by a computer, the computer can execute the construction method of the application problem solving knowledge base provided by the above-mentioned method, the method includes: constructing a mathematical knowledge sub-base and constructing a common sense knowledge sub-base; combining the mathematical knowledge sub-base and the common sense knowledge sub-base to form an application problem solving knowledge base, and configuring a corresponding query interface for the mathematical knowledge sub-base and the common sense knowledge sub-base, respectively.

[0183] In yet another aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program, which, when executed by a processor, implements the method for constructing an application problem solving knowledge base provided by each of the above embodiments, and the method comprises: constructing a mathematical knowledge base and constructing a common sense knowledge base; combining the mathematical knowledge base and the common sense knowledge base to form an application problem solving knowledge base, and configuring a corresponding query interface for the mathematical knowledge base and the common sense knowledge base, respectively.

[0184] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment scheme according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0185] From the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software and the necessary general hardware platform, and of course, can also be realized by hardware. Based on such understanding, the above technical solutions, essentially or in the form of software products, can be embodied in a computer software product, which can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0186] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for constructing an application problem solving knowledge base, characterized by, The application comprises the following steps: constructing a mathematical knowledge sub-library and a common sense knowledge sub-library; combining the mathematical knowledge sub-library and the common sense knowledge sub-library to form an application problem solving knowledge library, and configuring corresponding query interfaces for the mathematical knowledge sub-library and the common sense knowledge sub-library respectively; the construction of the mathematical knowledge sub-library specifically comprises the following steps: mining authoritative materials to obtain a mathematical knowledge entity set related to three types of mathematical knowledge entities, and building a skeleton of the mathematical knowledge sub-library; the three types of mathematical knowledge entities comprise mathematical knowledge topics, mathematical knowledge points and mathematical knowledge concepts; the authoritative materials comprise documents, books and test papers related to teaching materials and examination syllabuses; the subordination relationship of the three types of mathematical knowledge entities comprises any one of the following: a knowledge concept belongs to a knowledge topic, a knowledge concept belongs to a knowledge point, and a knowledge point belongs to a knowledge topic; mining authoritative materials and various teaching aid materials to obtain a mathematical knowledge content set related to the three types of mathematical knowledge entities, so as to fill the skeleton of the mathematical knowledge sub-library with the mathematical knowledge content set and create the mathematical knowledge sub-library.

2. The construction method of an application problem solving knowledge base according to claim 1, characterized by, the mining of the authoritative materials to obtain the mathematical knowledge entity set related to the three types of mathematical knowledge entities and the building of the skeleton of the mathematical knowledge sub-library specifically comprises the following steps: based on an OCR technology, converting paper versions of the authoritative materials into electronic documents; removing random code parts and meaningless data parts in the electronic documents to realize data cleaning of the electronic documents; based on a human-machine hybrid mode, extracting the mathematical knowledge entity set from the electronic documents after data cleaning to build the skeleton of the mathematical knowledge sub-library.

3. The method of claim 1 or 2, wherein the application problem solving knowledge base is constructed by: The application further comprises the following steps: obtaining a new version of the authoritative materials, expanding the mathematical knowledge sub-library according to new mathematical knowledge contents extracted from the new version of the authoritative materials; and / or, obtaining untreated teaching aid materials, expanding the mathematical knowledge sub-library according to new mathematical knowledge contents extracted from the untreated teaching aid materials; and / or, obtaining missing mathematical knowledge contents in the mathematical knowledge sub-library to expand the missing mathematical knowledge contents to the mathematical knowledge sub-library.

4. The construction method of an application problem solving knowledge base according to claim 1, characterized by, the construction of the common sense knowledge sub-library specifically comprises the following steps: constructing a concept-type common sense content set and other common sense content sets to form the common sense knowledge sub-library by integrating the concept-type common sense content set and the other common sense content sets; the other common sense content sets comprise a numerical-type common sense content set, a state-type common sense content set and a same-position-type common sense content set.

5. The method of claim 4, wherein the application problem solving knowledge base is constructed by, the construction of the concept-type common sense content set specifically comprises the following steps: based on the architecture of an open knowledge source and in combination with a candidate concept set extracted from a large number of application problem examples, building a concept hierarchical framework; obtaining candidate target attributes related to each candidate concept in the concept hierarchical framework to build a candidate target attribute set; according to the candidate target attribute set, respectively assigning target attribute values to target attributes of each candidate concept in the concept hierarchical framework to obtain a plurality of triplets composed of candidate concepts, target attributes and target attribute values; The triad set containing all the triads constitutes the concept type common sense content set.

6. The method of claim 5, wherein the application problem solving knowledge base is constructed by, The disclosed knowledge source includes at least one of Schema.org, Wikipedia, HowNet, and network encyclopedia.

7. The method of claim 4, wherein the application problem solving knowledge base is constructed by, The method for constructing other common sense content sets specifically includes: Extracting numerical type common sense content, state type common sense content and paratactic type common sense content from open data sources through heuristic rules to respectively constitute the numerical type common sense content set, the state type common sense content set and the paratactic type common sense content set; And / or, obtaining numerical type common sense content, state type common sense content and paratactic type common sense content from search engines through template retrieval to respectively constitute the numerical type common sense content set, the state type common sense content set and the paratactic type common sense content set; And / or, based on Chinese word segmentation, named entity recognition and noun phrase recognition, keyword extraction is performed on a large number of mathematical application question examples to obtain numerical type common sense content, state type common sense content and paratactic type common sense content, respectively constituting the numerical type common sense content set, the state type common sense content set and the paratactic type common sense content set.

8. An apparatus for constructing an application problem solving knowledge base, characterized by comprising: It includes: A data acquisition and processing module for acquiring data to construct a mathematical knowledge sub-library and a common sense knowledge sub-library; The construction of the mathematical knowledge sub-library specifically includes: mining authoritative materials to obtain a mathematical knowledge entity set related to three types of mathematical knowledge entities to build the skeleton of the mathematical knowledge sub-library; the three types of mathematical knowledge entities include mathematical knowledge topics, mathematical knowledge points and mathematical knowledge concepts; the authoritative materials include documents, books and test papers related to textbooks and examination syllabuses; the subordination relationship of the three types of mathematical knowledge entities includes any of the following: knowledge concepts subordinate to knowledge topics, knowledge concepts subordinate to knowledge points, and knowledge points subordinate to knowledge topics; mining authoritative materials and various teaching aids to obtain a mathematical knowledge content set related to the three types of mathematical knowledge entities to fill the skeleton of the mathematical knowledge sub-library with the mathematical knowledge content set and create the mathematical knowledge sub-library; A query interface configuration module for combining the mathematical knowledge sub-library and the common sense knowledge sub-library to constitute an application question solving knowledge base and configuring corresponding query interfaces for the mathematical knowledge sub-library and the common sense knowledge sub-library.

9. A problem solving robot for application problems, characterized by, It includes an application question solving robot ontology, wherein a reading unit and an output unit are arranged in the application question solving robot ontology; It also includes a memory and a program or instructions stored on the memory and executable on the processor; The reading unit reads the application question to be solved, so that the program or instructions are executed by the processor to access the application question solving knowledge base, match the query to the answer result, and output the answer result via the output unit; The application question solving knowledge base is created in advance based on the steps of the method for constructing the application question solving knowledge base according to any one of claims 1 to 7.

Images