A knowledge management method, device and system for urban physical examination knowledge

Abstract

The application discloses a kind of city physical examination knowledge knowledge management method, device and system.From theme, index and method three aspects, according to the preset multi-department analysis and evaluation index group, the city physical examination knowledge instance resource and association clustering method, to establish city physical examination knowledge base in the form of knowledge graph, to establish the standardization framework based on different analysis and evaluation index, and then realize the universality and compatibility of knowledge management, the knowledge management method, device and system of the present application improve the knowledge management efficiency of city physical examination knowledge;Further, the present application provides a kind of city physical examination knowledge knowledge management method, device and system also by according to the city physical examination knowledge base, timely to the city physical examination ontology base is supplemented and improved in reverse, to improve the accuracy of city physical examination knowledge knowledge management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of knowledge management of urban physical examination knowledge, and particularly relates to a knowledge management method, device, computer readable storage medium and system of urban physical examination knowledge. BACKGROUND

[0002] A knowledge graph is a semantic network that reveals the relationship between entities and can formally describe real things and their associated relationships. Knowledge graphs have been popularized in both academia and industry and play an important role in intelligent search, intelligent question answering, intelligent recommendation and other applications. Urban information is widely sourced, diverse in type and inconsistent in format, posing great challenges to data collection, correlation, fusion and analysis. At the same time, there is little useful data for urban physical examination evaluation in the vast amount of data, and important knowledge is easily overwhelmed by junk information. Using a knowledge graph to process, analyze and correlate and display multi-source heterogeneous data and knowledge helps to regularly evaluate the characteristics of urban development and the implementation of planning, helps to timely reveal problems and shortcomings in land space governance and urban function layout, and improves the quality of urban development. Specifically, the urban physical examination knowledge graph involves knowledge in multiple fields such as humanities, economy and environment.

[0003] In the prior art, a knowledge graph is used to process, analyze and correlate and display multi-source heterogeneous data and knowledge, wherein graph construction includes multi-level entity extraction, multi-level relationship extraction and knowledge graph storage.

[0004] However, the prior art still has the following defects: due to inconsistent analysis and evaluation indicators for urban physical examination, different data processing methods, and different characteristics of the subject, there is a lack of a unified framework based on semantics, and the generality and compatibility of urban physical examination knowledge graph construction and knowledge service application are not strong, resulting in low knowledge management efficiency.

[0005] Therefore, there is currently a need for a knowledge management method, device, computer readable storage medium and system of urban physical examination knowledge to overcome the above-mentioned defects in the prior art. SUMMARY

[0006] The embodiments of the present application provide a knowledge management method, device, computer readable storage medium and system of urban physical examination knowledge to improve the knowledge management efficiency of urban physical examination knowledge.

[0007] An embodiment of the present application provides a knowledge management method of city physical examination knowledge, the knowledge management method comprises: collecting city physical examination data through a subject crawler data collection tool, and cleaning the city physical examination data to obtain city physical examination knowledge; classifying and summarizing the city physical examination knowledge according to a preset field category and a preset expert experience library to obtain a city physical examination ontology library; constructing a city physical examination knowledge instance resource according to a preset first external knowledge library and the city physical examination ontology library; and establishing a city physical examination knowledge base in the form of a knowledge graph from three aspects of a theme, an index and a method according to a preset multi-department analysis and evaluation index group, the city physical examination knowledge instance resource and an association clustering method.

[0008] As an improvement of the above scheme, the city physical examination knowledge base is established from three aspects of a theme, an index and a method according to a preset multi-department analysis and evaluation index group, the city physical examination knowledge instance resource and an association clustering method, and specifically comprises: obtaining the preset multi-department analysis and evaluation index group; the multi-department analysis and evaluation index group comprises an analysis and evaluation index and a corresponding analysis and evaluation index connotation; an implicit Dirichlet distribution model is used to calculate the analysis and evaluation index to establish an index library; the analysis and evaluation index connotation is subjected to feature mining to establish a method library; the analysis and evaluation index is refined according to an association clustering method to obtain a theme group, and the theme group is subjected to clustering analysis to obtain a theme library; and the city physical examination knowledge base is established according to the index library, the method library and the theme library.

[0009] As an improvement of the above scheme, the city physical examination knowledge is subjected to classification and summary processing according to a preset field category and a preset expert experience library to obtain a city physical examination ontology library, and specifically comprises: obtaining the preset field category and the preset expert experience library; determining a city physical examination ontology range corresponding to each field category from the city physical examination knowledge according to the expert experience library and the field category; extracting a first keyword group and a second keyword group from the city physical examination knowledge range by a preset first high-frequency calculation method and a preset second high-frequency calculation method, respectively, and fusing and deduplicating the first keyword group and the second keyword group to correspondingly obtain a field concept corresponding to each field category; iteratively extracting according to a preset second external knowledge library, a preset matching method and a preset scoring evaluation formula to obtain a field relationship between each field category; and constructing the city physical examination ontology library according to the field concept and the field relationship.

[0010] As an improvement of the above scheme, the scoring evaluation formula is specifically:

[0011]

[0012] Wherein, N is the total number of candidate relations mined by the candidate word table P, and F is the number of relation entities in the entity dictionary mined by the candidate word table P; The precision of the candidate word table P is represented, and log2(F+1) represents the recall ability of the candidate word table P.

[0013] As an improvement of the above scheme, the implicit Dirichlet distribution model is used to calculate the analysis and evaluation indexes to establish an index library, specifically including: establishing an analysis and evaluation index set according to the analysis and evaluation indexes; data preprocessing is performed on the analysis and evaluation index set to obtain a department analysis and evaluation index set; the department analysis and evaluation index set includes a plurality of department analysis and evaluation indexes; according to the implicit Dirichlet distribution model and the preset keyword corresponding target function, the index keywords corresponding to the department analysis and evaluation indexes are calculated and confirmed, and according to the entanglement degree, the department analysis and evaluation indexes and the index keywords, an index library is established.

[0014] As an improvement of the above scheme, the preset keyword corresponding target function is specifically:

[0015]

[0016] Wherein, α and β are iterative hyperparameters.

[0017] As an improvement of the above scheme, the analysis and evaluation indexes are refined according to the correlation clustering method to obtain a theme group, and the theme group is clustered to obtain a theme library, specifically including: under different preset theme connotations, theme semantic similarity, index semantic similarity, index unit similarity and calculation method similarity are calculated according to the analysis and evaluation indexes through JS divergence; according to the preset theme connotation correlation degree formula, theme semantic similarity, index semantic similarity, index unit similarity and calculation method similarity, the theme connotation correlation degree between each theme connotation is calculated; according to the correlation clustering algorithm, the theme connotations are clustered to obtain a theme group; the theme group includes a plurality of theme keywords; according to the preset similarity calculation formula, the similarity between the theme keywords and the theme connotations is calculated, and according to the preset classification threshold and the similarity, the theme keywords and the theme connotations are clustered to obtain a theme library.

[0018] As an improvement of the above scheme, the preset theme connotation correlation degree formula is specifically:

[0019] SC=a*theme semantic similarity+b*index semantic similarity+c*index unit similarity+d*

[0020] Calculation method similarity;

[0021] Wherein, SC is the theme connotation correlation degree;a, b, c, d are weight factors respectively, and a, b, c, d belong to (0, 1).

[0022] As the improvement of the above scheme, the preset similarity calculation formula is specifically:

[0023]

[0024] Wherein, S is the theme vocabulary;C is the theme connotation.

[0025] As the improvement of the above scheme, according to the preset second external knowledge base, the preset matching method and the preset scoring evaluation formula, iterative extraction is carried out to obtain the domain relationship between each domain category, specifically including: referring to the preset second external knowledge base, the relationship type between each entity is inducted and obtained, and the relationship type and the city health examination knowledge are stored in the entity dictionary;By the method of pulling boots, the city health examination knowledge is matched with the relationship type, and the matching result is stored in the candidate word table;According to the preset scoring evaluation formula, the matching result in the candidate word table is evaluated and scored, and according to the preset iteration number and the preset convergence accuracy, it is judged whether the evaluation and scoring result converges;If convergence, the relationship type corresponding to the evaluation and scoring result is output as the domain relationship between the corresponding domains.

[0026] As the improvement of the above scheme, the knowledge management method further includes: according to the city health examination knowledge base, the city health examination ontology base is supplemented and improved in reverse.

[0027] Another embodiment of the application provides a knowledge management device for city health examination knowledge, which comprises a data acquisition unit, a classification induction unit, an instance construction unit and a knowledge base establishment unit, wherein the data acquisition unit is used for acquiring city health examination data through a subject crawler data acquisition tool, and cleaning the city health examination data to obtain city health examination knowledge;The classification induction unit is used for classifying and inducing the city health examination knowledge according to the preset domain category and the preset expert experience base, and obtaining the city health examination ontology base;The instance construction unit is used for constructing city health examination knowledge instance resources according to the preset first external knowledge base and the city health examination ontology base;The knowledge base establishment unit is used for establishing a city health examination knowledge base in the form of knowledge graph from three aspects of theme, index and method according to the preset multi-department analysis and evaluation index group, the city health examination knowledge instance resources and the association clustering method.

[0028] As an improvement of the above-mentioned scheme, the knowledge base establishing unit is further configured to: acquire a preset multi-department analysis and evaluation index group; the multi-department analysis and evaluation index group comprises analysis and evaluation indexes and corresponding analysis and evaluation index connotations; calculate the analysis and evaluation indexes by using an implicit Dirichlet distribution model to establish an index base; perform feature mining on the analysis and evaluation index connotations to establish a method base; refine the analysis and evaluation indexes according to an association clustering method to acquire a theme group, and perform clustering analysis on the theme group to acquire a theme base; and establish a city physical examination knowledge base according to the index base, the method base and the theme base.

[0029] As an improvement of the above-mentioned scheme, the knowledge base establishing unit is further configured to: establish an analysis and evaluation index set according to analysis and evaluation indexes; perform data preprocessing on the analysis and evaluation index set to acquire a department analysis and evaluation index set; the department analysis and evaluation index set comprises a plurality of department analysis and evaluation indexes; calculate and confirm index keywords corresponding to the department analysis and evaluation indexes according to an implicit Dirichlet distribution model and a preset target function corresponding to keywords, and establish an index base according to entanglement, the department analysis and evaluation indexes and the index keywords.

[0030] As an improvement of the above-mentioned scheme, the knowledge base establishing unit is further configured to: under different preset theme connotations, calculate theme semantic similarity, index semantic similarity, index unit similarity and calculation method similarity according to the analysis and evaluation indexes by using JS divergence; calculate theme connotation association degrees between various theme connotations according to a preset theme connotation association degree formula, theme semantic similarity, index semantic similarity, index unit similarity and calculation method similarity; perform clustering on the theme connotations according to an association clustering algorithm to obtain a theme group; the theme group comprises a plurality of theme keywords; calculate similarity between the theme keywords and the theme connotations according to a preset similarity calculation formula, and perform clustering on the theme keywords and the theme connotations according to a preset classification threshold and the similarity to obtain a theme base.

[0031] As an improvement of the above-mentioned scheme, the classification and induction unit is further configured to: obtain a preset domain category and a preset expert experience library; determine a city physical examination ontology range corresponding to each domain category from the city physical examination knowledge according to the expert experience library and the domain category; extract a first keyword group and a second keyword group from the city physical examination knowledge range respectively by a preset first high-frequency calculation method and a preset second high-frequency calculation method, and fuse and deduplicate the first keyword group and the second keyword group to obtain a domain concept corresponding to each domain category; obtain a domain relationship between each domain category by iterative extraction according to a preset second external knowledge base, a preset matching method and a preset scoring evaluation formula; and construct a city physical examination ontology library according to the domain concept and the domain relationship.

[0032] As an improvement of the above-mentioned scheme, the classification and induction unit is further configured to: refer to a preset second external knowledge base to obtain a relationship type between each entity and store the relationship type and city physical examination knowledge into an entity dictionary; match the city physical examination knowledge and the relationship type by a boot-removal method and store the matching result into a candidate word table; evaluate the matching result in the candidate word table according to a preset scoring evaluation formula, and determine whether the evaluation scoring result converges according to a preset iteration number and a preset convergence accuracy; if the evaluation scoring result converges, output the relationship type corresponding to the evaluation scoring result as a domain relationship between corresponding domains.

[0033] As an improvement of the above-mentioned scheme, the knowledge management device further comprises a supplementary feedback unit configured to: supplement and improve the city physical examination ontology library reversely according to the city physical examination knowledge base.

[0034] Another embodiment of the present application provides a computer readable storage medium comprising a stored computer program, wherein the computer readable storage medium controls a device in which the computer readable storage medium is located to execute the knowledge management method of city physical examination knowledge as described above when the computer program is executed.

[0035] Another embodiment of the present application provides a knowledge management system of city physical examination knowledge, which comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the knowledge management method of city physical examination knowledge as described above when the computer program is executed.

[0036] Compared with the prior art, the technical scheme has the following beneficial effects:

[0037] The application provides a city physical examination knowledge management method, device, computer readable storage medium and system, which establishes a city physical examination knowledge base in the form of a knowledge graph according to a preset multi-department analysis and evaluation index group, city physical examination knowledge instance resources and an association clustering method from three aspects of a theme, an index and a method, thereby establishing a standardized framework based on different analysis and evaluation indexes, and realizing the universality and compatibility of knowledge management, and improving the knowledge management efficiency of city physical examination knowledge.

[0038] Further, the application provides a city physical examination knowledge management method, device, computer readable storage medium and system, which further supplements and improves the city physical examination ontology base in a timely manner according to the city physical examination knowledge base, thereby improving the accuracy of city physical examination knowledge management. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 is a flowchart of a city physical examination knowledge management method provided by an embodiment of the application;

[0040] Figure 2 is a structural schematic diagram of a city physical examination knowledge management device provided by an embodiment of the application. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the application. Embodiment one

[0043] The embodiment of the application first describes a city physical examination knowledge management method. Figure 1 is a flowchart of a city physical examination knowledge management method provided by an embodiment of the application.

[0044] As shown in Figure 1 , the knowledge management method comprises:

[0045] S1: collecting city physical examination data by a subject crawler data collection tool, and cleaning the city physical examination data to obtain city physical examination knowledge.

[0046] Specifically, a subject crawler data collection tool is used to extract structured features of a webpage, and after data cleaning operation, city physical examination knowledge is extracted from a city physical examination field website. The city physical examination data includes city physical examination historical reports, city planning related data, city physical examination administrative evolution data, industry data, and external city knowledge base.

[0047] S2: According to the preset field category and the preset expert experience library, the city physical examination knowledge is classified and summarized to obtain a city physical examination ontology library.

[0048] Specifically, the field categories include humanities, ecological environment, and economy, etc. Specifically, the city physical examination ontology range is determined according to the three types of field knowledge of humanities, ecological environment, and economy; and then a semantic-based concept and relationship classification system is determined for the three types of field knowledge.

[0049] In one embodiment, according to the preset field category and the preset expert experience library, the city physical examination knowledge is classified and summarized to obtain a city physical examination ontology library, specifically including: obtaining a preset field category and a preset expert experience library; determining the city physical examination ontology range corresponding to each field category from the city physical examination knowledge according to the expert experience library and the field category; extracting a first keyword group and a second keyword group from the city physical examination knowledge range by a preset first high-frequency calculation method and a preset second high-frequency calculation method, respectively, and fusing and deduplicating the first keyword group and the second keyword group to obtain the field concept corresponding to each field category; iteratively extracting to obtain the field relationship between each field category according to a preset second external knowledge base, a preset matching method, and a preset scoring evaluation formula; and constructing a city physical examination ontology library according to the field concept and the field relationship.

[0050] In one embodiment, the preset second external knowledge base is Schema.org and DBpedia, etc. The preset first high-frequency calculation method is TF-IDF algorithm, and the preset second high-frequency calculation method is TextRank algorithm.

[0051] In one embodiment, the scoring evaluation formula is specifically:

[0052]

[0053] Wherein, N is the total number of candidate relationships mined by the candidate word table P, and F is the number of relationship entities in the candidate word table P that have been in the entity dictionary; The precision of the candidate word table P is represented by log2(F+1), and the recall ability of the candidate word table P is represented by log2(F+1).

[0054] S3: Constructing city health check knowledge instance resources according to the preset first external knowledge base and the city health check ontology base.

[0055] Specifically, the city health check knowledge in the city health check ontology base is segmented and stop words are removed, the part-of-speech of the city health check field knowledge is marked by referring to the preset first external knowledge base and the entities, attributes and relationships in the city health check ontology base, and the Chinese entities and relationships are extracted by using a Bi-GRU model based on character-level attention and sentence-level attention. The extracted knowledge instances are linked to external knowledge bases such as Baidu Encyclopedia and Wikipedia, to form a web version of structured links of knowledge instances, and to enrich the knowledge instance resources.

[0056] S4: Establishing a city health check knowledge base in the form of a knowledge graph from the aspects of theme, index and method according to the preset multi-department analysis and evaluation index group, the city health check knowledge instance resources and the associated clustering method.

[0057] By investigating the knowledge needs of city health check related departments, a city health check knowledge base construction method is proposed, which is oriented to multiple theme evaluation, has multiple index analysis and multiple method calculation, and forms hierarchical expression and information description of theme, index and method. A multi-level linkage operation closed loop is formed among the theme base, the index base and the method base.

[0058] In one embodiment, a city health check knowledge base is established from the aspects of theme, index and method according to the preset multi-department analysis and evaluation index group, the city health check knowledge instance resources and the associated clustering method, specifically including: obtaining the preset multi-department analysis and evaluation index group; calculating the analysis and evaluation index by using a latent Dirichlet allocation model to establish an index base; mining the features of the analysis and evaluation index connotation to establish a method base; refining the analysis and evaluation index according to the associated clustering method to obtain a theme group, and clustering analyzing the theme group to obtain a theme base; and establishing a city health check knowledge base according to the index base, the method base and the theme base. The multi-department analysis and evaluation index group includes analysis and evaluation indexes of different departments and corresponding analysis and evaluation index connotations.

[0059] In one embodiment, the analysis evaluation index is calculated by using an implicit Dirichlet distribution model to establish an index library, specifically including: establishing an analysis evaluation index set according to the analysis evaluation index; data preprocessing of the analysis evaluation index set to obtain a department analysis evaluation index set; the department analysis evaluation index set includes a plurality of department analysis evaluation indexes; according to the implicit Dirichlet distribution model and the preset keyword corresponding target function, the index keyword corresponding to the department analysis evaluation index is calculated and confirmed, and the index library is established according to the entanglement degree, the department analysis evaluation index and the index keyword. Wherein, the preprocessing process includes word segmentation and stop word removal.

[0060] In one embodiment, the preset keyword corresponding target function is specifically:

[0061]

[0062] Wherein, α and β are iterated hyperparameters. According to θ and k, the probability distribution of each index keyword in the index and the probability distribution of each index keyword in the theme can be obtained. Through iterative calculation, the numerical value converges, and the index keyword corresponding to the index is obtained by probability calculation.

[0063] The index connotation is a specific representation of the meaning of the index and how to calculate. The index connotation generally has characteristic words such as "account for", "proportion", "percentage", "average", "each", "total amount", "quantity", "rate" and the like. In one embodiment, the analysis evaluation index connotation is mined to establish a method library, specifically including: adding these characteristic words to the word segmentation vocabulary to ensure the quality of professional vocabulary word segmentation; applying the TextRank algorithm to mine the co-occurrence relationship of index connotation words and characteristic words, selecting 2-8 characteristic words as variable parameters; through the position relationship R-position of the variable parameters and the characteristic words, setting rule constraints to determine the logical operation relationship R-operation, combining the variable parameters V and the logical operation O into a calculation method ME.

[0064] For example, "the number of historical and cultural blocks in the city, county and district that have carried out protection and repair projects in the past 5 years accounts for a percentage of the total number of historical and cultural blocks", the variable parameters are composed of "city, county and district", "in the past 5 years", "protection and repair project", "historical and cultural block quantity", "account for", "historical and cultural block total amount", "percentage" by using the TextRank algorithm. From "percentage", it can be judged that the operation is division, the variable parameter before "account for" is the numerator, and the variable parameter after "account for" is the denominator. In order to meet the demand of variable parameter data retrieval from knowledge instances, the variable parameters are further combined according to the knowledge instances to realize variable adaptation; then, a plurality of calculation method entities ME are established to associate the calculation method word table MME of the index. i}, i < 170, meanwhile, according to the index evaluation intention, the positive and negative evaluation strategies S entities are evaluated, and the evaluation strategy interval table MS = {ms1, ms2,..., ms 10}; finally, the index library LI can call the calculation method and evaluation strategy from the method library LM according to the index keyword and feature word matching.

[0065] In one embodiment, the analysis evaluation index is refined to obtain a theme group according to an association clustering method, and the theme group is subjected to clustering analysis to obtain a theme library, specifically including: under different preset theme connotations, theme semantic similarity, index semantic similarity, index unit similarity and calculation method similarity are calculated according to the analysis evaluation index by JS divergence; theme connotation association degrees between each theme connotation are calculated according to a preset theme connotation association degree formula, theme semantic similarity, index semantic similarity, index unit similarity and calculation method similarity; the theme connotations are clustered according to an association clustering algorithm to obtain a theme group; the theme group includes multiple theme keywords; similarity between the theme keywords and the theme connotations is calculated according to a preset similarity calculation formula, and the theme keywords and the theme connotations are clustered to obtain a theme library according to a preset classification threshold and the similarity.

[0066] Under different preset theme connotations, theme semantic similarity, index semantic similarity, index unit similarity and calculation method similarity are calculated according to the analysis evaluation index by JS divergence, specifically referring to calculating the similarity of the index and the calculation method under the theme connotation from the perspectives of index entity and method entity using JS divergence. The probability distribution of entity e1 and entity e2 is P1 and P2 respectively, and the similarity between the entities can be expressed as JS(P1||P2):

[0067]

[0068] Wherein, D KL (P1||P2) represents the KL divergence of P1 and P2, and its expression is:

[0069]

[0070] In one embodiment, the preset theme connotation association degree formula is specifically:

[0071] SC = a * theme semantic similarity + b * index semantic similarity + c * index unit similarity + d

[0072] Calculation method similarity;

[0073] Wherein, SC is the theme connotation correlation degree; a, b, c, d are weight factors respectively, and a, b, c, d belong to (0, 1).

[0074] According to the preset classification threshold and the similarity, the theme vocabulary and the theme connotation are clustered to obtain a theme library, specifically, through bottom-up hierarchical clustering, a three-level knowledge of index / method-theme connotation-theme is organized, based on the content of the theme library, the index is associated and extracted to form an index library, and the response to the theme is realized. The calculation method of the index library is obtained from the method library, and the dynamic calculation and update of the index are realized. In an embodiment, the preset similarity calculation formula is specifically:

[0075]

[0076] Wherein, S is the theme vocabulary; C is the theme connotation.

[0077] In an embodiment, according to the preset second external knowledge base, the preset matching method and the preset scoring evaluation formula, iterative extraction is performed to obtain the domain relationship between each domain category, specifically including: referring to the preset second external knowledge base, the relationship types between each entity are inducted and obtained, and the relationship types and the city health examination knowledge are stored in an entity dictionary; through the boot method, the city health examination knowledge is matched with the relationship types, and the matching results are stored in a candidate word table; according to the preset scoring evaluation formula, the matching results in the candidate word table are evaluated and scored, and according to the preset iteration number and the preset convergence accuracy, it is judged whether the evaluation and scoring results converge; if converging, the relationship type corresponding to the evaluation and scoring results is output as the domain relationship between the corresponding domains.

[0078] In an embodiment, the knowledge management method further includes: according to the city health examination knowledge base, the city health examination ontology library is supplemented and improved in reverse.

[0079] Embodiments of the present application describe a knowledge management method of city health examination knowledge, which establishes a city health examination knowledge base in the form of a knowledge graph through three aspects of theme, index and method according to a preset multi-department analysis and evaluation index group, city health examination knowledge instance resources and an associated clustering method, thereby establishing a standardized framework based on different analysis and evaluation indexes, and realizing the universality and compatibility of knowledge management. The knowledge management method improves the knowledge management efficiency of city health examination knowledge. Further, the knowledge management method of city health examination knowledge described in embodiments of the present application further supplements and improves the city health examination ontology library in reverse in time according to the city health examination knowledge base, thereby improving the accuracy of knowledge management of city health examination knowledge. Specific embodiment two

[0081] In addition to the above method, the embodiment of the present application further discloses a knowledge management device of city physical examination knowledge. Figure 2 FIG. 1 is a structural schematic diagram of a knowledge management device of city physical examination knowledge provided by an embodiment of the present application.

[0082] As shown in the figure, the knowledge management device comprises a data acquisition unit, a classification and induction unit, an instance construction unit and a knowledge base establishment unit. Figure 2 The data acquisition unit is configured to acquire city physical examination data by using a subject crawler data acquisition tool, and clean the city physical examination data to obtain city physical examination knowledge. The classification and induction unit is configured to classify and induce the city physical examination knowledge according to a preset field category and a preset expert experience library, and obtain a city physical examination ontology library. The instance construction unit is configured to construct city physical examination knowledge instance resources according to a preset first external knowledge base and the city physical examination ontology library. The knowledge base establishment unit is configured to establish a city physical examination knowledge base in the form of a knowledge graph from three aspects of a theme, an index and a method according to a preset multi-department analysis and evaluation index group, the city physical examination knowledge instance resources and an association clustering method.

[0083] In one embodiment, the knowledge base establishment unit is further configured to: acquire a preset multi-department analysis and evaluation index group; the multi-department analysis and evaluation index group comprises analysis and evaluation indexes and corresponding analysis and evaluation index connotations; calculate the analysis and evaluation indexes by using a latent Dirichlet allocation model to establish an index library; perform feature mining on the analysis and evaluation index connotations to establish a method library; refine the analysis and evaluation indexes according to an association clustering method to obtain a theme group, and perform clustering analysis on the theme group to obtain a theme library; and establish a city physical examination knowledge base according to the index library, the method library and the theme library.

[0084] In one embodiment, the knowledge base establishment unit is further configured to: establish an analysis and evaluation index set according to analysis and evaluation indexes; perform data preprocessing on the analysis and evaluation index set to obtain a department analysis and evaluation index set; the department analysis and evaluation index set comprises a plurality of department analysis and evaluation indexes; calculate and confirm index keywords corresponding to the department analysis and evaluation indexes according to a latent Dirichlet allocation model and a preset keyword corresponding objective function, and establish an index library according to entanglement, the department analysis and evaluation indexes and the index keywords.

[0085] In an embodiment, the knowledge base establishing unit is further configured to: calculate, according to the analysis evaluation indexes, a theme semantic similarity, an index semantic similarity, an index unit similarity and a calculation method similarity respectively by JS divergence under different preset theme connotations; calculate a theme connotation correlation degree between each theme connotation according to a preset theme connotation correlation degree formula, the theme semantic similarity, the index semantic similarity, the index unit similarity and the calculation method similarity; cluster the theme connotations according to a correlation clustering algorithm to obtain a theme group; the theme group includes a plurality of theme glossaries; calculate a similarity between the theme glossaries and the theme connotations according to a preset similarity calculation formula, and cluster the theme glossaries and the theme connotations according to a preset classification threshold and the similarity to obtain a theme library.

[0086] In an embodiment, the classification induction unit is further configured to: obtain a preset domain category and a preset expert experience library; determine a city physical examination ontology range corresponding to each domain category from the city physical examination knowledge according to the expert experience library and the domain category; extract a first keyword group and a second keyword group from the city physical examination knowledge range respectively by a preset first high-frequency calculation method and a preset second high-frequency calculation method, fuse and remove duplicates of the first keyword group and the second keyword group to obtain a domain concept corresponding to each domain category; iteratively extract to obtain a domain relationship between each domain category according to a preset second external knowledge base, a preset matching method and a preset scoring evaluation formula; and construct a city physical examination ontology library according to the domain concept and the domain relationship.

[0087] In an embodiment, the classification induction unit is further configured to: refer to a preset second external knowledge base, induce to obtain a relationship type between each entity, and store the relationship type and the city physical examination knowledge into an entity dictionary; match the city physical examination knowledge and the relationship type by a boot-removing method, and store a matching result into a candidate word table; evaluate a score of the matching result in the candidate word table according to a preset scoring evaluation formula, and determine whether the evaluation score result converges according to a preset iteration number and a preset convergence accuracy; if the evaluation score result converges, output the relationship type corresponding to the evaluation score result as a domain relationship between corresponding domains.

[0088] In an embodiment, the knowledge management device further includes a supplementary feedback unit configured to: supplement and improve the city physical examination ontology library reversely according to the city physical examination knowledge base.

[0089] Another embodiment of the present application provides a computer readable storage medium, comprising a stored computer program, wherein the computer readable storage medium controls a device where the computer readable storage medium is located to execute the knowledge management method of urban health examination knowledge as described above when the computer program is running.

[0090] The units integrated in the knowledge management device can be stored in a computer readable storage medium if they are realized in the form of software functional units and sold or used as independent products. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. The computer program can implement the steps of each method embodiment when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0091] It should be noted that the above-described device embodiments are only schematic, and the units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment. In addition, the connection relationship between the units in the device embodiment provided by the present application indicates that there is a communication connection between them, which can be realized as one or more communication buses or signal lines. Those skilled in the art can understand and implement it without creative labor.

[0092] The embodiment of the present application describes a knowledge management device and computer readable storage medium for urban physical examination knowledge, which establishes an urban physical examination knowledge base in the form of a knowledge graph according to a preset multi-department analysis and evaluation index group, urban physical examination knowledge instance resources and an association clustering method from three aspects of a theme, an index and a method, thereby establishing a standardized framework based on different analysis and evaluation indexes, and realizing the universality and compatibility of knowledge management, and improving the knowledge management efficiency of urban physical examination knowledge. Further, the embodiment of the present application describes a knowledge management device and computer readable storage medium for urban physical examination knowledge, which timely supplements and improves the urban physical examination ontology base according to the urban physical examination knowledge base, thereby improving the accuracy of knowledge management of urban physical examination knowledge. Embodiment three

[0094] In addition to the above method and device, the embodiment of the present application also describes a knowledge management system for urban physical examination knowledge.

[0095] Specifically, the knowledge management system comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the knowledge management method for urban physical examination knowledge when executing the computer program.

[0096] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor is the control center of the device, and connects various parts of the device through various interfaces and lines.

[0097] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the device by running or executing the computer program and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function (such as a sound playing function, an image playing function, etc.), and the like; and the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), and the like. In addition, the memory can include a high-speed random access memory, and can also include a nonvolatile memory, for example, a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state memory devices.

[0098] The embodiment of the present application describes a knowledge management system of urban physical examination knowledge, which establishes an urban physical examination knowledge base in the form of a knowledge graph according to a preset multi-department analysis and evaluation index group, the urban physical examination knowledge instance resource and an associated clustering method from three aspects of a theme, an index and a method, thereby establishing a standardized framework based on different analysis and evaluation indexes, and realizing the universality and compatibility of knowledge management, and the knowledge management system improves the knowledge management efficiency of urban physical examination knowledge; further, the embodiment of the present application describes a knowledge management system of urban physical examination knowledge, which timely supplements and perfects the urban physical examination ontology base in reverse according to the urban physical examination knowledge base, thereby improving the accuracy of knowledge management of urban physical examination knowledge.

[0099] The above is the preferred embodiment of the present application, it should be noted that for those skilled in the art, without departing from the principles of the present application, can make a number of improvements and refinements, these improvements and refinements are also considered to be within the scope of protection of the present application.

Claims

1. A knowledge management method of urban physical examination knowledge, characterized in that, The knowledge management method comprises: collecting city physical examination data through a subject crawler data collection tool, and cleaning the city physical examination data to obtain city physical examination knowledge; according to a preset field category and a preset expert experience library, classifying and summarizing the city physical examination knowledge to obtain a city physical examination ontology library, specifically comprising: obtaining the preset field category and the preset expert experience library; determining the city physical examination ontology range corresponding to each field category from the city physical examination knowledge according to the expert experience library and the field category; extracting the first keyword group and the second keyword group from the city physical examination knowledge range respectively by using a preset first high-frequency calculation method and a preset second high-frequency calculation method, and fusing and deduplicating the first keyword group and the second keyword group to obtain the field concept corresponding to each field category; iteratively extracting to obtain the field relationship between each field category according to a preset second external knowledge base, a preset matching method and a preset scoring evaluation formula; and constructing the city physical examination ontology library according to the field concept and the field relationship; constructing a city physical examination knowledge instance resource according to a preset first external knowledge base and the city physical examination ontology library, specifically comprising: performing word segmentation and stop word removal on the city physical examination knowledge in the city physical examination ontology library, performing part-of-speech tagging on the city physical examination field knowledge by referring to the entities, attributes and relationships in the preset first external knowledge base and the city physical examination ontology library, and constructing the city physical examination knowledge instance resource; wherein a Bi-GRU model based on character-level attention and sentence-level attention is used for training to extract Chinese entities and relationships; from the aspects of theme, index and method, establishing a city physical examination knowledge base in the form of a knowledge graph according to a preset multi-department analysis and evaluation index group, the city physical examination knowledge instance resource and an association clustering method, specifically comprising: obtaining the preset multi-department analysis and evaluation index group; the multi-department analysis and evaluation index group comprises analysis and evaluation indexes and corresponding analysis and evaluation index connotations; calculating the analysis and evaluation indexes by using a latent Dirichlet allocation model to establish an index library; performing feature mining on the analysis and evaluation index connotations to establish a method library; refining the analysis and evaluation indexes according to the association clustering method to obtain a theme group, and performing clustering analysis on the theme group to obtain a theme library; and establishing a city physical examination knowledge base according to the index library, the method library and the theme library.

2. The knowledge management method of urban census knowledge according to claim 1, characterized in that, calculating the analysis and evaluation indexes by using a latent Dirichlet allocation model to establish an index library, specifically comprising: establishing an analysis and evaluation index set according to the analysis and evaluation indexes; performing data preprocessing on the analysis and evaluation index set to obtain a department analysis and evaluation index set; the department analysis and evaluation index set comprises a plurality of department analysis and evaluation indexes; calculating and confirming the index keywords corresponding to the department analysis and evaluation indexes according to the latent Dirichlet allocation model and a preset keyword corresponding target function, and establishing an index library according to entanglement, the department analysis and evaluation indexes and the index keywords.

3. The knowledge management method of urban census knowledge according to claim 1, characterized in that, According to the association clustering method, the analysis evaluation indexes are refined to obtain a theme group, and the theme group is subjected to clustering analysis to obtain a theme library, specifically including: Under different preset theme connotations, theme semantic similarity, index semantic similarity, index unit similarity and calculation method similarity are calculated according to the analysis evaluation indexes through JS divergence; According to the preset theme connotation association degree formula, theme semantic similarity, index semantic similarity, index unit similarity and calculation method similarity, theme connotation association degrees between each theme connotation are calculated; According to the association clustering algorithm, the theme connotations are clustered to obtain a theme group; the theme group includes multiple theme glossaries; According to a preset similarity calculation formula, the similarity between the theme glossary and the theme connotation is calculated, and according to a preset classification threshold and the similarity, the theme glossary and the theme connotation are clustered to obtain a theme library.

4. The knowledge management method of urban census knowledge according to claim 1, characterized in that, According to a preset second external knowledge base, a preset matching method and a preset scoring evaluation formula, iterative extraction is performed to obtain the domain relationship between each domain category, specifically including: Referring to the preset second external knowledge base, the relationship types between each entity are induced and obtained, and the relationship types and city physical examination knowledge are stored in an entity dictionary; The city physical examination knowledge is matched with the relationship types through the boot-removing method, and the matching results are stored in a candidate word table; According to the preset scoring evaluation formula, the matching results in the candidate word table are evaluated and scored, and according to a preset iteration number and a preset convergence accuracy, it is judged whether the evaluation and scoring results converge; If the convergence is obtained, the relationship type corresponding to the evaluation and scoring results is output as the domain relationship between the corresponding domains.

5. The knowledge management method of urban census knowledge according to any one of claims 1 to 4, characterized in that, The knowledge management method further includes: According to the city physical examination knowledge base, the city physical examination ontology base is supplemented and improved in reverse.

6. A knowledge management apparatus of city physical examination knowledge, characterized by, The knowledge management device includes a data acquisition unit, a classification induction unit, an instance construction unit and a knowledge base establishment unit, wherein, The data acquisition unit is used to acquire city physical examination data through a subject crawler data acquisition tool, and the city physical examination data is cleaned to obtain city physical examination knowledge; The classification and induction unit is configured to classify and induce the urban physical examination knowledge according to a preset domain category and a preset expert experience library to obtain an urban physical examination ontology library, specifically including: obtaining the preset domain category and the preset expert experience library; determining, from the urban physical examination knowledge, an urban physical examination ontology range corresponding to each domain category according to the expert experience library and the domain category; extracting, from the urban physical examination knowledge range, a first keyword group and a second keyword group respectively by using a preset first high-frequency calculation method and a preset second high-frequency calculation method, fusing and deduplicating the first keyword group and the second keyword group to obtain a domain concept corresponding to each domain category; iteratively extracting to obtain a domain relationship between each domain category according to a preset second external knowledge library, a preset matching method, and a preset scoring evaluation formula; and constructing the urban physical examination ontology library according to the domain concept and the domain relationship. The instance construction unit is configured to construct an urban physical examination knowledge instance resource according to a preset first external knowledge library and the urban physical examination ontology library, specifically including: performing word segmentation and stop word removal on the urban physical examination knowledge in the urban physical examination ontology library, performing part-of-speech tagging on the urban physical examination domain knowledge by referring to entities, attributes, and relationships in the preset first external knowledge library and the urban physical examination ontology library, and constructing the urban physical examination knowledge instance resource; and the Chinese entities and relationships are extracted by using a Bi-GRU model based on character-level attention and sentence-level attention. The knowledge base establishment unit is configured to establish an urban physical examination knowledge base in the form of a knowledge graph from three aspects of a theme, an index, and a method according to a preset multi-department analysis and evaluation index group, the urban physical examination knowledge instance resource, and an association clustering method, specifically including: obtaining the preset multi-department analysis and evaluation index group; the multi-department analysis and evaluation index group includes analysis and evaluation indexes and corresponding analysis and evaluation index connotations; calculating the analysis and evaluation indexes by using a latent Dirichlet allocation model to establish an index library; performing feature mining on the analysis and evaluation index connotations to establish a method library; refining the analysis and evaluation indexes by using an association clustering method to obtain a theme group, and performing clustering analysis on the theme group to obtain a theme library; and establishing the urban physical examination knowledge base according to the index library, the method library, and the theme library.

7. A computer-readable storage medium, characterized in that, The computer readable storage medium includes a stored computer program, wherein the computer readable storage medium controls a device in which the computer readable storage medium is located to execute the urban physical examination knowledge management method of any one of claims 1-5 when the computer program runs.

8. A knowledge management system for urban census knowledge, characterized by, The knowledge management system includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the urban physical examination knowledge management method of any one of claims 1-5 when the computer program is executed.

Images