Industry classification label construction method and device, medium, and electronic equipment

Abstract

The present application provides a method for constructing industry classification labels and related products, which belong to the field of artificial intelligence technology. It obtains scientific and technological innovation subject data, intellectual property data, and existing industry classification data, obtains the relationship data between the scientific and technological innovation subject data and the intellectual property data through data processing, extracts and integrates keyword data from the intellectual property data, extracts and synchronously stores the relationship data between keyword data and intellectual property data; constructs a knowledge graph, and classifies entities and extracts subject features based on the knowledge graph to obtain the final industry classification label; associates the final industry classification label with the preset industry classification label to form an industry classification label system. The technical solution of the present application establishes a knowledge graph, uses the scientific and technological innovation subject data, intellectual property data, and the relationship data to extract the final industry classification label, and combines the existing industry classification data to establish an industry classification label system.

Description

Technical Field

[0001] The present application belongs to the field of artificial intelligence technology, and in particular, relates to a method and device for constructing industry classification labels, a medium, and an electronic device. Background Art

[0002] At present, the rapid development of scientific and technological innovation and the segmentation of technology fields and industrial chains have made it impossible for traditional industry classifications to represent and divide various innovation entities (innovation entities are related companies that own intellectual property rights). It is necessary to manually summarize and analyze industry rankings and positioning information under multiple industry classification systems to classify various innovation entities, which will be used for the subsequent construction of indicator models such as comprehensive innovation capability and market influence for each innovation entity.

[0003] In addition, industry classification labels from different data sources cannot cover the subject characteristics of innovation entities (the subject characteristics involve keywords of main technologies and business contents). Summary of the invention

[0004] One of the purposes of the present application is to provide a method and device, medium, and electronic device for constructing industry classification labels, which can classify industries for innovative entities and construct industry classification labels for innovative entities.

[0005] To achieve the above-mentioned purpose and other related purposes, the first aspect of the present application provides a method for constructing industry classification labels, comprising the following steps:

[0006] Data acquisition step; the data acquisition step includes acquiring scientific and technological innovation subject data, intellectual property data, existing industry classification data, and industry labels preset according to the existing industry classification data; the scientific and technological innovation subject data includes the name of the enterprise subject;

[0007] Data processing step; the data processing step includes generating relationship data between the scientific and technological innovation subject data and the intellectual property data, extracting keyword data from the intellectual property data based on word frequency statistics, and extracting relationship data between the keyword data and the intellectual property data;

[0008] The step of constructing a knowledge graph includes dividing entity levels according to entity types, and defining relationship data between entities of different levels; the entity levels include primary entities, secondary entities, and tertiary entities; the entity whose entity type is the name of the enterprise entity is divided into the primary entity, the entity whose entity type is the intellectual property data is divided into the secondary entity, and the entity whose entity type is the keyword data is divided into the tertiary entity; the relationship data between entities of different levels includes the relationship data between the primary entity and the secondary entity, and the relationship data between the secondary entity and the tertiary entity;

[0009] A classification step based on a knowledge graph; the classification step based on the knowledge graph includes calculating a similarity RA; classifying the first-level entities according to the similarity RA into different first-level similar entity sets, wherein the first-level similar entity sets include multiple similar first-level entities, which are recorded as first-level similar entities; obtaining multiple second-level entities related to the first-level similar entity, which are recorded as second-level related entities, and obtaining multiple third-level entities related to the second-level related entity, which are recorded as third-level related entities, based on the relationship data between the first-level entity and the second-level entity and the relationship data between the second-level entity and the third-level entity; calculating the weight of the second-level related entity relative to the first-level similar entity and calculating the weight of the third-level related entity relative to the second-level related entity; calculating a co-occurrence index C according to the weight of the second-level related entity relative to the first-level similar entity and the weight of the third-level related entity relative to the second-level related entity; selecting multiple third-level related entities according to the co-occurrence index C as the final industry classification label of the first-level similar entity;

[0010] Industry classification label association step: The industry classification label association step includes associating the final industry classification label with the preset industry classification label to form an industry classification label system.

[0011] The second aspect of the present application provides an industry classification label construction device, including:

[0012] Acquisition data module; the acquisition data module is used to acquire scientific and technological innovation subject data, intellectual property data, existing industry classification data, and industry labels preset according to the existing industry classification data; the scientific and technological innovation subject data includes the name of the enterprise subject;

[0013] Data processing module: The data processing module is used to generate relationship data between the scientific and technological innovation subject data and the intellectual property data, extract keyword data from the intellectual property data based on word frequency statistics, and extract relationship data between the keyword data and the intellectual property data;

[0014] Constructing a knowledge graph module; constructing a knowledge graph module for dividing entity levels according to entity types, and defining relationship data between entities of different levels; the entity levels include first-level entities, second-level entities, and third-level entities; dividing entities whose entity type is the name of the enterprise entity into the first-level entity, dividing entities whose entity type is the intellectual property data into the second-level entity, and dividing entities whose entity type is the keyword data into the third-level entity; the relationship data between entities of different levels include the relationship data between the first-level entity and the second-level entity, and the relationship data between the second-level entity and the third-level entity;

[0015] A classification module based on a knowledge graph; the classification module based on the knowledge graph is used to calculate the similarity RA; the first-level entities are classified according to the similarity RA into different first-level similar entity sets, and the first-level similar entity sets include multiple similar first-level entities, which are recorded as first-level similar entities; based on the relationship data between the first-level entity and the second-level entity and the relationship data between the second-level entity and the third-level entity, multiple second-level entities related to the first-level similar entity are obtained, which are recorded as second-level related entities, and multiple third-level entities related to the second-level related entity are obtained, which are recorded as third-level related entities; the weight of the second-level related entity relative to the first-level similar entity is calculated, and the weight of the third-level related entity relative to the second-level related entity is calculated; the co-occurrence index C is calculated according to the weight of the second-level related entity relative to the first-level similar entity and the weight of the third-level related entity relative to the second-level related entity; according to the co-occurrence index C, multiple third-level related entities are selected as the final industry classification labels of the first-level similar entity;

[0016] Industry classification label association module: The industry classification label association module is used to associate the final industry classification label with the preset industry classification label to form an industry classification label system.

[0017] A third aspect of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, the processor executes the above-mentioned industry classification label construction method.

[0018] A fourth aspect of the present application provides an electronic device, comprising: at least one memory for storing a program;

[0019] At least one processor is used to execute the program stored in the memory. When the program stored in the memory is executed, the processor is used to execute the above-mentioned industry classification label construction method.

[0020] This application has at least the following beneficial effects:

[0021] The present application provides a method for constructing industry classification labels and related products, which obtains scientific and technological innovation subject data, intellectual property data, and existing industry classification data, obtains the relationship data between the scientific and technological innovation subject data and the intellectual property data through data processing, extracts and integrates keyword data from the intellectual property data, extracts and synchronously stores the relationship data between keyword data and intellectual property data; constructs a knowledge graph, and performs entity classification and subject feature extraction based on the knowledge graph to obtain the final industry classification label; associates the final industry classification label with the preset industry classification label to form an industry classification label system. The technical solution of the present application can efficiently establish an industry classification label system by establishing a knowledge graph, extracting the relationship data between the scientific and technological innovation subject data, intellectual property data, and the relationship data as the final industry classification label, and combining the existing industry classification data. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 The figure is a flow chart of a method for constructing industry classification labels according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] The following describes the implementation methods of the present application through specific specific embodiments, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific implementation methods, and the details in the present application can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0025] The term "and / or" in this application is a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. In the present invention, the symbol " / " indicates that the associated objects are in an "or" relationship, for example, "A / B" means "A or B".

[0026] The terms "first" and "second" in the specification and claims of this application are used to distinguish different objects rather than to describe a specific order of the objects. For example, "first response message" and "second response message" are used to distinguish different response messages rather than to describe a specific order of the response messages.

[0027] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be considered as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0028] In the description of the embodiments of the present application, unless otherwise specified, "multiple" means two or more than two. For example, multiple processing units refer to two or more processing units, etc., and multiple elements refer to two or more elements, etc.

[0029] The first aspect of the present application provides a method for constructing an industry classification label, comprising the following steps:

[0030] Get data step;

[0031] The data acquisition step includes acquiring scientific and technological innovation subject data and intellectual property data.

[0032] The data on science and technology innovation entities include the names of corporate entities; these corporate entities are corporate entities with intellectual property data, and these corporate entities include approximately 1.6 million enterprises and institutions.

[0033] Intellectual property data includes patent data and copyright data by type. Copyright data includes work copyright data and software copyright data. Intellectual property data includes bibliographic data (including abstract data) and full-text data by content.

[0034] The step of obtaining data also includes obtaining existing industry classification data and industry labels preset based on the existing industry classification data. Existing industry classification data includes data of national economic industry classification and / or data of Shenwan Hongyuan industry classification. On the basis of existing industry classification data, preset industry labels are obtained through deletion, integration and addition. Specifically, some label names will be directly quoted, but their label positions, upper and lower label affiliations, etc. have basically been adjusted, so the preset industry labels do not completely include data of national economic industry classification and Shenwan Hongyuan industry classification. For example: the first-level category label "electronic information technology" and its second-level category label are set with reference to the current national key support for high-tech fields, emerging industries and other trends, and after integrating national economic industry classification data such as computers and information technology.

[0035] Industry labels are used to classify and identify the industry field in which the scientific and technological innovation subject of this application is located, so as to reflect the main technology and business content of the scientific and technological innovation subject. Industry labels include first-level category labels, second-level category labels, and third-level category labels. For example,

[0036]

[0037] The step of acquiring data also includes classifying and storing the acquired data according to the data type; among them, the subject data of scientific and technological innovation that belongs to structured data (data logically expressed and realized by two-dimensional table structure) can be stored in TIDB database; the full-text data that belongs to unstructured data (data with irregular or incomplete data structure, no predefined data model, and data that is not convenient to be expressed by two-dimensional logical table of database) can be stored in MINIO distributed object storage server; the entry data that belongs to semi-structured data (a form between structured data and unstructured data) is stored in the form of files in MONGODB database. For the existing industry classification data, it is stored in the local Excel file.

[0038] Data processing steps;

[0039] The data processing steps include producing relationship data between scientific and technological innovation subject data and intellectual property data, extracting keyword data from intellectual property data based on word frequency statistics, and extracting relationship data between keyword data and intellectual property data.

[0040] The steps of generating relationship data between the scientific and technological innovation subject data and the intellectual property data include parsing, splitting, merging and filtering the scientific and technological innovation subject data and the intellectual property data.

[0041] The relationship data between keyword data and intellectual property data can be extracted through the configuration of the database primary key. The database primary key is used to identify the original source of the keyword data.

[0042] The relationship data between the subject data of scientific and technological innovation and the intellectual property data, the keyword data, and the relationship data between the keyword data and the intellectual property data are stored in the structured data table of the HIVE database. The relationship data between the subject data of scientific and technological innovation and the intellectual property data can be extracted and stored synchronously. The relationship data between the subject data of scientific and technological innovation and the intellectual property data includes: numbering the enterprise subject name to obtain the enterprise subject number, the intellectual property data includes patent data and work data, numbering the patents and works in the intellectual property data to obtain the intellectual property entity number, extracting the patent name or work name in the entry data of the intellectual property data as the intellectual property entity name, extracting the type of patent or work in the entry data as the intellectual property type, extracting the patent owner or copyright owner from the entry data (because the entry data is semi-structured data, it is necessary to split the semi-structured data to obtain it), matching the patent owner or copyright owner with the enterprise subject name, and forming a data storage structure including the enterprise subject number, enterprise subject name, intellectual property entity number, intellectual property entity name and intellectual property type. For example, the data storage structure is in the following form:

[0043]

[0044] The keyword data can be extracted from the intellectual property data based on word frequency statistics. The keyword data can include keyword strings. In the above table, the keyword data corresponds to the database primary key, that is, the keyword data is configured with a corresponding database primary key.

[0045] Extracting keyword data includes data preprocessing, which includes: noise cleaning, word segmentation, and stop word removal.

[0046] Noise cleaning includes filtering out name data in the bibliographic data and information that does not contain Chinese characters and repeated garbled data in the summary data.

[0047] The word segmentation process includes the word segmentation process on the name data and the abstract data in the entry data.

[0048] Stop word removal includes removing words that do not have representative meanings, such as "a kind of", "of", "based on", etc., from the name data and summary data after word segmentation.

[0049] The above-mentioned example of processing the name data (including patent names and book names) and abstract data (including patent abstracts and book abstracts) of the bibliographic data based on word frequency statistics is as follows:

[0050]

[0051] The above data preprocessing based on word frequency statistics is also applicable to summary data and full-text data. For summary data and full-text data, the traditional TF-IDF algorithm can be used to extract keywords based on word frequency statistics.

[0052] Steps to build a knowledge graph;

[0053] The steps of building a knowledge graph include dividing entity levels according to entity types and defining relationship data between entities at different levels.

[0054] The step of dividing entity levels according to entity types includes defining different levels of entities, including first-level entities, second-level entities and third-level entities; dividing entities whose entity types are corporate entity names into first-level entities, dividing entities whose entity types are intellectual property data into second-level entities, and dividing entities whose entity types are keyword data into third-level entities.

[0055] The storage process corresponding to the step of dividing entity levels according to entity types includes: extracting enterprise entity name data from the TIDB database storing scientific and technological innovation subject data; extracting full-text data from the MINIO distributed object storage server; extracting bibliographic data from the MONGODB database; extracting keyword data from the HIVE database; storing the extracted enterprise entity name data, full-text data, bibliographic data, and keyword data in the Neo4j database. The specific implementation of the above steps includes: creating first-level entities, second-level entities, and third-level entities through Python's Node function; using the graph.create function of Python's py2neo library to store first-level entities, second-level entities, and third-level entities in the Neo4j database.

[0056] In some embodiments of the present application, the step of storing the extracted keyword data in the Neo4j database further includes comparing the keyword data, and merging keyword data with similar meanings into one keyword data. The specific implementation method includes: comparing the keyword data through Python's hash function and eq function.

[0057] The step of defining the relationship data between entities of different levels includes defining the relationship data between the first-level entity and the second-level entity and defining the relationship data between the second-level entity and the third-level entity. Specifically, the content of the definition includes defining the categories and boundaries of the relationship data between entities of different levels.

[0058] Defining the relationship data between the primary entity and the secondary entity includes: extracting the enterprise subject name and the intellectual property entity name subordinate to the enterprise subject name from the data storage structure, and defining the enterprise subject name and the intellectual property entity name as a subordinate relationship. It is required that the entity of the enterprise subject name corresponds to at least one entity of the intellectual property entity name, and the corresponding relationship belongs to the relationship data between the primary entity and the secondary entity.

[0059] Defining the relationship data between the second-level entity and the third-level entity includes: corresponding each intellectual property data entity to at least one keyword data entity, and the corresponding relationship belongs to the relationship data between the second-level entity and the third-level entity. The relationship data between the second-level entity and the third-level entity can be a combination of intellectual property data and keyword data, such as "patent keywords" and "book keywords".

[0060] The specific implementation of the above steps includes: using the Relationship function in Python to define the relationship data between the first-level entity and the second-level entity, and the relationship data between the second-level entity and the third-level entity. Using the graph.create function in the py2neo library in Python to store the defined relationship data between the first-level entity and the second-level entity, and the relationship data between the second-level entity and the third-level entity in the Neo4j graph database.

[0061] In summary, the Neo4j graph database is used to store entities of different levels in the knowledge graph, as well as the relationship data between the first-level entities and the second-level entities, and the relationship data between the second-level entities and the third-level entities.

[0062] Classification steps based on knowledge graph;

[0063] The classification step based on the knowledge graph includes calculating the similarity RA; the similarity RA calculation formula is as follows:

[0064]

[0065] Among them, x and y represent the first-level entity (i.e., the name of the enterprise entity), u represents the second-level entity (i.e., intellectual property data) shared by the first-level entities x and y, the RA (x, y) value represents the similarity between the first-level entities x and y, N (x) represents the set of second-level entities adjacent to the first-level entity x, N (y) represents the set of second-level entities adjacent to the first-level entity y, and N (u) represents the set of second-level entities adjacent to the second-level entity (i.e., intellectual property data) shared by the first-level entities x and y. The larger the RA (x, y) value, the higher the similarity between the first-level entities x and y. The higher the similarity, the more second-level entities x and y have in common.

[0066] The classification step based on the knowledge graph includes classifying the first-level entities according to the similarity RA into different first-level similar entity sets, wherein the first-level similar entity sets include multiple similar first-level entities, which are recorded as first-level similar entities. For example, when the similarity RA (x, y) value is greater than 1.5, these similar first-level entities are recorded as first-level similar entities, and these first-level similar entities can be composed into different first-level similar entity sets according to the set similarity RA threshold standard. If different first-level similar entity sets have the same first-level similar entity, the first-level similar entity is divided into the first-level similar entity set that can obtain the highest similarity RA value.

[0067] The classification step based on the knowledge graph includes obtaining multiple second-level entities related to the first-level similar entity based on the relationship data between the first-level entity and the second-level entity, which are recorded as second-level related entities; and obtaining multiple third-level entities related to the second-level related entity based on the relationship data between the second-level entity and the third-level entity, which are recorded as third-level related entities.

[0068] In some embodiments of the present application, the similarities between a plurality of first-level entities corresponding to one of the second-level related entities are calculated to determine whether these first-level entities are first-level similar entities.

[0069] The knowledge graph-based classification step includes calculating the weight of the second-level related entities relative to the first-level similar entities and calculating the weight of the third-level related entities relative to the second-level related entities.

[0070] For example, if a first-level similar entity M1 corresponds to multiple second-level related entities N (including second-level related entity N1), and second-level related entity N1 corresponds to X1 first-level similar entities M (including first-level similar entity M1), then the weight of second-level related entity N1 relative to first-level similar entity M1 is for: .

[0071] It can be seen that the more the number of second-level related entities N1 corresponds to the first-level similar entities M, the greater the weight of the second-level related entity N1 relative to the first-level similar entity M1. The smaller.

[0072] Similarly, the second-level related entity N1 corresponds to multiple third-level related entities S (including the third-level related entity S1), and the third-level related entity S1 corresponds to X2 second-level related entities N (including the second-level related entity N1). Then the weight of the third-level related entity S1 relative to the second-level related entity N1 is for: .

[0073] The knowledge graph-based classification step includes calculating the co-occurrence index C based on the weight of the second-level related entities relative to the first-level similar entities and the weight of the third-level related entities relative to the second-level related entities.

[0074] The co-occurrence index C is used to measure the degree to which two or more things co-occur in a data set. In this application, the co-occurrence index C represents the degree to which the second-level related entities and the third-level related entities co-occur with respect to the same first-level similar entity M.

[0075] For example, a first-level similar entity M1 corresponds to multiple second-level related entities N (including second-level related entity N1) and multiple third-level related entities S (including third-level related entity S1) corresponding to the second-level related entity N. According to the above calculation formula, the weight of the second-level related entity N1 to the first-level similar entity M1 is for: .

[0076] The weight of the third-level related entity S1 relative to the second-level related entity N1 for: .

[0077] Then the co-occurrence index of the second-level related entity N1 and the third-level related entity S1 relative to the same first-level similar entity M1 is The calculation formula is as follows:

[0078] in, is the weight of all second-level related entities N corresponding to the third-level related entity S1 (including the weight of the second-level related entity N1 to the first-level similar entity M1 ), is the weight of all third-level related entities S corresponding to the second-level related entity N1 (including the weight of the third-level related entity S1 relative to the second-level related entity N1 ) is the mean of .

[0079] According to the value of the co-occurrence index C, multiple first-level similar entities with the top values ​​of the co-occurrence index C (for example, the top three in each set) are selected from all the third-level related entities corresponding to the same first-level similar entity as the final industry classification label of the first-level similar entity. This process is also called topic feature extraction.

[0080] Industry classification label association steps;

[0081] The industry classification label association step includes associating the final industry classification label with the preset industry classification label according to the industry classification label preset in the data acquisition step to form an industry classification label system. For example, if the preset industry labels are a first-level category label, a second-level category label, and a third-level category label. Then the final industry classification label is a fourth-level category label. If there is no fourth-level category label that can be associated under one of the third-level category labels, then the third-level category label is merged with the third-level category labels of other associated fourth-level category labels.

[0082] A second aspect of the present application provides an industry classification label construction device, the industry classification label construction device comprising:

[0083] Acquisition data module; the acquisition data module is used to acquire scientific and technological innovation subject data, intellectual property data, existing industry classification data, and industry labels preset according to the existing industry classification data; the scientific and technological innovation subject data includes the name of the enterprise subject;

[0084] Data processing module: The data processing module is used to generate relationship data between the scientific and technological innovation subject data and the intellectual property data, extract keyword data from the intellectual property data based on word frequency statistics, and extract relationship data between the keyword data and the intellectual property data;

[0085] Constructing a knowledge graph module; constructing a knowledge graph module for dividing entity levels according to entity types, and defining relationship data between entities of different levels; the entity levels include first-level entities, second-level entities, and third-level entities; dividing entities whose entity type is the name of the enterprise entity into the first-level entity, dividing entities whose entity type is the intellectual property data into the second-level entity, and dividing entities whose entity type is the keyword data into the third-level entity; the relationship data between entities of different levels include the relationship data between the first-level entity and the second-level entity, and the relationship data between the second-level entity and the third-level entity;

[0086] A classification module based on a knowledge graph; the classification module based on the knowledge graph is used to calculate the similarity RA; the first-level entities are classified according to the similarity RA into different first-level similar entity sets, and the first-level similar entity sets include multiple similar first-level entities, which are recorded as first-level similar entities; based on the relationship data between the first-level entity and the second-level entity and the relationship data between the second-level entity and the third-level entity, multiple second-level entities related to the first-level similar entity are obtained, which are recorded as second-level related entities, and multiple third-level entities related to the second-level related entity are obtained, which are recorded as third-level related entities; the weight of the second-level related entity relative to the first-level similar entity is calculated, and the weight of the third-level related entity relative to the second-level related entity is calculated; the co-occurrence index C is calculated according to the weight of the second-level related entity relative to the first-level similar entity and the weight of the third-level related entity relative to the second-level related entity; according to the co-occurrence index C, multiple third-level related entities are selected as the final industry classification labels of the first-level similar entity;

[0087] The role of the knowledge graph-based classification module is to determine the similarity (or intimacy) between different entities and extract subject features for the construction of industry classification labels in the industry classification label association module.

[0088] Industry classification label association module: The industry classification label association module is used to associate the final industry classification label with the preset industry classification label to form an industry classification label system.

[0089] The third aspect of the present application provides a computer-readable storage medium, on which a program of an industry classification label construction method is stored, so that the industry classification label construction method provided in the first aspect of the present application is executed. As for the aforementioned computer-readable storage medium, those skilled in the art can understand that the implementation example of the above-mentioned system and each unit function can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, it executes an implementation example including the above-mentioned system and each unit function. The aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM or optical disk.

[0090] The fourth aspect of the present application provides an electronic device, comprising: at least one memory for storing programs; at least one processor for executing the programs stored in the memory. The electronic device may be a server. The electronic device comprises a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device comprises a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the electronic device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, the industry classification label construction function or step of the present application is implemented.

[0091] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and substitutions can be made without departing from the technical principles of the present application. These improvements and substitutions should also be regarded as the scope of protection of the present application.

Claims

1. A method for constructing industry classification labels, characterized in that: The following steps are involved: Data acquisition step; the data acquisition step includes acquiring scientific and technological innovation subject data, intellectual property data, existing industry classification data, and industry labels preset according to the existing industry classification data; the scientific and technological innovation subject data includes the name of the enterprise subject; Data processing step; the data processing step includes generating relationship data between the scientific and technological innovation subject data and the intellectual property data, extracting keyword data from the intellectual property data based on word frequency statistics, and extracting relationship data between the keyword data and the intellectual property data; The step of constructing a knowledge graph includes dividing entity levels according to entity types, and defining relationship data between entities of different levels; the entity levels include primary entities, secondary entities, and tertiary entities; the entity whose entity type is the name of the enterprise entity is divided into the primary entity, the entity whose entity type is the intellectual property data is divided into the secondary entity, and the entity whose entity type is the keyword data is divided into the tertiary entity; the relationship data between entities of different levels includes the relationship data between the primary entity and the secondary entity, and the relationship data between the secondary entity and the tertiary entity; The classification step based on the knowledge graph includes calculating the similarity. The similarity calculation formula is as follows: ; where x and y represent first-level entities respectively, and u represents the second-level entity shared by the first-level entities x and y. The value represents the similarity between the first-level entities x and y. represents the set of secondary entities adjacent to the primary entity x, represents the set of secondary entities adjacent to the primary entity y, A set of secondary entities adjacent to the secondary entities shared by the primary entities x and y; classifying the primary entities according to the similarity RA into different sets of primary similar entities, wherein the sets of primary similar entities include multiple similar primary entities, which are recorded as primary similar entities; obtaining multiple secondary entities related to the primary similar entity, which are recorded as secondary related entities, and obtaining multiple tertiary entities related to the secondary related entity, which are recorded as tertiary related entities, according to the relationship data between the primary entity and the secondary entity and the relationship data between the secondary entity and the tertiary entity; calculating the weight of the secondary related entity relative to the primary similar entity and the weight of the tertiary related entity relative to the secondary related entity; calculating the co-occurrence index according to the weight of the secondary related entity relative to the primary similar entity and the weight of the tertiary related entity relative to the secondary related entity; selecting multiple tertiary related entities according to the co-occurrence index as the final industry classification label of the primary similar entity; an industry classification label association step; the industry classification label association step includes associating the final industry classification label with the preset industry classification label to form an industry classification label system.

2. The method for constructing industry classification labels according to claim 1, characterized in that: The existing industry classification data include the data of national economic industry classification and / or the data of Shenwan Hongyuan industry classification.

3. The method for constructing industry classification labels according to claim 1, characterized in that: The step of extracting keywords from the intellectual property data based on word frequency statistics includes data preprocessing.

4. The method for constructing industry classification labels according to claim 1, characterized in that: The relationship data between the production of the scientific and technological innovation subject data and the intellectual property data includes: numbering the enterprise subject name to obtain the enterprise subject number; the intellectual property data includes patent data and work data, numbering the patents and works in the intellectual property data to obtain the intellectual property entity number, extracting the patent name or work name in the entry data of the intellectual property data as the intellectual property entity name, extracting the type of the patent or work in the entry data as the intellectual property type, extracting the patent right owner or copyright owner from the entry data, matching the patent right owner or copyright owner with the enterprise subject name, and forming a data storage structure including the enterprise subject number, enterprise subject name, intellectual property entity number, intellectual property entity name and intellectual property type.

5. The method for constructing industry classification labels according to claim 4, characterized in that: The extracting of the relationship data between the keyword data and the intellectual property data also includes configuring a database primary key, and the data storage structure also includes the keyword data and the database primary key.

6. The method for constructing industry classification labels according to claim 4, characterized in that: Defining the relationship data between the primary entity and the secondary entity includes: extracting the enterprise entity name and the intellectual property entity name subordinate to the enterprise entity name from the data storage structure, and defining the enterprise entity name and the intellectual property entity name as a subordinate relationship.

7. The method for constructing industry classification labels according to claim 5, characterized in that: Defining the relationship data between the second-level entity and the third-level entity includes: making each entity of the intellectual property data correspond to at least one entity of the keyword data.

8. An industry classification label construction device, characterized in that: include: Get data module; The data acquisition module is used to acquire the scientific and technological innovation subject data, intellectual property data, existing industry classification data, and industry labels preset according to the existing industry classification data; the scientific and technological innovation subject data includes the name of the enterprise subject; data processing module; The data processing module is used to produce the relationship data between the scientific and technological innovation subject data and the intellectual property data, extract keyword data from the intellectual property data based on word frequency statistics, and extract the relationship data between the keyword data and the intellectual property data; construct a knowledge graph module; construct a knowledge graph module to divide entity levels according to entity types, and define the relationship data between entities of different levels; the entity levels include first-level entities, second-level entities and third-level entities; the entity whose entity type is the name of the enterprise subject is divided into the first-level entity, the entity whose entity type is the intellectual property data is divided into the second-level entity, and the entity whose entity type is the keyword data is divided into the third-level entity; the relationship data between entities of different levels include the relationship data between the first-level entity and the second-level entity and the relationship data between the second-level entity and the third-level entity; Classification module based on knowledge graph; the classification module based on knowledge graph is used to calculate similarity; the similarity calculation formula is as follows: ; where x and y represent first-level entities respectively, and u represents the second-level entity shared by the first-level entities x and y. The value represents the similarity between the first-level entities x and y. represents the set of secondary entities adjacent to the primary entity x, represents the set of secondary entities adjacent to the primary entity y, Represents a set of secondary entities adjacent to the secondary entities shared by the primary entities x and y; classifies the primary entities according to the similarity RA into different sets of primary similar entities, wherein the sets of primary similar entities include multiple similar primary entities, which are recorded as primary similar entities; obtains multiple secondary entities related to the primary similar entity, which are recorded as secondary related entities, and obtains multiple third-level entities related to the second-level related entity, which are recorded as third-level related entities, according to the relationship data between the primary entity and the secondary entity and the relationship data between the secondary entity and the third-level entity; calculates the weight of the second-level related entity relative to the primary similar entity and the weight of the third-level related entity relative to the second-level related entity; calculates the co-occurrence index according to the weight of the second-level related entity relative to the primary similar entity and the weight of the third-level related entity relative to the second-level related entity; selects multiple third-level related entities as the final industry classification label according to the co-occurrence index; an industry classification label association module; the industry classification label association module is used to associate the final industry classification label with the preset industry classification label to form an industry classification label system.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program runs on a processor, the processor executes the industry classification label construction method as described in any one of claims 1 to 7.

10. An electronic device, characterized in that: include: at least one memory for storing a program; At least one processor is used to execute the program stored in the memory. When the program stored in the memory is executed, the processor is used to execute the industry classification label construction method as described in any one of claims 1 to 7.

Images