CN113901238A - City physical examination index knowledge graph construction method and system - Google Patents

City physical examination index knowledge graph construction method and system Download PDF

Info

Publication number
CN113901238A
CN113901238A CN202111482376.7A CN202111482376A CN113901238A CN 113901238 A CN113901238 A CN 113901238A CN 202111482376 A CN202111482376 A CN 202111482376A CN 113901238 A CN113901238 A CN 113901238A
Authority
CN
China
Prior art keywords
index
entity
pair
category
classification
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.)
Granted
Application number
CN202111482376.7A
Other languages
Chinese (zh)
Other versions
CN113901238B (en
Inventor
李颖
陈胜鹏
刘高
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Geospace Information Technology Co ltd
Original Assignee
Wuda Geoinformatics Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Wuda Geoinformatics Co ltd filed Critical Wuda Geoinformatics Co ltd
Priority to CN202111482376.7A priority Critical patent/CN113901238B/en
Publication of CN113901238A publication Critical patent/CN113901238A/en
Application granted granted Critical
Publication of CN113901238B publication Critical patent/CN113901238B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • 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
    • G06F16/35—Clustering; Classification
    • G06F16/353—Clustering; Classification into predefined classes
    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services

Abstract

The invention relates to the field of knowledge maps, and provides a method and a system for constructing a knowledge map of an urban physical examination index, which comprises the following steps: performing first fusion on the knowledge triples to obtain an index entity set, an index category entity set and an index belonging index category relation set; performing second fusion on the index category entity set to obtain a fused index category entity set; establishing an incidence relation between index entities in an index entity set; and constructing the city physical examination index knowledge graph by the index entity set, the fused index category entity set, the incidence relation set among the index entities and the index belonging to the index category relation set. The city physical examination index is stored through the graph structure, so that the city physical examination index retrieval efficiency is improved, index recommendation is facilitated, and city physical examination work is facilitated; by simplifying the association index pair set, the redundant relation among index entities is removed, and the graph database relation searching efficiency is greatly improved.

Description

City physical examination index knowledge graph construction method and system
Technical Field
The invention relates to the field of knowledge maps, in particular to a method and a system for constructing a knowledge map of urban physical examination indexes.
Background
The assessment of urban physical examination (hereinafter referred to as urban physical examination) by national space planning means that characteristics of urban development stages and overall planning implementation effects are periodically analyzed and evaluated in a mode of one-year-one-physical examination and five-year-one-evaluation. The traditional mode of storing the city physical examination resource entries through the relational database is difficult to accurately express the association strength between the indexes and six dimensions of safety, innovation, coordination, green, openness and sharing of city physical examination, and is inconvenient for calculation of the association relationship between the indexes. And when the relational database is used for multi-layer nested connection retrieval, the problems of long time consumption, low performance and the like exist.
The above is only for the purpose of assisting understanding of the technical aspects of the present invention, and does not represent an admission that the above is prior art.
Disclosure of Invention
The invention mainly aims to solve the problems that in the prior art, the relation database stores urban physical examination resource items, the retrieval performance is low, and the relevance among indexes is difficult to mine.
In order to achieve the purpose, the invention provides a method for constructing a knowledge graph of urban physical examination indexes, which comprises the following steps:
s1: acquiring city physical examination index data, extracting knowledge triples in the city physical examination index data, and fusing the knowledge triples for the first time to obtain an index entity set, an index category entity set and an index category relation set belonging to indexes;
s2: performing second fusion on the index category entity set to obtain a fused index category entity set;
step S2 specifically includes:
s21: calculating and obtaining a first classification vector and a second classification vector of the index entity set;
s22: calculating all first vectors and all second vectors of the index category entity set according to the fact that the first classification vector, the second classification vector and the index belong to the index category relation set;
s23: performing second fusion on the index category entity set through all the first vectors and all the second vectors to obtain the fused index category entity set;
s3: establishing an incidence relation between index entities in the index entity set;
s4: and constructing the urban physical examination index knowledge graph through the index entity set, the fused index category entity set, the incidence relation set among the index entities and the index belonging index category relation set.
Preferably, step S21 is specifically:
s211: the attributes of each index entity in the index entity set comprise: the index entity comprises an index name attribute, a numerical value size attribute and a numerical value unit attribute, wherein the index name attribute of the index entity is divided into n =6 categories for labeling;
s212: deep learning training and model tuning are carried out on the index name attribute values of the labeled index entities to obtain a trained index name classification model;
s213: removing the top softmax layer of the trained index name classification model, reasoning the index names of the index entities in the index entity set to obtain a first classification vector
Figure 569383DEST_PATH_IMAGE002
=[xi1, xi2……xin]Wherein i represents the index entity number, xijIndicates the probability that the index entity is of class j, j being [1, n]N represents the total number of classifications;
s214: obtaining the maximum item in the first classification vector, and using the maximum item xijMarking the corresponding classification of j as a first classification, and storing the first classification vector and the first classification into the attribute of the index entity;
s215: setting n to be 23, repeating the steps S211-S214, obtaining the second classification vector and the second classification, and storing the second classification vector and the second classification into the attribute of the index entity.
Preferably, step S22 is specifically:
s221: the index category entity set is marked as M (item), the number of the included index category entities is N (item), a certain index category entity is selected, all the index entities belonging to the index category entity and a first classification vector of the index entity are obtained through the index belonging to the index category relationship;
s222: arranging the first classification vectors into a matrix A according to rows, wherein the expression is as follows:
Figure 516610DEST_PATH_IMAGE004
wherein m represents the total number of index entities, n represents the total number of classifications, xmnRepresenting the probability that the index entity m is of the nth class,
Figure 832185DEST_PATH_IMAGE006
a first classification vector representing a pointer entity m;
s223: record the first vector of the pointer category entity as
Figure 606499DEST_PATH_IMAGE008
The expression is:
Figure 7525DEST_PATH_IMAGE010
wherein k represents a count of the number of the index entity;
s224: iterating the steps S221-S223 for N (item) times to obtain all first vectors of the index category entity set;
s225: selecting a certain index category entity, obtaining the index entity belonging to the index category entity and the second classification vector of the index entity through the index belonging to the index category relationship, and repeating the steps S222-S224 to obtain all the second vectors of the index category entity set.
Preferably, step S23 is specifically:
s231: calculating cosine similarity between the second vectors, and executing
Figure 617498DEST_PATH_IMAGE012
Calculating the sub-cosine similarity to obtain
Figure 787579DEST_PATH_IMAGE013
A second degree of similarity between the individual index category entities;
s232: for the second similarity being greater than a preset threshold k1The two index category entities calculate cosine similarity between first vectors of the two as first similarity, and for the first similarity larger than a preset threshold k2And merging the two index category entities to complete the second fusion.
Preferably, step S3 is specifically:
s31: recording the index entity set as M (indicator), recording the number of the included index entities as N (indicator), and calculating the association strength among the index entities;
s32: acquiring a set M (pair) of association index pairs according to the association strength;
s33: simplifying the association index pair set M (pair)' to obtain a simplified association index pair set M (pair);
s34: and establishing an association relation between the index entities through the simplified association index pair set M (pair).
Preferably, step S31 is specifically:
s311: for two index entities with the numbers of x and y, acquiring index name semantic similarity of the index entity x and the index entity y through a deep learning model;
obtaining a first classification vector and a second classification vector of an index entity x and a first classification vector and a second classification vector of an index entity y;
s312: if the index entity x and the index entity y have the same second classification vector, the index classification similarity is 1; if the second classification vectors of the index entity x and the index entity y are different, but the first classification vectors are the same, the index classification similarity is 0.5; if the first classification vector and the second classification vector of the index entity x and the index entity y are different, the index classification similarity is 0;
s313: the value size attribute value of the index entity x is denoted as val (x), the value size attribute value of the index entity y is denoted as val (y), and the calculation formula of the index value size similarity between x and y is as follows:
Figure 496909DEST_PATH_IMAGE015
s314: establishing a Chinese-English mapping table, converting the numerical unit attribute values of the index entity x and the index entity y into Chinese, marking the numerical units converted into Chinese as fields, and if the numerical units of the index entity x and the index entity y belong to the same field, the index numerical unit similarity is the character similarity between the numerical units converted into Chinese of the index entity x and the index entity y; otherwise, the unit similarity of the index numerical value is 0;
s315: and calculating to obtain the correlation strength, wherein a calculation formula is as follows:
correlation strength S (x, y) = a index name semantic similarity + b index classification similarity + c index numerical value size similarity + d index numerical value unit similarity
Wherein a, b, c and d are preset weights, a + b + c + d =1, and a, b, c, d ∈ (0, 1).
Preferably, step S32 is specifically:
s321: recording a set of related index pairs as M (pair), wherein M (pair) is a set of subsets of a plurality of index entity sets M (indicator), and recording the subset numbers of M (indicator) contained in M (pair) as N (M (pair));
let the index entity number be z, the initial value of z be 1, and mark the z index entity as Mz, Mz belongs to M (indicator);
s322: if the value of z is less than n (indicator), then step S323 is entered, otherwise, a set of associated index pairs m (pair)';
s323: calculating the correlation strength between Mz and M (z +1) -M (N) (indicator), wherein the correlation strength is greater than a preset threshold k3The marker entity of (2) is denoted as M (pair) 'as a set of Mz-related markers'MzWill beCorrelation index set M (pair)'MzStoring into m (pair)'; let z = z +1 and return to step S322.
Preferably, step S33 is embodied as
S331: setting the initial value of the index entity number z as 1; the number of subsets of m (indicator) included in the set m (pair) 'of association indicator pairs is N (m (pair)');
s332: if the value of z is less than or equal to N (m (pair)'), proceeding to step S333, otherwise, outputting a simplified set of association index pairs m (pair);
s333: for the z < th > index entity, acquiring a related index set M (pair) 'from a related index pair set M (pair)'.MzIf M (pair)'MzIf the result is null, let z = z +1 and return to step S332; if M (pair)'MzIf not, go to step S334;
s334: collecting M (pair) 'of related indexes'MzThe number of middle index entities is recorded as N (M (pair)'Mz)；
S335: setting the initial value of the count p to 1;
s336: if p has a value less than N (M (pair)'Mz) Step S337 is entered, otherwise let z = z +1 and return to step S332;
s337: if M (pair)'MzIf the p-th index entity in (b) is Ms, obtaining M (pair) ' from M (pair) ', and 'MsIf M (pair)'MsIf not, the process proceeds to step S338; if M (pair)'MsIf no, let p = p +1 and return to step S336;
s338: if M (pair)'MzThe (p +1) -th to N (M (pair)'s of (1)'Mz) Item entity, present in M (pair)'MsIn (c), M (pair) 'is deleted'MsThe index entity corresponding to (1) and simultaneously updating M (pair) 'in M (pair)'.MsLet p = p +1 and return to step S336.
Preferably, step S34 is specifically:
s341: for the simplified association index pair set M (pair), counting the number of M (indicator) subsets in M (pair) as N (M (pair));
s342: setting the initial value of the index entity number z as 1;
s343: if z is less than or equal to N (m) (pair)), then step S344 is performed, otherwise, the association relationship between the index entities is completed;
s344: obtaining a simplified associated index set M (pair) from M (pair)MzIf M (pair)MzIf not, the step S345 is executed, otherwise, z = z +1 is made and the step S343 is returned to;
s345: insertion of marker entity Mz into M (pair)MzFor updated M (pair)MzThe index entities in (b) sequentially establish an association relationship between two adjacent index entities in sequence, add an association strength attribute to the association relationship, where the value of the association strength attribute is the value of the association strength between two adjacent index entities, make z = z +1, and return to step S343.
A system for constructing a knowledge graph of urban physical examination indexes comprises the following components:
the first fusion module is used for acquiring city physical examination index data, extracting knowledge triples in the city physical examination index data, and performing first fusion on the knowledge triples to acquire an index entity set, an index category entity set and an index category relation set belonging to indexes;
the second fusion module is used for carrying out second fusion on the index category entity set to obtain a fused index category entity set;
the incidence relation building module is used for building incidence relations among the index entities in the index entity set;
and the city physical examination index knowledge graph construction module is used for constructing a city physical examination index knowledge graph through the index entity set, the fused index category entity set, the incidence relation set among the index entities and the index belonging to the index category relation set.
The invention has the following beneficial effects:
1. the city physical examination indexes are stored through the graph structure, so that the city physical examination index retrieval efficiency is improved, index recommendation is facilitated, and city physical examination work is facilitated;
2. the performance and the accuracy of map searching are improved through the second fusion of the index category entity set;
3. by simplifying the association index pair set, the redundant relationship between index entities is removed, and the graph database relationship searching efficiency is greatly improved on the premise of ensuring that the removed relationship can still be retrieved by depending on the searching characteristic of the graph database.
Drawings
FIG. 1 is a flow chart of a method according to an embodiment of the present invention;
FIG. 2 is a simplified structural diagram of the association between index entities;
FIG. 3 is a schematic view of a map structure of the physical indicators of the city;
FIG. 4 is a diagram of a map of knowledge of physical indicators in a city;
FIG. 5 is a system block diagram of an embodiment of the present invention;
the implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, the invention provides a method for constructing a knowledge graph of urban physical examination indexes, which comprises the following steps:
s1: acquiring city physical examination index data, extracting knowledge triples in the city physical examination index data, and fusing the knowledge triples for the first time to obtain an index entity set, an index category entity set and an index category relation set belonging to indexes;
s2: performing second fusion on the index category entity set to obtain a fused index category entity set;
s3: establishing an incidence relation between index entities in the index entity set;
s4: and constructing the urban physical examination index knowledge graph through the index entity set, the fused index category entity set, the incidence relation set among the index entities and the index belonging index category relation set.
In this embodiment, step S1 specifically includes:
s11: the sources of the urban physical examination index data comprise national state and soil survey, annual change survey, special survey of natural resources, general survey of geographical national conditions, statistical yearbook, statistical bulletin, space-time big data and the like; taking methods including but not limited to named entity identification, entity relation identification, keyword extraction and topic analysis for the unstructured data, taking methods including but not limited to wrapper for the semi-structured data, taking methods including but not limited to ETL (data warehouse technology) for the structured data, and extracting knowledge triples from the urban physical examination index data;
s12: the knowledge triplets include: the initial index entity set, the initial index category entity set and the initial index belong to an index category relation set;
the initial set of pointer entities comprises at least the following attributes: the index name attribute, the numerical value size attribute, the numerical value unit attribute and the index unique identification attribute;
the initial index category entity set includes at least the following attributes: the index category is unique identification and index category name;
the initial index belongs to the index category relation set and represents the relation between the index entity and the index category entity;
s13: the index name attribute and the index category name attribute are cleaned and duplicated by adopting a mode including but not limited to semantic analysis and regular matching, then knowledge fusion is carried out on knowledge triples from different sources by adopting a mode including but not limited to data mapping, entity matching and ontology fusion, and finally an index entity set, an index category entity set and an index category relation set are obtained;
in this embodiment, step S2 specifically includes:
s21: calculating and obtaining a first classification vector and a second classification vector of the index entity set;
s22: calculating all first vectors and all second vectors of the index category entity set according to the fact that the first classification vector, the second classification vector and the index belong to the index category relation set;
s23: and performing second fusion on the index category entity set through all the first vectors and all the second vectors to obtain the fused index category entity set.
In this embodiment, step S21 specifically includes:
s211: the attributes of each index entity in the index entity set at least comprise: the index name attribute, the numerical value size attribute and the numerical value unit attribute are used for dividing the index name attribute value of the index entity into n index classification dimensions specified in TD/T1063-2021 for marking, and according to the current specification, n =6, the index name attribute value is respectively safe, innovative, coordinated, green, open and shared;
in a specific implementation, for example, a certain index entity in the index entity set is:
the index unique identification: 1
Index name attribute: number of permanent population
The numerical size attribute: 2189
Numerical Unit Attribute: all people
Index recording time: 2020 to
Index region: beijing City;
s212: deep learning training and model tuning are carried out on the index name attribute values of the labeled index entities to obtain a trained index name classification model; in the concrete implementation, a TextCNN or Bert + Sigmoid model is adopted for training;
s213: removing the top softmax layer of the trained index name classification model, reasoning the index names of the index entities in the index entity set to obtain a first classification vector
Figure 978706DEST_PATH_IMAGE002
=[xi1, xi2……xin]Wherein i represents the index entity number, xijIndicates the probability that the index entity is of class j, j being [1, n]N represents the total number of classifications;
s214: obtaining the maximum item in the first classification vector, and using the maximum item xijThe corresponding classification of j is marked as a first classification, and the first classification vector and the first classification are stored to the fingerIn the attributes of the subject entity;
s215: setting n as n index secondary classifications specified in TD/T1063-2021, repeating the steps S211-S214 according to the current specification, n =23, obtaining the second classification vector and the second classification, and storing the second classification vector and the second classification into the attribute of the index entity.
In this embodiment, step S22 specifically includes:
s221: the index category entity set is marked as M (item), the number of the included index category entities is N (item), a certain index category entity is selected, all the index entities belonging to the index category entity and a first classification vector of the index entity are obtained through the index belonging to the index category relationship;
s222: arranging the first classification vectors into a matrix A according to rows, wherein the expression is as follows:
Figure 297430DEST_PATH_IMAGE004
wherein m represents the total number of index entities, n represents the total number of classifications, xmnRepresenting the probability that the index entity m is of the nth class,
Figure 322017DEST_PATH_IMAGE006
a first classification vector representing a pointer entity m;
s223: record the first vector of the pointer category entity as
Figure 936669DEST_PATH_IMAGE008
The expression is:
Figure 312287DEST_PATH_IMAGE010
wherein k represents a count of the number of the index entity;
s224: iterating the steps S221-S223 for N (item) times to obtain all first vectors of the index category entity set;
s225: selecting a certain index category entity, obtaining the index entity belonging to the index category entity and the second classification vector of the index entity through the index belonging to the index category relationship, and repeating the steps S222-S224 to obtain all the second vectors of the index category entity set.
In this embodiment, step S23 specifically includes:
s231: calculating cosine similarity between the second vectors, and executing
Figure 906473DEST_PATH_IMAGE012
Calculating the sub-cosine similarity to obtain
Figure 175780DEST_PATH_IMAGE013
A second degree of similarity between the individual index category entities;
s232: for the second similarity being greater than a preset threshold k1The two index category entities calculate cosine similarity between first vectors of the two as first similarity, and for the first similarity larger than a preset threshold k2And merging the two index category entities to complete the second fusion.
Referring to fig. 2, in this embodiment, the association relationship between the index entities obtained in step S3 greatly improves the efficiency of searching the graph database relationship on the premise of ensuring that the removed relationship can still be retrieved, depending on the search characteristics of the graph database;
step S3 specifically includes:
s31: recording the index entity set as M (indicator), recording the number of the included index entities as N (indicator), and calculating the association strength among the index entities;
s32: acquiring a set M (pair) of association index pairs according to the association strength;
s33: simplifying the association index pair set M (pair)' to obtain a simplified association index pair set M (pair);
s34: and establishing an association relation between the index entities through the simplified association index pair set M (pair).
In this embodiment, step S31 specifically includes:
s311: for two index entities with the numbers of x and y, acquiring index name semantic similarity of the index entity x and the index entity y through a deep learning model;
obtaining a first classification vector and a second classification vector of an index entity x and a first classification vector and a second classification vector of an index entity y;
s312: if the index entity x and the index entity y have the same second classification vector, the index classification similarity is 1; if the second classification vectors of the index entity x and the index entity y are different, but the first classification vectors are the same, the index classification similarity is 0.5; if the first classification vector and the second classification vector of the index entity x and the index entity y are different, the index classification similarity is 0;
s313: the value size attribute value of the index entity x is denoted as val (x), the value size attribute value of the index entity y is denoted as val (y), and the calculation formula of the index value size similarity between x and y is as follows:
Figure 226913DEST_PATH_IMAGE015
s314: establishing a Chinese-English mapping table, converting the numerical unit attribute values of the index entity x and the index entity y into Chinese, marking the numerical units converted into Chinese as fields, and if the numerical units of the index entity x and the index entity y belong to the same field, the index numerical unit similarity is the character similarity between the numerical units converted into Chinese of the index entity x and the index entity y; otherwise, the unit similarity of the index numerical value is 0;
in specific implementation, for example, a chinese-english mapping table shown in table 1 is established
TABLE 1
Numerical Unit before processing Processed numerical Unit
% Percentage of
cm3 Cubic centimeter
Jin/mu Each mu of jin
Marking the numerical units as fields, wherein the fields comprise length, weight, area, number, amount, ratio and the like, marking the numerical units in a mode of regularly matching keywords in the fields, and respectively marking the fields to which the numerators and the denominators belong for the mixed numerical units;
for example, "kilogram/hectare", the "weight" of the field to which the "kilogram" of the marker molecule belongs and the "area" of the field to which the "hectare" of the denominator belongs;
for the index entity x and the index entity y, if the index entity x and the index entity y belong to the same field (for a mixed numerical unit, the field to which a numerator or a denominator belongs is the same as the field of a numerical unit of another index entity, that is, the index entity x and the index entity y belong to the same field), the similarity of the index numerical unit is the character similarity between the numerical units of Chinese, and the calculation method of the character similarity is not specially limited here; otherwise, it is 0, and the calculation formula is as follows:
Figure 824247DEST_PATH_IMAGE017
s315: and calculating to obtain the correlation strength, wherein a calculation formula is as follows:
correlation strength S (x, y) = a index name semantic similarity + b index classification similarity + c index numerical value size similarity + d index numerical value unit similarity
Wherein a, b, c and d are preset weights, a + b + c + d =1, and a, b, c, d ∈ (0, 1).
In this embodiment, step S32 specifically includes:
s321: recording a set of related index pairs as M (pair), wherein M (pair) is a set of subsets of a plurality of index entity sets M (indicator), and recording the subset numbers of M (indicator) contained in M (pair) as N (M (pair));
let the index entity number be z, the initial value of z be 1, and mark the z index entity as Mz, Mz belongs to M (indicator);
s322: if the value of z is less than n (indicator), then step S323 is entered, otherwise, a set of associated index pairs m (pair)';
s323: calculating the correlation strength between Mz and M (z +1) -M (N) (indicator), wherein the correlation strength is greater than a preset threshold k3The marker entity of (2) is denoted as M (pair) 'as a set of Mz-related markers'MzM (pair) 'is set of related indexes'MzStoring into m (pair)'; let z = z +1 and return to step S322.
In a specific implementation, for example, n (indicator) =5, S321 to S323 are executed, and a correlation index set of Mz (z ∈ [1,5), where z is an integer) is obtained by calculation as follows:
m1 relevance index set M (pair)'M1Comprises the following steps: m2, M3;
m2 relevance index set M (pair)'M2Comprises the following steps: m3;
m3 relevance index set M (pair)'M3Comprises the following steps: m4, M5;
m4 relevance index set M (pair)'M4Comprises the following steps: m5; then the relevance metric pair set m (pair)' is as follows:
Figure 720659DEST_PATH_IMAGE019
in this embodiment, step S33 specifically includes
S331: setting the initial value of the index entity number z as 1; the number of subsets of m (indicator) included in the set m (pair) 'of association indicator pairs is N (m (pair)');
s332: if the value of z is less than or equal to N (m (pair)'), proceeding to step S333, otherwise, outputting a simplified set of association index pairs m (pair);
s333: for the z < th > index entity, acquiring a related index set M (pair) 'from a related index pair set M (pair)'.MzIf M (pair)'MzIf the result is null, let z = z +1 and return to step S332; if M (pair)'MzIf not, go to step S334;
s334: collecting M (pair) 'of related indexes'MzThe number of middle index entities is recorded as N (M (pair)'Mz)；
S335: setting the initial value of the count p to 1;
s336: if p has a value less than N (M (pair)'Mz) Step S337 is entered, otherwise let z = z +1 and return to step S332;
s337: if M (pair)'MzIf the p-th index entity in (b) is Ms, obtaining M (pair) ' from M (pair) ', and 'MsIf M (pair)'MsIf not, the process proceeds to step S338; if M (pair)'MsIf no, let p = p +1 and return to step S336;
s338: if M (pair)'MzThe (p +1) -th to N (M (pair)'s of (1)'Mz) Item entity, present in M (pair)'MsIn (c), M (pair) 'is deleted'MsThe index entity corresponding to (1) and simultaneously updating M (pair) 'in M (pair)'.MsLet p = p +1 and return to step S336.
In specific implementations, for example:
(1) performing S331-S338 for the first time;
z=1；
obtaining M (pair) 'from M (pair)'M1=[M2,M3]，M(pair)’M1Is not empty;
N(M(pair)’M1)=2；
let P = 1;
M(pair)’M1the index entity of item 1 is M2, and M (pair) 'is obtained from M (pair)'.M2=[M3]Is not empty;
M(pair)’M1the 2 nd to 2 nd entities in (a) are M3, M3 is present in M (pair)'M2In (1), M (pair) 'is deleted'M2M3 in (1), and simultaneously updating M (pair)' to
Figure 952795DEST_PATH_IMAGE021
；
(2) Performing S331-S338 a second time;
z = 2; at this time, M (pair) 'is obtained from M (pair)'M2=[]，M(pair)’M2Is an empty set;
(3) performing S331-S338 for a third time;
z=3；
at this time, M (pair) 'is obtained from M (pair)'M3=[M4,M5]，M(pair)’M3Is not empty;
N(M(pair)’M3)=2；
let P = 1;
M(pair)’M3the index entity of item 1 is M4, and M (pair) 'is obtained from M (pair)'.M4=[M5]Is not empty;
M(pair)’M3the 2 nd to 2 nd entities in (a) are M5, M5 is present in M (pair)'M4In (1), M (pair) 'is deleted'M4M5 in (1), and simultaneously updating M (pair)' to
Figure 33884DEST_PATH_IMAGE023
；
(4) Fourth performing S331-S338;
z = 4; at this time, M (pair) 'is obtained from M (pair)'M4=[]，M(pair)’M4Is an empty set;
the set of the output simplified association index pairs is as follows:
Figure 118514DEST_PATH_IMAGE025
。
in this embodiment, step S34 specifically includes:
s341: for the simplified association index pair set M (pair), counting the number of M (indicator) subsets in M (pair) as N (M (pair));
s342: setting the initial value of the index entity number z as 1;
s343: if z is less than or equal to N (m) (pair)), then step S344 is performed, otherwise, the association relationship between the index entities is completed;
s344: obtaining a simplified associated index set M (pair) from M (pair)MzIf M (pair)MzIf not, the step S345 is executed, otherwise, z = z +1 is made and the step S343 is returned to;
s345: insertion of marker entity Mz into M (pair)MzFor updated M (pair)MzThe index entities in (b) sequentially establish an association relationship between two adjacent index entities in sequence, add an association strength attribute to the association relationship, where the value of the association strength attribute is the value of the association strength between two adjacent index entities, make z = z +1, and return to step S343.
In specific implementations, for example:
(1) executing S341-S345 for the first time;
z=1；
obtaining M (pair) from M (pair)M1=[M2,M3]，M(pair)M1Is not empty;
inserting M1 into M (pair)M1At the very front, will M (pair)M1Updated to [ M1, M2, M3]Sequentially establishing association relations between M1 and M2 and between M2 and M3;
(2) performing S341-S345 a second time;
z=2；
obtaining M (pair) from M (pair)M2=[]，M(pair)M2Is an empty set;
(3) executing S341-S345 for the third time;
z=3；
obtaining M (pair) from M (pair)M3=[M4,M5]，M(pair)M3Is not empty;
inserting M3 into M (pair)M3At the very front, will M (pair)M3Updated to [ M3, M4, M5]Sequentially establishing association relations between M3 and M4 and between M4 and M5;
(4) fourth execution of S341-S345;
z=4；
obtaining M (pair) from M (pair)M4=[]，M(pair)M4Is an empty set;
and finishing the establishment of the association relationship among the index entities.
Referring to fig. 3, a schematic diagram of the structure of the urban physical examination indicator knowledge graph constructed in step S4 in this embodiment is shown;
referring to fig. 4, the local map of the example of the urban physical examination index knowledge graph constructed in step S4 of the present embodiment can clearly show the association relationship between the index entities and the affiliation relationship between the index entities and the index category entities;
the invention has the following advantages:
(1) by establishing the urban physical examination index knowledge map and establishing the relationship between index data resources and six dimensions which are safe, innovative, coordinated, green, open and shared and secondary classification thereof, the index search result based on the map is strongly correlated with the six dimensions, the search result is more in line with the regulation of an index system, and the development of urban physical examination work is facilitated.
(2) By establishing the urban physical examination index knowledge graph and establishing the link relation among index data resources, the index relevance is favorably mined, the index exploration and index recommendation are favorably realized, and the working efficiency of the index system construction step in the urban physical examination process is improved.
(3) By removing the redundant relationship between the index entities, the efficiency of searching the graph database relationship is greatly improved on the premise of ensuring that the removed relationship can still be retrieved by depending on the searching characteristic of the graph database.
Referring to fig. 5, the invention provides a system for constructing a knowledge graph of urban physical examination indicators, comprising:
the first fusion module 10 is configured to acquire city physical examination index data, extract a knowledge triplet in the city physical examination index data, and perform first fusion on the knowledge triplet to obtain an index entity set, an index category entity set, and an index category relationship set to which an index belongs;
a second fusing module 20, configured to perform second fusion on the index category entity set to obtain a fused index category entity set;
an association relationship establishing module 30, configured to establish an association relationship between each index entity in the index entity set;
and the city physical examination index knowledge graph building module 40 is configured to build a city physical examination index knowledge graph through the index entity set, the fused index category entity set, the association relationship set between the index entities and the index belonging to the index category relationship set.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, third and the like do not denote any order, but rather the words first, second and the like may be interpreted as indicating any order.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims (10)

1. A city physical examination index knowledge graph construction method is characterized by comprising the following steps:
s1: acquiring city physical examination index data, extracting knowledge triples in the city physical examination index data, and fusing the knowledge triples for the first time to obtain an index entity set, an index category entity set and an index category relation set belonging to indexes;
s2: performing second fusion on the index category entity set to obtain a fused index category entity set;
step S2 specifically includes:
s21: calculating and obtaining a first classification vector and a second classification vector of the index entity set;
s22: calculating all first vectors and all second vectors of the index category entity set according to the fact that the first classification vector, the second classification vector and the index belong to the index category relation set;
s23: performing second fusion on the index category entity set through all the first vectors and all the second vectors to obtain the fused index category entity set;
s3: establishing an incidence relation between index entities in the index entity set;
s4: and constructing the urban physical examination index knowledge graph through the index entity set, the fused index category entity set, the incidence relation set among the index entities and the index belonging index category relation set.
2. The urban physical examination index knowledge graph construction method according to claim 1, wherein the step S21 specifically comprises:
s211: the attributes of each index entity in the index entity set comprise: the index entity comprises an index name attribute, a numerical value size attribute and a numerical value unit attribute, wherein the index name attribute of the index entity is divided into n =6 categories for labeling;
s212: deep learning training and model tuning are carried out on the index name attribute values of the labeled index entities to obtain a trained index name classification model;
s213: removing the top softmax layer of the trained index name classification model, reasoning the index names of the index entities in the index entity set to obtain a first classification vector
Figure 476801DEST_PATH_IMAGE001
=[xi1, xi2……xin]Wherein i represents the index entity number, xijIndicates the probability that the index entity is of class j, j being [1, n]N represents the total number of classifications;
s214: obtaining the maximum item in the first classification vector, and using the maximum item xijMarking the corresponding classification of j as a first classification, and storing the first classification vector and the first classification into the attribute of the index entity;
s215: setting n to be 23, repeating the steps S211-S214, obtaining the second classification vector and the second classification, and storing the second classification vector and the second classification into the attribute of the index entity.
3. The urban physical examination index knowledge graph construction method according to claim 1, wherein the step S22 specifically comprises:
s221: the index category entity set is marked as M (item), the number of the included index category entities is N (item), a certain index category entity is selected, all the index entities belonging to the index category entity and a first classification vector of the index entity are obtained through the index belonging to the index category relationship;
s222: arranging the first classification vectors into a matrix A according to rows, wherein the expression is as follows:
Figure 111920DEST_PATH_IMAGE002
wherein m represents the total number of index entities, n represents the total number of classifications, xmnRepresenting the probability that the index entity m is of the nth class,
Figure 452902DEST_PATH_IMAGE003
a first classification vector representing a pointer entity m;
s223: record the first vector of the pointer category entity as
Figure 56053DEST_PATH_IMAGE004
The expression is:
Figure 607120DEST_PATH_IMAGE005
wherein k represents a count of the number of the index entity;
s224: iterating the steps S221-S223 for N (item) times to obtain all first vectors of the index category entity set;
s225: selecting a certain index category entity, obtaining the index entity belonging to the index category entity and the second classification vector of the index entity through the index belonging to the index category relationship, and repeating the steps S222-S224 to obtain all the second vectors of the index category entity set.
4. The urban physical examination index knowledge graph construction method according to claim 1, wherein the step S23 specifically comprises:
s231: calculating cosine similarity between the second vectors, and executing
Figure 529420DEST_PATH_IMAGE006
Calculating the sub-cosine similarity to obtain
Figure 724909DEST_PATH_IMAGE007
A second degree of similarity between the individual index category entities;
s232: for the second similarity being greater than a preset threshold k1The two index category entities calculate cosine similarity between first vectors of the two as first similarity, and for the first similarity larger than a preset threshold k2And merging the two index category entities to complete the second fusion.
5. The urban physical examination index knowledge graph construction method according to claim 1, wherein the step S3 specifically comprises:
s31: recording the index entity set as M (indicator), recording the number of the included index entities as N (indicator), and calculating the association strength among the index entities;
s32: acquiring a set M (pair) of association index pairs according to the association strength;
s33: simplifying the association index pair set M (pair)' to obtain a simplified association index pair set M (pair);
s34: and establishing an association relation between the index entities through the simplified association index pair set M (pair).
6. The urban physical examination index knowledge graph construction method according to claim 5, wherein the step S31 specifically comprises:
s311: for two index entities with the numbers of x and y, acquiring index name semantic similarity of the index entity x and the index entity y through a deep learning model;
obtaining a first classification vector and a second classification vector of an index entity x and a first classification vector and a second classification vector of an index entity y;
s312: if the index entity x and the index entity y have the same second classification vector, the index classification similarity is 1; if the second classification vectors of the index entity x and the index entity y are different, but the first classification vectors are the same, the index classification similarity is 0.5; if the first classification vector and the second classification vector of the index entity x and the index entity y are different, the index classification similarity is 0;
s313: the value size attribute value of the index entity x is denoted as val (x), the value size attribute value of the index entity y is denoted as val (y), and the calculation formula of the index value size similarity between x and y is as follows:
Figure 826857DEST_PATH_IMAGE008
s314: establishing a Chinese-English mapping table, converting the numerical unit attribute values of the index entity x and the index entity y into Chinese, marking the numerical units converted into Chinese as fields, and if the numerical units of the index entity x and the index entity y belong to the same field, the index numerical unit similarity is the character similarity between the numerical units converted into Chinese of the index entity x and the index entity y; otherwise, the unit similarity of the index numerical value is 0;
s315: and calculating to obtain the correlation strength, wherein a calculation formula is as follows:
correlation strength S (x, y) = a index name semantic similarity + b index classification similarity + c index numerical value size similarity + d index numerical value unit similarity
Wherein a, b, c and d are preset weights, a + b + c + d =1, and a, b, c, d ∈ (0, 1).
7. The urban physical examination index knowledge graph construction method according to claim 5, wherein the step S32 specifically comprises:
s321: recording a set of related index pairs as M (pair), wherein M (pair) is a set of subsets of a plurality of index entity sets M (indicator), and recording the subset numbers of M (indicator) contained in M (pair) as N (M (pair));
let the index entity number be z, the initial value of z be 1, and mark the z index entity as Mz, Mz belongs to M (indicator);
s322: if the value of z is less than n (indicator), then step S323 is entered, otherwise, a set of associated index pairs m (pair)';
s323: calculating the correlation strength between Mz and M (z +1) -M (N) (indicator), wherein the correlation strength is greater than a preset threshold k3The marker entity of (2) is denoted as M (pair) 'as a set of Mz-related markers'MzM (pair) 'is set of related indexes'MzStoring into m (pair)'; let z = z +1 and return to step S322.
8. The method for constructing a city physical examination index knowledge graph according to claim 5, wherein the step S33 is specifically to
S331: setting the initial value of the index entity number z as 1; the number of subsets of m (indicator) included in the set m (pair) 'of association indicator pairs is N (m (pair)');
s332: if the value of z is less than or equal to N (m (pair)'), proceeding to step S333, otherwise, outputting a simplified set of association index pairs m (pair);
s333: for the z-th index entity, obtaining the associated index set M (pa) from the associated index pair set M (pair)ir)’MzIf M (pair)'MzIf the result is null, let z = z +1 and return to step S332; if M (pair)'MzIf not, go to step S334;
s334: collecting M (pair) 'of related indexes'MzThe number of middle index entities is recorded as N (M (pair)'Mz)；
S335: setting the initial value of the count p to 1;
s336: if p has a value less than N (M (pair)'Mz) Step S337 is entered, otherwise let z = z +1 and return to step S332;
s337: if M (pair)'MzIf the p-th index entity in (b) is Ms, obtaining M (pair) ' from M (pair) ', and 'MsIf M (pair)'MsIf not, the process proceeds to step S338; if M (pair)'MsIf no, let p = p +1 and return to step S336;
s338: if M (pair)'MzThe (p +1) -th to N (M (pair)'s of (1)'Mz) Item entity, present in M (pair)'MsIn (c), M (pair) 'is deleted'MsThe index entity corresponding to (1) and simultaneously updating M (pair) 'in M (pair)'.MsLet p = p +1 and return to step S336.
9. The urban physical examination index knowledge graph construction method according to claim 5, wherein the step S34 specifically comprises:
s341: for the simplified association index pair set M (pair), counting the number of M (indicator) subsets in M (pair) as N (M (pair));
s342: setting the initial value of the index entity number z as 1;
s343: if z is less than or equal to N (m) (pair)), then step S344 is performed, otherwise, the association relationship between the index entities is completed;
s344: obtaining a simplified associated index set M (pair) from M (pair)MzIf M (pair)MzIf not, the step S345 is executed, otherwise, z = z +1 is made and the step S343 is returned to;
s345: insertion of marker entity Mz into M (pair)MzFor updated M (pair)MzEach index entity in (1) is in sequenceEstablishing an association relationship between two adjacent index entities, adding an association strength attribute to the association relationship, wherein the value of the association strength attribute is the value of the association strength between the two adjacent index entities, letting z = z +1, and returning to step S343.
10. A system for constructing a knowledge graph of urban physical examination indexes is characterized by comprising the following steps:
the first fusion module is used for acquiring city physical examination index data, extracting knowledge triples in the city physical examination index data, and performing first fusion on the knowledge triples to acquire an index entity set, an index category entity set and an index category relation set belonging to indexes;
the second fusion module is used for carrying out second fusion on the index category entity set to obtain a fused index category entity set;
the incidence relation building module is used for building incidence relations among the index entities in the index entity set;
and the city physical examination index knowledge graph construction module is used for constructing a city physical examination index knowledge graph through the index entity set, the fused index category entity set, the incidence relation set among the index entities and the index belonging to the index category relation set.
CN202111482376.7A 2021-12-07 2021-12-07 City physical examination index knowledge graph construction method and system Active CN113901238B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111482376.7A CN113901238B (en) 2021-12-07 2021-12-07 City physical examination index knowledge graph construction method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111482376.7A CN113901238B (en) 2021-12-07 2021-12-07 City physical examination index knowledge graph construction method and system

Publications (2)

Publication Number Publication Date
CN113901238A true CN113901238A (en) 2022-01-07
CN113901238B CN113901238B (en) 2022-02-18

Family

ID=79025603

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111482376.7A Active CN113901238B (en) 2021-12-07 2021-12-07 City physical examination index knowledge graph construction method and system

Country Status (1)

Country Link
CN (1) CN113901238B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115099722A (en) * 2022-08-24 2022-09-23 自然资源部第三航测遥感院 Knowledge pedigree-based homeland space planning index model management and application method
CN115186111A (en) * 2022-09-13 2022-10-14 中国医学科学院医学信息研究所 Index data semantic association and fusion method, system and storage medium

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180159876A1 (en) * 2016-12-05 2018-06-07 International Business Machines Corporation Consolidating structured and unstructured security and threat intelligence with knowledge graphs
CN110347843A (en) * 2019-07-10 2019-10-18 陕西师范大学 A kind of Chinese tour field Knowledge Service Platform construction method of knowledge based map
CN111026883A (en) * 2019-12-11 2020-04-17 南方电网数字电网研究院有限公司 Knowledge graph construction method, device, equipment and medium of power business data
CN112948595A (en) * 2021-03-30 2021-06-11 中国科学院自动化研究所 Method, system and equipment for building urban group operation state knowledge graph
CN113065003A (en) * 2021-04-22 2021-07-02 国际关系学院 Knowledge graph generation method based on multiple indexes
CN113535986A (en) * 2021-09-02 2021-10-22 中国医学科学院医学信息研究所 Data fusion method and device applied to medical knowledge graph

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180159876A1 (en) * 2016-12-05 2018-06-07 International Business Machines Corporation Consolidating structured and unstructured security and threat intelligence with knowledge graphs
CN110347843A (en) * 2019-07-10 2019-10-18 陕西师范大学 A kind of Chinese tour field Knowledge Service Platform construction method of knowledge based map
CN111026883A (en) * 2019-12-11 2020-04-17 南方电网数字电网研究院有限公司 Knowledge graph construction method, device, equipment and medium of power business data
CN112948595A (en) * 2021-03-30 2021-06-11 中国科学院自动化研究所 Method, system and equipment for building urban group operation state knowledge graph
CN113065003A (en) * 2021-04-22 2021-07-02 国际关系学院 Knowledge graph generation method based on multiple indexes
CN113535986A (en) * 2021-09-02 2021-10-22 中国医学科学院医学信息研究所 Data fusion method and device applied to medical knowledge graph

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
张鹏飞等: "面向绿色标准的知识图谱构建方法的应用研究", 《标准科学》 *
李智星等: "电网企业运营指标知识图谱构建及其应用", 《农村电气化》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115099722A (en) * 2022-08-24 2022-09-23 自然资源部第三航测遥感院 Knowledge pedigree-based homeland space planning index model management and application method
CN115099722B (en) * 2022-08-24 2022-11-18 自然资源部第三航测遥感院 Knowledge pedigree-based homeland space planning index model management and application method
CN115186111A (en) * 2022-09-13 2022-10-14 中国医学科学院医学信息研究所 Index data semantic association and fusion method, system and storage medium

Also Published As

Publication number Publication date
CN113901238B (en) 2022-02-18

Similar Documents

Publication Publication Date Title
CN113901238B (en) City physical examination index knowledge graph construction method and system
CN106777274B (en) A kind of Chinese tour field knowledge mapping construction method and system
Martınez et al. Scalable learning of Bayesian network classifiers
Gustafsson et al. Comparison and validation of community structures in complex networks
CN110990590A (en) Dynamic financial knowledge map construction method based on reinforcement learning and transfer learning
CN109635083B (en) Document retrieval method for searching topic type query in TED (tele) lecture
CN106951498A (en) Text clustering method
CN112380325B (en) Knowledge graph question-answering system based on joint knowledge embedded model and fact memory network
CN112527938A (en) Chinese POI matching method based on natural language understanding
CN104881689A (en) Method and system for multi-label active learning classification
CN106991127A (en) A kind of knowledget opic short text hierarchy classification method extended based on topological characteristic
CN109271486A (en) A kind of similitude reservation cross-module state Hash search method
CN115599899B (en) Intelligent question-answering method, system, equipment and medium based on aircraft knowledge graph
CN112988917A (en) Entity alignment method based on multiple entity contexts
Sugiharti et al. Predictive evaluation of performance of computer science students of unnes using data mining based on naÏve bayes classifier (NBC) algorithm
Elayidom et al. A generalized data mining framework for placement chance prediction problems
CN116775661A (en) Big space data storage and management method based on Beidou grid technology
Luo et al. Exploring destination image through online reviews: an augmented mining model using latent Dirichlet allocation combined with probabilistic hesitant fuzzy algorithm
CN114048305A (en) Plan recommendation method for administrative penalty documents based on graph convolution neural network
Galbrun Methods for redescription mining
CN109033746B (en) Protein compound identification method based on node vector
CN109460467B (en) Method for constructing network information classification system
Guesmi et al. Community detection in multi-relational social networks based on relational concept analysis
Luo et al. Chinese address standardisation of POIs based on GRU and spatial correlation and applied in multi-source emergency events fusion
CN113553402B (en) Automatic question-answering method for reading and understanding examination based on graph neural network

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
GR01 Patent grant
GR01 Patent grant
CP01 Change in the name or title of a patent holder

Address after: 430000 Wuda science and Technology Park, Jiangxia Avenue, Miaoshan District, Donghu Development Zone, Wuhan City, Hubei Province

Patentee after: Geospace Information Technology Co.,Ltd.

Address before: 430000 Wuda science and Technology Park, Jiangxia Avenue, Miaoshan District, Donghu Development Zone, Wuhan City, Hubei Province

Patentee before: WUDA GEOINFORMATICS Co.,Ltd.

CP01 Change in the name or title of a patent holder