CN115098696A - Method and device for constructing urban physical examination knowledge graph and storage medium - Google Patents
Method and device for constructing urban physical examination knowledge graph and storage medium Download PDFInfo
- Publication number
- CN115098696A CN115098696A CN202210695086.9A CN202210695086A CN115098696A CN 115098696 A CN115098696 A CN 115098696A CN 202210695086 A CN202210695086 A CN 202210695086A CN 115098696 A CN115098696 A CN 115098696A
- Authority
- CN
- China
- Prior art keywords
- scene
- physical examination
- knowledge
- city
- urban
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 239000002131 composite material Substances 0.000 claims abstract description 31
- 238000010276 construction Methods 0.000 claims abstract description 25
- 230000001502 supplementing effect Effects 0.000 claims abstract description 12
- 238000007499 fusion processing Methods 0.000 claims abstract description 10
- 238000000605 extraction Methods 0.000 claims description 21
- 238000012549 training Methods 0.000 claims description 14
- 238000013136 deep learning model Methods 0.000 claims description 9
- 230000004927 fusion Effects 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 239000000463 material Substances 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 abstract description 13
- 238000011156 evaluation Methods 0.000 description 17
- 238000010586 diagram Methods 0.000 description 10
- 238000004364 calculation method Methods 0.000 description 7
- 230000008859 change Effects 0.000 description 6
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 6
- 238000002372 labelling Methods 0.000 description 5
- 230000036541 health Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 230000033228 biological regulation Effects 0.000 description 3
- 238000010219 correlation analysis Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 239000000203 mixture Substances 0.000 description 3
- 238000013441 quality evaluation Methods 0.000 description 3
- 239000013589 supplement Substances 0.000 description 3
- RAHZWNYVWXNFOC-UHFFFAOYSA-N Sulphur dioxide Chemical compound O=S=O RAHZWNYVWXNFOC-UHFFFAOYSA-N 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000005180 public health Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- MGWGWNFMUOTEHG-UHFFFAOYSA-N 4-(3,5-dimethylphenyl)-1,3-thiazol-2-amine Chemical compound CC1=CC(C)=CC(C=2N=C(N)SC=2)=C1 MGWGWNFMUOTEHG-UHFFFAOYSA-N 0.000 description 1
- UGFAIRIUMAVXCW-UHFFFAOYSA-N Carbon monoxide Chemical compound [O+]#[C-] UGFAIRIUMAVXCW-UHFFFAOYSA-N 0.000 description 1
- 108091026890 Coding region Proteins 0.000 description 1
- 240000002853 Nelumbo nucifera Species 0.000 description 1
- 235000006508 Nelumbo nucifera Nutrition 0.000 description 1
- 235000006510 Nelumbo pentapetala Nutrition 0.000 description 1
- 244000183278 Nephelium litchi Species 0.000 description 1
- CBENFWSGALASAD-UHFFFAOYSA-N Ozone Chemical compound [O-][O+]=O CBENFWSGALASAD-UHFFFAOYSA-N 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000009411 base construction Methods 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 229910002091 carbon monoxide Inorganic materials 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 239000010985 leather Substances 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- JCXJVPUVTGWSNB-UHFFFAOYSA-N nitrogen dioxide Inorganic materials O=[N]=O JCXJVPUVTGWSNB-UHFFFAOYSA-N 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 229910052717 sulfur Inorganic materials 0.000 description 1
- 239000002352 surface water Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Tourism & Hospitality (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Biophysics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Animal Behavior & Ethology (AREA)
- Databases & Information Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a construction method, a device and a storage medium of a city physical examination knowledge graph, wherein the method comprises the following steps: acquiring city physical examination knowledge resources, and constructing a city physical examination knowledge base according to the city physical examination knowledge resources; constructing a multi-dimensional knowledge structure model of the urban physical examination according to an urban physical examination knowledge base based on the knowledge representation form and the knowledge structure of the urban physical examination scene; integrating various types of information in the urban physical examination multi-dimensional knowledge structure model to construct a urban physical examination network model fusing hypergraphs; extracting scene elements from data in a city physical examination knowledge base according to entity characteristics in the city physical examination network model to obtain meta-scene information; scene fusion processing is carried out on the meta-scene information to form a composite scene; and constructing the composite scene in a hierarchy mode to generate scene maps in different hierarchies, and supplementing and improving scene semantics in the scene maps to obtain the city physical examination knowledge map. The embodiment of the invention can effectively improve the comprehensiveness of city physical examination analysis.
Description
Technical Field
The invention relates to the technical field of knowledge graphs, in particular to a method and a device for constructing an urban physical examination knowledge graph and a storage medium.
Background
The urban physical examination work is carried out in partial cities in China, and results are formed. The achievements comprise abundant special basic data, indexes and index data in an evaluation system, complete analysis reports, a visual display system and the like. The urban physical examination work has been carried out for years, the results are also processed in a knowledge mode, and a knowledge base and a corresponding system about the urban physical examination knowledge are built. The knowledge contents, time span and granularity are not equal, the space span and granularity are not equal, and the urban physical examination scene relates to the complex field and the large span, but has an incidence relation. For example, the population age structure is associated with the endowment service, the endowment service is associated with the medical resource and the endowment facility, and the evaluation analysis of the medical resource and the endowment facility is associated with the structure, the density and the spatial distribution of the population. The complexity of the knowledge structure needs to be expressed by a reasonable knowledge model, and the complexity of the relation between the knowledge needs to be stored based on a graph database, so that the knowledge is reasonably constructed and expressed by adopting a knowledge graph.
The knowledge graph is based on semantic network, ontology, Web, semantic Web, and link data, and realizes the link from text to object. The knowledge graph is based on graph theory, and compared with the traditional relational database, the knowledge graph has the advantages that: new classes, entities and relationships can be added very conveniently; the computer can also process the knowledge in the computer language while storing and expressing the knowledge based on the semantics of the human language; the intelligence of human brain storage knowledge is simulated; in the face of complicated knowledge content, perfect logic is not required, and expression capacity is highlighted. The heat of the construction and application research of the knowledge graph continues, and the progress and development of the knowledge graph related technology are continuously promoted.
The existing construction method of the city physical examination knowledge graph cannot represent the mutual connection among multiple types of nodes, so that the city physical examination analysis is difficult to carry out comprehensively.
Disclosure of Invention
The invention provides a construction method, a device and a storage medium of an urban physical examination knowledge graph, and aims to solve the technical problem that the existing construction method of the urban physical examination knowledge graph cannot represent the interconnection among multiple types of nodes, so that the urban physical examination analysis is difficult to carry out comprehensively.
The embodiment of the invention provides a method for constructing a city physical examination knowledge graph, which comprises the following steps:
acquiring city physical examination knowledge resources, and constructing a city physical examination knowledge base according to the city physical examination knowledge resources; the urban physical examination knowledge resources comprise historical reports, administrative leather-following materials and industrial data materials in the urban physical examination field;
constructing a multi-dimensional knowledge structure model of the urban physical examination according to the urban physical examination knowledge base based on the knowledge representation form and the knowledge structure of the urban physical examination scene;
integrating various types of information in the urban physical examination multi-dimensional knowledge structure model to construct a hypergraph fused urban physical examination network model; wherein the plurality of types of models include: metrics, knowledge instances, attributes, and scenarios;
extracting scene elements of the structured data, the semi-structured data and the unstructured data in the urban physical examination knowledge base according to entity characteristics in the urban physical examination network model to obtain element scene information;
scene fusion processing is carried out on the meta scene information to form a composite scene;
and constructing the composite scene in layers to generate scene maps in different layers, and supplementing and perfecting scene semantics in the scene maps to obtain the city physical examination knowledge map.
Further, the acquiring city physical examination knowledge resources and constructing a city physical examination knowledge base according to the city physical examination knowledge resources include:
acquiring urban physical examination knowledge resources by adopting a crawler tool;
establishing a classification system based on the city physical examination ontology range to obtain a city physical examination initial body library;
and constructing an urban physical examination knowledge base according to the urban physical examination knowledge resources and the urban physical examination initial body base, wherein the urban physical examination knowledge base comprises knowledge instance resources, an index base and a subject base.
Further, the urban physical examination multi-dimensional knowledge structure model comprises a basic dimension and a scene dimension, wherein the basic dimension comprises an entity dimension, a time dimension and a space dimension, and the scene dimension comprises a knowledge instance dimension, an index dimension and a subject dimension.
Further, the method for constructing the urban physical examination network model fusing the hypergraph by integrating various types of information in the urban physical examination multidimensional knowledge structure model comprises the following steps:
constructing six relation sets according to the various types of information, wherein the six relation sets comprise: an index-index relationship, a knowledge instance-knowledge instance relationship, a knowledge instance-attribute relationship, an index-attribute relationship, and an index-knowledge instance relationship;
and constructing a city physical examination network model fusing the hypergraph by taking the various types of information as a vertex set of the hypergraph and taking the six types of relation sets as a super edge set of the hypergraph.
Further, the extracting scene elements from the structured data, the semi-structured data and the unstructured data in the urban physical examination knowledge base according to the entity characteristics in the urban physical examination network model to obtain meta scene information includes:
designing scene trigger words according to semantic rules, extracting scene elements of structured data, semi-structured data and unstructured data in the city physical examination knowledge base according to entity characteristics in the city physical examination network model based on the scene trigger words to obtain scene sentences, segmenting the scene sentences according to scene types and scene vocabularies, obtaining enhanced scene sentence data through marking and combining, and deeply inputting the enhanced scene sentence data into a deep learning model for training to obtain meta-scene information.
Further, the performing scene fusion processing on the meta-scene information to form a composite scene includes:
and calculating the relevance among the meta-scene information by adopting a similarity algorithm, and performing scene fusion on the meta-scene information according to the relevance to form a composite scene.
Further, the step of constructing the composite scene in layers to generate scene maps in different layers, and the step of supplementing and perfecting scene semantics in the scene maps to obtain the city physical examination knowledge map includes:
judging whether scene entities in the composite scene are aligned according to a preset scene level grading strategy, and judging the scene entities which are not aligned as unmatched meta-scenes;
if the unmatched meta-scenes are a single class of scenes, constructing a hierarchical scene spectrogram of the unmatched meta-scenes;
if the unmatched meta-scenes are not the single scene, taking the unmatched scenes as other scenes, and reconstructing a scene graph after a new scene is added into the unmatched scenes;
scene semantics in the scene map are supplemented and perfected in a scene condensing mode, and the city physical examination knowledge map is obtained.
One embodiment of the invention provides a construction device of an urban physical examination knowledge graph, which comprises the following steps:
the city physical examination knowledge base building module is used for acquiring city physical examination knowledge resources and building a city physical examination knowledge base according to the city physical examination knowledge resources; the urban physical examination knowledge resources comprise historical reports, administrative leather-following materials and industrial data materials in the urban physical examination field;
the city physical examination multi-dimensional knowledge structure model building module is used for building a city physical examination multi-dimensional knowledge structure model according to the city physical examination knowledge base based on the knowledge representation form and the knowledge structure of the city physical examination scene;
the city physical examination network model building module is used for building a city physical examination network model fusing hypergraphs by integrating various types of information in the city physical examination multi-dimensional knowledge structure model; wherein the plurality of types of models include: metrics, knowledge instances, attributes, and scenarios;
the scene element extraction module is used for extracting scene elements from the structured data, the semi-structured data and the unstructured data in the urban physical examination knowledge base according to entity characteristics in the urban physical examination network model to obtain meta scene information;
the scene fusion module is used for carrying out scene fusion processing on the meta-scene information to form a composite scene;
and the knowledge map construction module is used for constructing the composite scene in a hierarchy manner to generate scene maps in different hierarchies, and supplementing and improving scene semantics in the scene maps to obtain the urban physical examination knowledge map.
An embodiment of the present invention provides a computer storage medium, on which a computer program is stored, and the computer program, when executed by a processor, implements the method for constructing the physical examination knowledge graph of the city as described above.
The embodiment of the invention is based on the knowledge representation form and the knowledge structure of the city physical examination scene, the city physical examination multi-dimensional knowledge structure model is constructed according to the city physical examination knowledge base, and the city physical examination information can be uniformly expressed, so that the multi-dimensional knowledge of the city physical examination can be comprehensively expressed; according to the embodiment of the invention, the urban physical examination network model fused with the hypergraph is constructed by integrating various types of information in the urban physical examination multi-dimensional knowledge structure model, and the co-occurrence among scene element nodes can be effectively depicted by connecting the hyperedges of the scene element nodes, so that the convenience of carrying out correlation analysis on the urban physical examination scene can be effectively improved; according to the embodiment of the invention, scene element extraction is carried out on the structured data, the semi-structured data and the unstructured data in the urban physical examination knowledge base according to the entity characteristics in the urban physical examination network model, so that the accuracy of scene element extraction can be effectively improved.
Drawings
FIG. 1 is a schematic flow chart of a method for constructing a knowledge graph of urban physical examination according to an embodiment of the present invention;
FIG. 2 is another schematic flow chart of a method for constructing a city physical examination knowledge graph according to an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a multidimensional knowledge structural model for urban physical examination according to an embodiment of the present invention;
FIG. 4 is a schematic structural diagram of a physical examination network model of a city with a merged supergraph according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a formalized multi-square hypergraph according to an embodiment of the invention;
FIG. 6 is a schematic diagram of a scene recognition and extraction process of a BERT-Bi LSTM-CRF model training provided by the embodiment of the present invention;
fig. 7 is a schematic flowchart of a method for extracting an urban physical examination scene according to an embodiment of the present invention;
FIG. 8 is a schematic structural diagram of a city health knowledge map system according to an embodiment of the present invention;
fig. 9 is a schematic structural diagram of a construction apparatus for an urban physical examination knowledge graph according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application without making any creative effort belong to the protection scope of the present application.
In the description of the present application, it is to be understood that the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or as implying that the number of indicated technical features is indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present application, the meaning of "a plurality" is two or more unless otherwise specified.
In the description of the present application, it is to be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meaning of the above terms in the present application can be understood in a specific case by those of ordinary skill in the art.
Referring to fig. 1, an embodiment of the present invention provides a method for constructing a city physical examination knowledge graph, including:
s1, acquiring urban physical examination knowledge resources, and constructing an urban physical examination knowledge base according to the urban physical examination knowledge resources; the urban physical examination knowledge resources comprise historical reports, administrative leather-following materials and industrial data materials in the urban physical examination field;
s2, constructing a multi-dimensional knowledge structure model of the urban physical examination according to an urban physical examination knowledge base based on the knowledge representation form and the knowledge structure of the urban physical examination scene;
s3, integrating various types of information in the urban physical examination multi-dimensional knowledge structure model to construct an urban physical examination network model fusing hypergraphs; wherein the plurality of types of models include: metrics, knowledge instances, attributes, and scenarios;
s4, extracting scene elements of the structured data, the semi-structured data and the unstructured data in the urban physical examination knowledge base according to the entity characteristics in the urban physical examination network model to obtain meta scene information;
s5, carrying out scene fusion processing on the meta scene information to form a composite scene;
s6, constructing the composite scene in a hierarchical manner to generate scene maps in different levels, and supplementing and improving scene semantics in the scene maps to obtain the city physical examination knowledge map.
The embodiment of the invention is based on the knowledge representation form and the knowledge structure of the city physical examination scene, the city physical examination multi-dimensional knowledge structure model is constructed according to the city physical examination knowledge base, and the city physical examination information can be uniformly expressed, so that the multi-dimensional knowledge of the city physical examination can be comprehensively expressed; according to the embodiment of the invention, the urban physical examination network model fused with the hypergraph is constructed by integrating various types of information in the urban physical examination multi-dimensional knowledge structure model, and the co-occurrence among scene element nodes can be effectively drawn through the hyperedges connecting the scene element nodes, so that the convenience of carrying out correlation analysis on the urban physical examination scene can be effectively improved; according to the embodiment of the invention, scene element extraction is carried out on the structured data, the semi-structured data and the unstructured data in the urban physical examination knowledge base according to the entity characteristics in the urban physical examination network model, so that the accuracy of scene element extraction can be effectively improved.
Referring to fig. 2, another flow chart of the method for constructing a city physical examination knowledge graph according to the embodiment of the present invention is shown.
In one embodiment, acquiring the urban physical examination knowledge resources, and constructing the urban physical examination knowledge base according to the urban physical examination knowledge resources comprises:
s11, acquiring urban physical examination knowledge resources by using a crawler tool;
in the embodiment of the invention, the urban physical examination knowledge resources also comprise resources of an external knowledge base, and the urban knowledge resources are crawled by adopting a crawler tool.
S12, establishing a classification system based on the urban physical examination ontology range to obtain an urban physical examination initial body library;
in the embodiment of the invention, the city physical examination ontology range can be determined by the domain expert, the city physical examination ontology range comprises three types of domain knowledge of humanity, ecological environment and economy, and a classification system is further established for the concept and the relation of the city physical ontology range to obtain the city physical examination initial body library. The embodiment of the invention also utilizes the TF-IDF algorithm and the TextRank algorithm to jointly calculate high-frequency vocabularies for the urban physical examination knowledge resources, and supplements and perfects concepts.
S13, constructing an urban physical examination knowledge base according to the urban physical examination knowledge resources and the urban physical examination initial body base, wherein the urban physical examination knowledge base comprises knowledge instance resources, an index base and a subject base.
In the embodiment of the invention, the establishing of the knowledge instance resources specifically comprises the steps of carrying out part-of-speech tagging on the city physical examination knowledge resources by referring to entities, attributes and relations in a city physical examination ontology library, adopting a Bi-GRU model based on character-level attention and sentence-level attention to train and extract Chinese entities and relations, and forming webpage version structured links of the knowledge instances and establishing the knowledge instance resources by linking the extracted knowledge instances with external knowledge bases such as encyclopedia, Wikipedia and the like.
The construction mode of the index library is as follows: index sets are respectively established for the acquired urban physical examination analysis and evaluation indexes of the natural resource part and the living establishment part, data preprocessing operations such as word segmentation and word removal and stop are respectively carried out, word, index key words and index three-layer structure potential theme information are extracted from the index sets by adopting an LDA model, integration is realized according to the theme information reflected by 2 sets of urban physical examination analysis and evaluation indexes of the natural resource part and the living establishment part, and an index library is established.
The method library is constructed in the following way: feature mining is conducted on index connotations of the urban physical examination analysis and evaluation indexes, a TextRank algorithm is used for mining the co-occurrence relation between index connotation words and feature words, then a rule constraint is set through the position relation R-position of variable parameters and feature words to determine a logic operation relation R-operation, the variable parameters V and the logic operation O are combined into a calculation method ME, finally, calculation method word lists of correlation indexes are established for multiple calculation method entities ME, and a method library is established.
The construction mode of the subject library is as follows: and further refining the city physical examination analysis and evaluation indexes by using an associated clustering method to form theme connotations, carrying out similarity analysis on the theme and the theme connotations to obtain the theme and corresponding connotation classifications thereof, and organizing the contents of the theme and the corresponding connotations and mapping relations thereof to establish a theme library.
After the knowledge instance resources, the index library, the method library and the subject library are respectively constructed, the urban physical examination knowledge base which is supported by the underlying knowledge instance and is evaluated by multiple subjects, has multi-type index analysis and multiple methods for calculation is formed, and knowledge supplement and the ontology library can be further refined through the constructed urban physical examination knowledge base.
In one embodiment, the urban physical examination multi-dimensional knowledge structure model comprises a base dimension and a scene dimension, wherein the base dimension comprises an entity dimension, a time dimension and a space dimension, and the scene dimension comprises a knowledge instance dimension, an index dimension and a subject dimension.
In the embodiment of the invention, the urban physical examination knowledge is dependent on the variation characteristics of the urban physical examination entity, so that the urban physical examination entity has space-time dynamics. Therefore, in the embodiment of the present invention, an urban physical Examination Multi-dimensional Knowledge structure model (Multi-dimensional Knowledge of City exploration, MKCE) is constructed, in which a measurement entity is an entity dimension of urban physical Examination, a life cycle is a time dimension of urban physical Examination, a geometric form is a space dimension of urban physical Examination, and an organization unit carrying measurement evaluation is a scene dimension of urban physical Examination.
Please refer to fig. 3, wherein S, T, E belongs to the basic dimensions, which respectively represent Space (Space), Time (Time), and city health Entity (Entity); sc belongs to scene dimensions and comprises three knowledge structures of knowledge Instance (Instance), index (Metrics) and subject (same).
The mathematical expression of MKCE is as follows:
MKCE＝{T,S,E,Sc＝(I,M,T)}#(1)。
the embodiment of the invention analyzes each dimension of the urban physical examination knowledge, and performs mathematical definition and expression, and the mathematical definition and expression are specifically expressed as follows:
basic dimension:
the entity dimension is an important composition and a reference of the city physical examination entity description and describes the city physical examination entity classification. Referring to the ontology structure and the knowledge system in the urban physical examination knowledge base, the entities are divided into three categories according to three fields of humanity, ecological environment and economy. Wherein, the entity is divided into 5 types of population structure, science and education and civilization, industry, facility and zone bit by taking humanity as a core concept; dividing the ecological environment into 4 types of water, air, land and disasters; and the entity is divided into an economic type, a GDP type and a employment 3 type by taking the economy as a core. Each type is subdivided into subclasses, and each subclass corresponds to each entity. The Entity Information (Entity Information) is expressed in Class (Category) -Class (Class) -Subclass (Subclass) -Entity (Entity) level four scale axis, namely:
EI＝(E,Subc,C,Cat)#(2)。
the time dimension is an important composition and a reference of the description of the urban physical examination entity, and the changes of the space, the geometry, the state and the like of the entity are continuously generated and accumulated along with the time. And the ratio of UTC + 08: 00 is a time reference, and a time system is established according to three-level time scales of year, month, day and the like. Wherein T (T) b ,T e ) Time domain of physical status of city, i.e. time of occurrence T of change b And time of death T e . For three time scales, time-operated operations are respectively established, i.e.
The spatial dimension is an important composition and a reference of city physical examination entity description, and describes geometrical characteristics such as spatial position, shape and size of the entity. The city physical examination knowledge granularity is greatly influenced by the space granularity, and the knowledge generation is also based on the space analysis. Therefore, with reference to the national standard 'geospatial network coding rule', the coding rule, the coding sequence and the coding calculation method of the urban physical examination entity are established; on the basis of referring to the coding of Chinese provinces, cities, counties and regions, the two-level administrative division scale description of the cities, the counties and the regions is established. The MKCE model is regarded as a closed set space, and set operation is introduced into MKCE space operation. The variation information in the MKCE model can be obtained through space-time operation. For example the entity "permeable floor" corresponding to the index term "urban permeable floor area ratio". Supposing that a built area of a city is impermeable ground, the original impermeable ground is reconstructed into permeable ground with a space-time operation relation of ' erasing ', ' erased ' and ' intersection ', a ' union ' relation of ' permeable ground ' is newly added on the original basis, and a ' symmetric difference ' relation of space-time change of the permeable ground ' in a period of time exists. The increased and decreased 'permeable ground' in 2020-2021 can be obtained by the above 5 arithmetic operations.
The MKCE spatial operation operator comprises: erase (\), erased (/), intersection (#), union (#), symmetric differential
Then the entity set E of the MKCE model is expressed as TSE — Geo (T, S, E) in a spatio-temporal framework. The incremental set of entities may be expressed as TSE ═ GeoI ((T) ═ GeoI i ，S i ，E i )，(T j ，S j ，E j )). Wherein, T i To start time, T j For the end time, S i And S j Are respectively entity E i And E j The spatial characteristics of (1).
In one embodiment, assume Geo (T) i ，S i ，E i ) As set A, Geo (T) j ，S j ，E i ) As set B, Geo (T) j ，S j ，E j ) To set C, five basic spatial operation algorithms of the MKCE model can be obtained:
Δ + GeoI(T i ，S i ，E i )，(T j ，S j ，E i )＝B-A#(3)
Δ - GeoI((T j ，S j ，E i )，(T i ，S i ，E i ))＝A-B#(4)
Δ ∩ GeoI((T i ，S i ，E i )，(T j ，S j ，E i )＝C-A#(5)
Δ ∪ GeoI(T i ，S i ，E i )，(T j ，S j ，E i ))＝A+B#(6)
scene dimension:
with reference to fig. 3, in the coordinate system of three basic dimensions of time, space and entity, there are many scenarios of urban physical examination, and each scenario contains rich information, including specific knowledge instances and evaluation indexes supported by the knowledge instances, and topic information reflected by the scenarios. The scene dimension also constitutes a composite dimension. For example, the air quality evaluation scene not only relates to the entity corresponding to the air quality and the space-time distribution thereof, but also comprises the air quality evaluation index and the theme information of the urban ecological environment.
The knowledge instance dimension is specific facts and data corresponding to the physical entities of the city, and provides multi-modal information of Text expression (Text), graphic Charts (Graphical Charts), Video and Audio (Video & Audio), Web page links (Web link) and Database links (Database link) related to the entities. An Instance of knowledge (Instance) can be expressed as a collection of these 5 pieces of information, i.e.
I＝(T，GC，VA，W，D)#(8)。
The index dimension is an evaluation measurement standard of urban physical examination and provides an evaluation index and an index calculation method thereof. Indexes in the urban physical examination knowledge base embody two functions. On one hand, the method is an index-associated and evaluated urban physical examination entity and a knowledge example thereof, is used for determining an evaluation object indicated in index connotation, and is convenient for directly calling method operation according to the evaluation object and an index calculation method. For example, the index item 'air quality good day number ratio' is related to the city air quality index entity class and is subdivided into 6 entities of ground ozone, PM10, PM2.5, carbon monoxide, sulfur dioxide and nitrogen dioxide, the air quality index and the corresponding time domain are calculated by calling the method, and the corresponding result of the index is obtained. And on the other hand, the related topic information can help to acquire indexes and knowledge instance contents when the topic positions knowledge resources. For example, the index item 'air quality good day ratio' is related to the city ecological environment of the theme, and the analysis and evaluation text, index data, a space-time statistical chart and other information of the air quality evaluation are linked under the theme. The Index (Index) is expressed as a set of indices (Metrics, M) 1 ) And Methods set (M) 2 ) A set of (i) i
M＝(M 1 ,M 2 )#(9)。
The topic dimension is multi-aspect information carried by the urban physical examination evaluation content, and the front 5 dimensions are represented by topics. The topic information is beneficial to classifying the city physical examination knowledge and is beneficial to retrieval and application of the knowledge. The topic (the) and the topic Connotation (the conversion) of hierarchical clustering and concretion in the urban physical examination knowledge base are expressed together, namely
T＝(T,TC)#(10)。
Referring to fig. 5, in an embodiment, the building of the urban physical examination network model fusing hypergraphs by integrating multiple types of information in the urban physical examination multidimensional knowledge structure model includes:
s31, constructing six relation sets according to the various types of information, wherein the six relation sets comprise: an index-index relationship, a knowledge instance-knowledge instance relationship, a knowledge instance-attribute relationship, an index-attribute relationship, and an index-knowledge instance relationship;
in the embodiment of the invention, the multiple types of information comprise four types of information including indexes, knowledge instances, attributes and scenes.
The indexes are index items of two index systems of a natural resource department and a housing and a city and countryside construction department integrated in the city physical examination knowledge base, and are used for quantitatively evaluating the problems of urban diseases and the like in a city.
The knowledge example is multi-modal information of specific facts and data corresponding to the urban physical examination index association entity and is used for expressing the association relation and the time-space change of the scene where the entity is located from various information forms.
The attribute is data content corresponding to the urban physical examination index. For example, under the theme information of convenient traffic, there are information of public transport, track, shared single car and the like in the prefecture built-up area of the city corresponding to the index item of 'green traffic trip sharing rate'. The public transportation data has attribute information such as 'line code', 'line name', 'vehicle license plate', 'vehicle model code', and the like, and can comprehensively know entity information expressed by the data.
The scene is third-party information corresponding to the urban physical examination indexes. Besides environmental information such as time, space geometry and the like, government-related policy provisions and regulations are regarded as contextual information and reflect the relevance of urban physical examination conditions and government policies.
The embodiment of the present invention defines the following six forms for the relationship of the above four types of information in pairs or groups.
Index term-index term (MM) relationship. This relationship represents the correlation between index items. For example, the sub-index items of the rail transit commute amount are included between the index items of the rail station surrounding coverage commute proportion and the green transit trip sharing rate.
Knowledge example-knowledge example (II) relationship. The relation represents shared connection between specific facts and data corresponding to the associated entities, and one knowledge instance can be associated with a plurality of index items and can be in a many-to-many relation.
Knowledge instance-attribute (IA) relationship. The knowledge examples described by the facts and the data are provided with various attribute descriptions, and the incidence relation between the knowledge examples and the attributes is collected, so that the expression of the rich knowledge examples is facilitated.
Index-attribute (MA) relationship. The index connotation carried by the index item designates a knowledge instance with specific attributes, and the first choice relation between the indexes and the attributes of the knowledge instance is collected as an important relation.
Index-knowledge example (MI) relationship. The knowledge examples of the urban physical examination entity evaluated by the index item mainly reflect the co-occurrence relationship between the index and the knowledge examples in a specific scene. For example, under the theme information of urban ecological environment, the water quality is measured by an index item that the surface water reaches or is better than the III-class water body proportion (%), and the water quality conditions of national assessment objects of Guangzhou lotus mountain and Guangzhou Yagang are associated as co-occurrence knowledge examples.
Index-context-knowledge instance (MCI) relationships. The third-party context information reflects and influences the index evaluation result and knowledge instance situation representation and the co-occurrence relation between the two. If government-related policy provisions and regulations are adjusted, then there must also be a change between the indicators and knowledge instances.
And S32, constructing the urban physical examination network model fusing the hypergraph by taking various types of information as a vertex set of the hypergraph and taking six types of relation sets as a super edge set of the hypergraph.
Fig. 4 is a schematic structural diagram of an urban physical examination network model incorporating a hypergraph according to an embodiment of the present invention.
Referring to fig. 5, in the embodiment of the present invention, four types of information and six relationship sets are constructed by using the hypergraph G ═ (V, E). The vertex set V ═ M ═ I { < C { < A { [ means ] respectively, and represents an index, a knowledge instance, a scenario, an attribute, and a super-set E ═ MM { [ means ] II { [ means ] IA { [ means ] MI { [ means ] MCI { [ means ] respectively, and represents a super-relationship of an index item-index item, a knowledge instance-knowledge instance, a knowledge instance-attribute, an index-knowledge instance, and an index-scenario-knowledge instance.
Referring to fig. 6, in the embodiment of the present invention, a scene entity identification method using a BERT + BiLSTM + CRF model is designed according to the characteristics of entities in the urban physical examination scene network model, which specifically includes:
determining scene types according to the entity characteristics in the urban physical examination scene network model, and designing a manual labeling mode;
the scene type is determined according to the entity set evaluation topic information corresponding to the index items of the urban physical examination. Taking the main theme of urban ecological environment as an example, the main theme has four scenes of water quality, air quality, noise level and green open space, wherein the scene of green open space comprises four scene elements of indexes, knowledge examples, scenes and attributes, and the associated indexes are 'urban ecological corridor standard reaching rate' and 'park green space service radius coverage rate'; the knowledge example comprises a text, a numerical value and a map corresponding to an entity, wherein the entity relates to ecological corridors and park green land types and is subdivided into roads, rivers, parks, greenbelts, forest shade belts and the like, the text relates to introduction of urban green open spaces, functions of the urban ecological corridors and the like, the numerical value comprises the standard reaching rate of the ecological corridors in each area and the radius coverage rate of the park green land service of each area under the size of Guangzhou city area county, and the map comprises an interactive map of the standard reaching rate of the ecological corridors in each area and the radius coverage rate of the park green land service of each area generated on the basis of the numerical value item; the situation has space-time distribution of the occurrence of the knowledge instance and relevant regulations and requirements, such as 'ecological corridor construction standard', 'ecological corridor space definition and construction implementation requirement' and the like.
Determining scene types according to natural languages, wherein the scene types are determined as two types of tasks: scene classification and scene element extraction. Only the operation on the text is described here, and the extraction of the diagrams, maps and data is realized according to the designed scene trigger words.
And (3) carrying out scene trigger word labeling on various knowledge resources in the urban physical examination knowledge base, and adopting a BIOE (Begin, Inside, Other and End) mode. In the scene classification task, the whole texts such as the city physical examination planning text, the policy leather information and the like are marked according to the scene types, and the training unit is divided by the whole texts. And the chart and the map are directly labeled with scene words. In the scene element extraction task, scene elements in the whole text are extracted, and the training unit divides and identifies the scene elements in a single sentence.
Model training is performed by using a deep learning model, such as a BERT + BilSTM + CRF model training procedure shown in FIG. 6.
And training the sequence labels by using a deep learning model. For example, the BERT + BiLSTM + CRF model, consists of a BERT layer, a BiLSTM layer, and a CRF layer. The BERT model converts the word sequence into word vectors, the BilSTM model calculates probability distribution for each label, and the CRF model scores the results after the processing of the BilSTM layer through a transfer matrix to obtain the labels corresponding to the words. And training m epochs, performing F1 value verification on the training result of each epoch, and selecting a precision threshold value of 70% to perform result retention. And finally, selecting an optimal precision result through precision comparison.
In one embodiment, extracting scene elements from the structured data, the semi-structured data and the unstructured data in the physical examination knowledge base according to the entity characteristics in the physical examination network model to obtain meta scene information includes:
the method comprises the steps of designing scene trigger words according to semantic rules, extracting scene elements from structured data, semi-structured data and unstructured data in a city physical examination knowledge base according to entity features in a city physical examination network model based on the scene trigger words to obtain scene sentences, segmenting the scene sentences according to scene types and scene vocabularies, obtaining enhanced scene sentence data through marking and combining, and inputting the enhanced scene sentence data into a deep learning model for training to obtain meta-scene information.
In the embodiment of the invention, scene element extraction is carried out on the sentence-level scene information, and the primary condition is to determine the scene type and the trigger words of the scene. According to semantic rule matching, the embodiment of the invention provides a user-friendly operation design scene trigger word mode which is selected by a scene type and custom-built by a scene model. The scene type selection is to select the divided scene types according to the city physical examination evaluation subject information. For example, under the theme information of urban ecological environment, four scenes including water quality, air quality, noise level and green open space are divided; under the public health theme information, three scenes of public sports construction, community public health and medical resources are divided. The self-defining construction of the scene model is carried out according to indexes, knowledge examples and scenes in the scene. Taking a certain 'public sports construction' scene as an example, the text describes that in 2020, the per-capita community sports field area in Guangzhou city reaches 0.93 square meters per person, and the change is not obvious compared with the change in the past year. Wherein, the sports field area of the people community in the old cities such as Yuxiu, litchi bay, Haizhu, Tianhe and Baiyun is generally lower. The result of the decomposition of the scene network model with the fusion hypergraph is shown in the following table 1.
TABLE 1 Single scene element decomposition
It can be seen from the above example that "past year" cannot clearly define time, and time and space can be customized, for example, "2015 year to 2020", "guangzhou city and each region", and scene information corresponding to time and space is obtained through semantic rule matching. In addition, the scope of scene recognition can be expanded by adding a trigger word, such as "fitness".
The extraction of charts, maps and data is realized according to a hierarchical extraction strategy, and a scene of public sports construction is taken as an example for explanation.
And extracting the chart and the map from coarse to fine according to indexes, knowledge examples and scenes corresponding to the scenes of public sports construction by semantic rule matching and hierarchical retrieval. Firstly, acquiring a statistical chart and a map of 'per-capita community sports field area' of a city and a county scale corresponding to indexes, acquiring the number, the area statistical chart and the distribution map of relevant community sports fields of the city and the county scale corresponding to knowledge examples, and acquiring the number, the area and the three-dimensional space-time distribution map of the community sports fields of a specific plot scale according to scenes.
And extracting data according to the attribute list corresponding to the indexes and the knowledge examples, and extracting the data from the urban physical examination database by semantic rule matching. Wherein, the achievement data directly corresponding to the index is directly extracted and applied; and (3) configuring a data computation flow diagram of a required result in a method library in the urban physical examination knowledge base according to indirect data corresponding to the knowledge example, and calling a computation method to realize computation conversion from the indirect data to the result data by using data field semantics to match data computation elements.
The embodiment of the invention adopts the custom construction of the scene model, so that the deficiency of scene elements can be made up and the scene information can be acquired as required.
Referring to fig. 7, the embodiment of the present invention jointly implements scene element extraction based on a scene trigger word and a deep learning model.
In order to improve the efficiency and accuracy of scene element extraction, the embodiment of the invention jointly extracts the scene element extraction based on the scene trigger words and the deep learning model.
Preferably, two extraction modes are formed through scene type selection and scene model custom construction. One is that a scene vocabulary forms and supplements a scene trigger word bank, a relevant scene new sentence is recognized from a knowledge base by using pattern matching, a new scene element is extracted by using a trigger word and a context rule, and the new scene element is combined with a scene model fused with a hypergraph to be structured; and the other method comprises the steps of segmenting words of a scene sentence according to the scene type and the scene vocabulary, labeling and combining the words by using a labeling tool, expanding and supplementing training data, and training by using a deep learning model to continuously improve the scene element recognition effect.
In one embodiment, the scene fusion processing of the meta-scene information to form a composite scene includes:
and calculating the relevance among the meta-scene information by adopting a similarity algorithm, and carrying out scene fusion on the meta-scene information according to the relevance to form a composite scene.
In the embodiment of the invention, the scene relevance strength is measured by adopting a similarity function. Assume that the association of two meta-scenes e1 and e2 is Sim sum :
Sim sum ＝a×Sims(e 1 ，e 2 )+b×Simt(e 1 ，e 2 )+ c×Simm(e 1 ，e 2 )+d×Simc(e 1 ，e 2 )#(11)。
Wherein Sims (e) 1 ,e 2 ) Semantic similarity, Simt (e) corresponding to MetaScenario entities 1 ,e 2 ) Class similarity, Simm (e), corresponding to meta-scene entities 1 ,e 2 ) Index similarity, Simc (e) corresponding to Meta scene 1 ,e 2 ) For the scene similarity of meta-scenes, a, b, c, d ∈ (0, 1).
Calculating the relevance score, setting two similarity thresholds, and judging the total relevance score Sim sum In which association degree interval, formalizing 3 cases shows:
1) if Sim sum >t 2 Then e is 1 And e 2 Matching;
2) if t 1 <Sim sum <t 2 Then e 1 And e 2 A possible match;
3) if Sim sum <t 1 Then e is 1 And e 2 And not matched.
Wherein t is 1 And t 2 For the lower and upper bounds of the similarity threshold, in one particular embodiment, Sim is set sum >t 2 All meta-scenes of (a) are aggregated.
In one embodiment, the method comprises the steps of constructing a composite scene in a hierarchy mode to generate scene maps of different hierarchies, supplementing and perfecting scene semantics in the scene maps to obtain an urban physical examination knowledge map, and comprises the following steps:
s61, judging whether scene entities in the composite scene are aligned according to a preset scene hierarchy grading strategy, and judging the scene entities which are not aligned as unmatched meta-scenes;
s62, if the unmatched meta-scenes are a single scene, constructing a hierarchical scene spectrogram of the unmatched meta-scenes;
s63, if the unmatched meta-scene is not a single scene, taking the unmatched scene as other scenes, and reconstructing a scene graph after adding the unmatched scene into the new scene;
and S64, supplementing and perfecting scene semantics in the scene map in a scene condensing mode to obtain the city physical examination knowledge map.
For matching meta-scenes, embodiments of the present invention consider that entity alignment has been achieved. For meta-scenes which are possibly matched, the embodiment of the invention verifies and judges whether scene entities are aligned or not according to a scene level grading strategy by utilizing manual experience, namely 'scene level-index level-knowledge instance level', otherwise, the meta-scenes are treated according to unmatched meta-scenes, whether the unmatched meta-scenes are independently classified into one type of scenes or not is judged, otherwise, the unmatched meta-scenes are classified into other scenes, and the new scenes are waited to be added for carrying out re-aggregation. And finally, performing sampling verification to edit and replace the places with errors in a human-computer interaction manner.
Furthermore, the embodiment of the invention summarizes and concurs the aggregated meta-scenes on the vocabulary to form comprehensive semantic descriptions, namely composite scenes. The meta scenes such as "subway trip", "shared bicycle trip", "bus trip", "taxi trip" and the like can be aggregated into a composite scene "public transport".
According to the embodiment of the invention, after the scene maps of different levels are constructed, the scene semantics in the scene maps are supplemented and perfected to obtain the city physical examination knowledge map.
Please refer to fig. 8, which is a schematic structural diagram of a city health knowledge map system according to an embodiment of the present invention.
The embodiment of the invention has the following beneficial effects:
the embodiment of the invention is based on the knowledge representation form and the knowledge structure of the city physical examination scene, the city physical examination multi-dimensional knowledge structure model is constructed according to the city physical examination knowledge base, and the city physical examination information can be uniformly expressed, so that the multi-dimensional knowledge of the city physical examination can be comprehensively expressed; according to the embodiment of the invention, the urban physical examination network model fused with the hypergraph is constructed by integrating various types of information in the urban physical examination multi-dimensional knowledge structure model, and the co-occurrence among scene element nodes can be effectively drawn through the hyperedges connecting the scene element nodes, so that the convenience of carrying out correlation analysis on the urban physical examination scene can be effectively improved; according to the embodiment of the invention, scene element extraction is carried out on the structured data, the semi-structured data and the unstructured data in the urban physical examination knowledge base according to the entity characteristics in the urban physical examination network model, so that the accuracy of scene element extraction can be effectively improved.
Referring to fig. 9, based on the same inventive concept as the above embodiment, an embodiment of the present invention provides an apparatus for constructing an urban physical examination knowledge graph, including:
the city physical examination knowledge base construction module is used for acquiring city physical examination knowledge resources and constructing a city physical examination knowledge base according to the city physical examination knowledge resources; the urban physical examination knowledge resources comprise historical reports, administrative leather-following data and industrial data in the urban physical examination field;
the city physical examination multi-dimensional knowledge structure model building module is used for building a city physical examination multi-dimensional knowledge structure model according to a city physical examination knowledge base based on the knowledge representation form and the knowledge structure of a city physical examination scene;
the city physical examination network model building module is used for building a city physical examination network model fusing hypergraphs by integrating various types of information in the city physical examination multi-dimensional knowledge structure model; wherein the plurality of types of models include: metrics, knowledge instances, attributes, and scenarios;
the scene element extraction module is used for extracting scene elements from the structured data, the semi-structured data and the unstructured data in the urban physical examination knowledge base according to the entity characteristics in the urban physical examination network model to obtain meta scene information;
the scene fusion module is used for carrying out scene fusion processing on the meta-scene information to form a composite scene;
and the knowledge map construction module is used for constructing the composite scene in layers to generate scene maps in different layers, and supplementing and perfecting scene semantics in the scene maps to obtain the urban physical examination knowledge map.
In one embodiment, the city health knowledge base building module 10 is specifically configured to:
acquiring urban physical examination knowledge resources by adopting a crawler tool;
establishing a classification system based on the city physical examination ontology range to obtain a city physical examination initial body library;
and constructing an urban physical examination knowledge base according to the urban physical examination knowledge resources and the urban physical examination initial body base, wherein the urban physical examination knowledge base comprises knowledge instance resources, an index base and a subject base.
In one embodiment, the urban physical examination multi-dimensional knowledge structure model comprises a base dimension and a scene dimension, wherein the base dimension comprises an entity dimension, a time dimension and a space dimension, and the scene dimension comprises a knowledge instance dimension, an index dimension and a subject dimension.
In one embodiment, the city physical examination network model building module 30 is specifically configured to:
constructing six relation sets according to various types of information, wherein the six relation sets comprise: an index-index relationship, a knowledge instance-knowledge instance relationship, a knowledge instance-attribute relationship, an index-attribute relationship, and an index-knowledge instance relationship;
and constructing the urban physical examination network model fusing the hypergraph by taking various types of information as a vertex set of the hypergraph and taking the six types of relation sets as a super edge set of the hypergraph.
In one embodiment, the scene element extraction module 40 is specifically configured to:
designing a scene trigger word according to a semantic rule, extracting scene elements of structured data, semi-structured data and unstructured data in a city physical examination knowledge base according to entity characteristics in a city physical examination network model based on the scene trigger word to obtain a scene sentence, segmenting the scene sentence according to a scene type and scene vocabularies, obtaining enhanced scene sentence subdata through labeling and combination, and inputting the enhanced scene sentence subdata into a deep learning model for training to obtain meta-scene information.
In one embodiment, the scene fusion module 50 is specifically configured to:
and calculating the relevance among the meta-scene information by adopting a similarity algorithm, and performing scene fusion on the meta-scene information according to the relevance to form a composite scene.
In one embodiment, knowledge-graph building module 60 is specifically configured to:
judging whether scene entities in the composite scene are aligned according to a preset scene level grading strategy, and judging the scene entities which are not aligned as unmatched meta-scenes;
if the unmatched meta-scenes are a single scene, constructing a hierarchical scene spectrogram of the unmatched meta-scenes;
if the unmatched meta-scene is not a single scene, taking the unmatched scene as other scenes, and reconstructing a scene map after adding the unmatched scene into the new scene;
scene semantics in the scene map are supplemented and perfected in a scene condensing mode, and the city physical examination knowledge map is obtained.
One embodiment of the present invention provides a computer storage medium, on which a computer program is stored, and the computer program, when executed by a processor, implements the method for constructing the physical examination knowledge graph as described above.
The foregoing is a preferred embodiment of the present invention, and it should be noted that it would be apparent to those skilled in the art that various modifications and enhancements can be made without departing from the principles of the invention, and such modifications and enhancements are also considered to be within the scope of the invention.
Claims (9)
1. A construction method of a city physical examination knowledge graph is characterized by comprising the following steps:
acquiring city physical examination knowledge resources, and constructing a city physical examination knowledge base according to the city physical examination knowledge resources; the urban physical examination knowledge resources comprise historical reports, administrative leather-following data and industrial data in the urban physical examination field;
constructing a multi-dimensional knowledge structure model of the urban physical examination according to the urban physical examination knowledge base based on the knowledge representation form and the knowledge structure of the urban physical examination scene;
integrating various types of information in the urban physical examination multi-dimensional knowledge structure model to construct a urban physical examination network model fusing hypergraphs; the multiple types of models comprise indexes, knowledge instances, attributes and situations;
extracting scene elements of the structured data, the semi-structured data and the unstructured data in the urban physical examination knowledge base according to entity characteristics in the urban physical examination network model to obtain meta scene information;
scene fusion processing is carried out on the meta scene information to form a composite scene;
and constructing the composite scene in layers to generate scene maps in different layers, and supplementing and perfecting scene semantics in the scene maps to obtain the city physical examination knowledge map.
2. The method for constructing a city physical examination knowledge graph as claimed in claim 1, wherein the step of acquiring city physical examination knowledge resources and constructing a city physical examination knowledge base according to the city physical examination knowledge resources comprises:
acquiring urban physical examination knowledge resources by adopting a crawler tool;
establishing a classification system based on the city physical examination ontology range to obtain a city physical examination initial body library;
and constructing an urban physical examination knowledge base according to the urban physical examination knowledge resources and the urban physical examination initial body base, wherein the urban physical examination knowledge base comprises knowledge instance resources, an index base and a subject base.
3. The method for constructing a city physical examination knowledge graph as claimed in claim 1, wherein the city physical examination multi-dimensional knowledge structure model comprises a basic dimension and a scene dimension, the basic dimension comprises an entity dimension, a time dimension and a space dimension, and the scene dimension comprises a knowledge instance dimension, an index dimension and a subject dimension.
4. The method for constructing a city physical examination knowledge graph as claimed in claim 1, wherein the step of constructing a city physical examination network model fusing hypergraphs by integrating various types of information in the city physical examination multi-dimensional knowledge structure model comprises the following steps:
constructing six relation sets according to the various types of information, wherein the six relation sets comprise: an index-index relationship, a knowledge instance-knowledge instance relationship, a knowledge instance-attribute relationship, an index-attribute relationship, and an index-knowledge instance relationship;
and constructing a city physical examination network model fusing the hypergraph by taking the various types of information as a vertex set of the hypergraph and taking the six types of relation sets as a super edge set of the hypergraph.
5. The method for constructing a physical examination knowledge graph of a city according to claim 1, wherein the extracting scene elements from the structured data, the semi-structured data and the unstructured data in the physical examination knowledge base according to the entity features in the physical examination network model to obtain meta scene information comprises:
designing scene trigger words according to semantic rules, extracting scene elements of structured data, semi-structured data and unstructured data in the city physical examination knowledge base according to entity characteristics in the city physical examination network model based on the scene trigger words to obtain scene sentences, segmenting the scene sentences according to scene types and scene vocabularies, obtaining enhanced scene sentence data through marking and combining, and deeply inputting the enhanced scene sentence data into a deep learning model for training to obtain meta-scene information.
6. The method for constructing a city physical examination knowledge graph as claimed in claim 1, wherein the scene fusion processing is performed on the meta-scene information to form a composite scene, comprising:
and calculating the relevance among the meta-scene information by adopting a similarity algorithm, and performing scene fusion on the meta-scene information according to the relevance to form a composite scene.
7. The method for constructing an urban physical examination knowledge graph according to claim 1, wherein the step of constructing the composite scene in a hierarchy manner to generate scene graphs in different hierarchies, and the step of supplementing and improving scene semantics in the scene graphs to obtain the urban physical examination knowledge graph comprises the following steps:
judging whether scene entities in the composite scene are aligned according to a preset scene level grading strategy, and judging the unaligned scene entities as unmatched meta-scenes;
if the unmatched meta-scenes are a single class of scenes, constructing a hierarchical scene spectrogram of the unmatched meta-scenes;
if the unmatched meta-scene is not a single scene, taking the unmatched scene as other scenes, and reconstructing a scene graph after a new scene is added into the unmatched scene;
scene semantics in the scene graph are supplemented and perfected in a scene condensing mode, and the city physical examination knowledge graph is obtained.
8. A construction device of a city physical examination knowledge graph is characterized by comprising:
the city physical examination knowledge base building module is used for acquiring city physical examination knowledge resources and building a city physical examination knowledge base according to the city physical examination knowledge resources; the urban physical examination knowledge resources comprise historical reports, administrative leather-following materials and industrial data materials in the urban physical examination field;
the city physical examination multi-dimensional knowledge structure model building module is used for building a city physical examination multi-dimensional knowledge structure model according to the city physical examination knowledge base based on the knowledge representation form and the knowledge structure of the city physical examination scene;
the city physical examination network model building module is used for building a city physical examination network model fusing hypergraphs by integrating various types of information in the city physical examination multi-dimensional knowledge structure model; wherein the plurality of types of models include: metrics, knowledge instances, attributes, and scenarios;
the scene element extraction module is used for extracting scene elements from the structured data, the semi-structured data and the unstructured data in the urban physical examination knowledge base according to the entity characteristics in the urban physical examination network model to obtain meta scene information;
the scene fusion module is used for carrying out scene fusion processing on the meta-scene information to form a composite scene;
and the knowledge map construction module is used for constructing the composite scene in layers to generate scene maps in different layers, and supplementing and perfecting scene semantics in the scene maps to obtain the urban physical examination knowledge map.
9. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when being executed by a processor, realizes the steps of the method for constructing a city physical examination knowledge graph according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210695086.9A CN115098696A (en) | 2022-06-20 | 2022-06-20 | Method and device for constructing urban physical examination knowledge graph and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210695086.9A CN115098696A (en) | 2022-06-20 | 2022-06-20 | Method and device for constructing urban physical examination knowledge graph and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115098696A true CN115098696A (en) | 2022-09-23 |
Family
ID=83290182
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210695086.9A Pending CN115098696A (en) | 2022-06-20 | 2022-06-20 | Method and device for constructing urban physical examination knowledge graph and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115098696A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115544275A (en) * | 2022-11-29 | 2022-12-30 | 合肥工业大学 | Natural disaster knowledge hypergraph construction method considering time-space process and disaster mechanism |
| CN116028645A (en) * | 2023-01-30 | 2023-04-28 | 正元地理信息集团股份有限公司 | Urban municipal infrastructure emergency knowledge graph determination method, system and equipment |
| CN116089628A (en) * | 2023-02-14 | 2023-05-09 | 成都市城市建设和自然资源档案馆 | City construction and natural resource archive knowledge graph construction method |
| CN116307792A (en) * | 2022-10-12 | 2023-06-23 | 广州市阿尔法软件信息技术有限公司 | Urban physical examination subject scene-oriented evaluation method and device |
| CN118313451A (en) * | 2024-06-11 | 2024-07-09 | 中国科学院沈阳应用生态研究所 | Ecological protection red line risk knowledge graph structure optimization method based on space-time hypergraph |
| CN118505064A (en) * | 2024-07-12 | 2024-08-16 | 中南大学 | Urban physical examination method and system based on knowledge graph |
-
2022
- 2022-06-20 CN CN202210695086.9A patent/CN115098696A/en active Pending
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116307792A (en) * | 2022-10-12 | 2023-06-23 | 广州市阿尔法软件信息技术有限公司 | Urban physical examination subject scene-oriented evaluation method and device |
| CN116307792B (en) * | 2022-10-12 | 2024-03-12 | 广州市阿尔法软件信息技术有限公司 | Urban physical examination subject scene-oriented evaluation method and device |
| CN115544275A (en) * | 2022-11-29 | 2022-12-30 | 合肥工业大学 | Natural disaster knowledge hypergraph construction method considering time-space process and disaster mechanism |
| CN116028645A (en) * | 2023-01-30 | 2023-04-28 | 正元地理信息集团股份有限公司 | Urban municipal infrastructure emergency knowledge graph determination method, system and equipment |
| CN116028645B (en) * | 2023-01-30 | 2024-04-12 | 正元地理信息集团股份有限公司 | Urban municipal infrastructure emergency knowledge graph determination method, system and equipment |
| CN116089628A (en) * | 2023-02-14 | 2023-05-09 | 成都市城市建设和自然资源档案馆 | City construction and natural resource archive knowledge graph construction method |
| CN118313451A (en) * | 2024-06-11 | 2024-07-09 | 中国科学院沈阳应用生态研究所 | Ecological protection red line risk knowledge graph structure optimization method based on space-time hypergraph |
| CN118313451B (en) * | 2024-06-11 | 2024-09-17 | 中国科学院沈阳应用生态研究所 | Ecological protection red line risk knowledge graph structure optimization method based on space-time hypergraph |
| CN118505064A (en) * | 2024-07-12 | 2024-08-16 | 中南大学 | Urban physical examination method and system based on knowledge graph |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115098696A (en) | Method and device for constructing urban physical examination knowledge graph and storage medium | |
| Wu et al. | Mapping the knowledge domain of smart city development to urban sustainability: a scientometric study | |
| CN106919689B (en) | Professional domain knowledge mapping dynamic fixing method based on definitions blocks of knowledge | |
| Niu et al. | Delineating urban functional use from points of interest data with neural network embedding: A case study in Greater London | |
| Mark | Geographic information science: Defining the field | |
| Wu et al. | Analyzing spatial heterogeneity of housing prices using large datasets | |
| CN109684440A (en) | Address method for measuring similarity based on level mark | |
| Kokla et al. | A review of geospatial semantic information modeling and elicitation approaches | |
| CN102508871B (en) | Method for constructing geographic object ontology oriented to disaster event | |
| Simoff et al. | Ontology-based multimedia data mining for design information retrieval | |
| Ahlqvist | In search of classification that supports the dynamics of science: the FAO Land Cover Classification System and proposed modifications | |
| Woźniak et al. | Hex2vec: Context-aware embedding h3 hexagons with openstreetmap tags | |
| CN113157931B (en) | Fusion map construction method and device | |
| Bales et al. | Bibliometric visualization and analysis software: State of the art, workflows, and best practices | |
| CN112948595A (en) | Method, system and equipment for building urban group operation state knowledge graph | |
| Tripathy et al. | An open-source tool to extract natural continuity and hierarchy of urban street networks | |
| CN117875412A (en) | Method for constructing computer education knowledge graph based on knowledge graph | |
| Métral et al. | An ontology-based model for urban planning communication | |
| Zheng et al. | Understanding the city-transport system of urban agglomeration through improved space syntax analysis | |
| CN115495594A (en) | Knowledge graph fusion method and system based on urban public facility decision case | |
| CN113392147B (en) | VR scene knowledge graph representation and dynamic update method | |
| Partyka et al. | Enhanced geographically typed semantic schema matching | |
| Zhañay et al. | A Text Mining Approach to Discover Real-Time Transit Events from Twitter | |
| Guo et al. | Identifying up-to-date urban land-use patterns with visual and semantic features based on multisource geospatial data | |
| CN116304011A (en) | Method, device and storage medium for generating regional industry chain |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |