CN111597349A - Rail transit standard entity relation automatic completion method based on artificial intelligence - Google Patents

Rail transit standard entity relation automatic completion method based on artificial intelligence Download PDF

Info

Publication number
CN111597349A
CN111597349A CN202010363261.5A CN202010363261A CN111597349A CN 111597349 A CN111597349 A CN 111597349A CN 202010363261 A CN202010363261 A CN 202010363261A CN 111597349 A CN111597349 A CN 111597349A
Authority
CN
China
Prior art keywords
entity
sentence
rail transit
entity relationship
relationship
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
CN202010363261.5A
Other languages
Chinese (zh)
Other versions
CN111597349B (en
Inventor
朱磊
冯林林
黑新宏
刘尧林
吕泓瑾
张晋源
林泓
刘瑞
刘旭华
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.)
Xian University of Technology
Original Assignee
Xian University of Technology
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 Xian University of Technology filed Critical Xian University of Technology
Priority to CN202010363261.5A priority Critical patent/CN111597349B/en
Publication of CN111597349A publication Critical patent/CN111597349A/en
Application granted granted Critical
Publication of CN111597349B publication Critical patent/CN111597349B/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
    • G06F16/33—Querying
    • G06F16/3331—Query processing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/237—Lexical tools
    • G06F40/242—Dictionaries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/279—Recognition of textual entities

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Machine Translation (AREA)

Abstract

The invention discloses an automatic track traffic standard entity relationship completion method based on artificial intelligence, which comprises the steps of constructing an entity relationship completion model, inputting track traffic standards and noun part-of-speech participles into the entity relationship completion model, judging whether the input standards are simple sentences, searching entity related attributes in the track traffic standards if the input standards are the simple sentences, generating entity relationship triplets, and extracting track traffic standard back sentence attribute words and entities if the input standards are not the simple sentences to ensure that the front sentence entities and the back sentence attribute words are n: n, matching or judging whether the grammar of the front sentence is a principal object and the grammar of the rear sentence is an object supplement, if so, directly matching the entity of the front sentence with the object, directly matching the keyword of the rear sentence with the entity of the object, generating an entity relationship triple, if not, outputting the entity and entity relationship of which the vocabulary relevancy exceeds a threshold value, generating an entity relationship triple, obtaining a complete semantic structure entity specification, and finishing the automatic completion of the track traffic specification entity relationship.

Description

Rail transit standard entity relation automatic completion method based on artificial intelligence
Technical Field
The invention belongs to the technical field of artificial intelligence natural language processing, and relates to an automatic track traffic standard entity relationship completion method based on artificial intelligence.
Background
The knowledge graph is a semantic knowledge base, and the knowledge graph stores knowledge by adopting triples. The knowledge graph can promote a computer to better understand natural language, provide better service for people, and realize natural switching of human and machine, and entity identification and relation extraction are important processes for constructing the knowledge graph.
Because Chinese grammar is complex, sentence structure is disordered, no strict limitation exists, entity components in part of standard documents are missing, and description texts can be generated by meeting basic semantics. These reasons have led to large differences in writing of specifications in various professional areas. Therefore, when a knowledge graph in the professional field is constructed, although the entities can be extracted from the specification entries, the relationship between the entities cannot be judged due to the missing entity components.
In recent years, the rail transit of China is rapidly developed, so that the production and living modes of people are deeply influenced, and the development of social productivity is greatly promoted. The rail transit design standard is a main basis for rail transit design, construction, detection and maintenance, and provides clear requirements and explanation for each part of rail transit design. Therefore, the entities and the relations in the specification are extracted to construct a knowledge graph, and then the knowledge graph is manufactured into the design, detection and construction of the subway by combining the technologies of retrieval, reasoning, query and the like of the knowledge graph. And a plurality of constraints of each single object can be inquired on a website or an application program generated by the correlation of the knowledge graph according to keyword search. The construction personnel can design and construct according to the query result when constructing each independent object in the rail transit, and the detection personnel can directly compare according to the displayed constraint when detecting whether the project meets the standard. Therefore, the construction personnel can more conveniently design and detect whether the subway design meets the standard or not.
The missing entity relation components are supplemented, and the method is a very important basis in the construction work of the knowledge graph. The missing entity relationship directly affects the information extraction, and further affects the structure of nodes and edges in the knowledge graph, so that the reasoning performance of the knowledge graph is greatly reduced. The automatic map construction of the rail transit standard and the later inquiry, reasoning and intelligent question and answer can be carried out by complementing the entity relationship.
Because the natural language processing for map construction and specification of each domain is in the bud stage, the existing probability model and dictionary-based method lack a large amount of accurate labeled documents. However, these problems require professional practitioners to perform a large amount of analysis and processing, and perform manual entity relationship completion, which is time-consuming and labor-consuming; and the design specification relates to more than 30 industries, so that personnel can only carry out completion treatment from the prior experience of the personnel, and the problem of low accuracy of the completion entity relationship is caused.
Disclosure of Invention
The invention aims to provide an artificial intelligence-based automatic completion method for a rail transit standard entity relationship, which solves the problems that the conventional method for completing the rail transit standard entity relationship can only be manually performed, is time-consuming and labor-consuming and has low accuracy.
The technical scheme adopted by the invention is that an artificial intelligence-based automatic track traffic regulation entity relationship completion method comprises the following steps:
step 1: constructing an entity relation completion model according to the rail transit standard;
step 2: performing part-of-speech tagging on the rail transit specification, and extracting noun part-of-speech participles in the rail transit specification;
and step 3: inputting all rail transit specifications and extracted noun part-of-speech participles into an entity relationship completion model, wherein the extracted noun part-of-speech participles serve as entities to be completed; judging whether each input rail transit standard is a simple sentence or not by using a symbol detection method, if so, performing a step 4, and if not, performing a step 5;
and 4, step 4: searching entity related attributes in the rail transit specification, judging the relationship between the entity and the entity, generating entity relationship triples and storing the entity relationship triples;
and 5: performing dependency syntax analysis based on deep learning on the rail transit specification, if the former sentence is a Noun Phrase (Noun Phrase, NP) with a parallel structure, performing step 6, and if not, performing step 7;
step 6: extracting the attribute words and entities of the back sentence, matching the attribute words n of the front sentence with the attribute words n of the back sentence, generating entity relation triples and storing the entity relation triples;
and 7: judging whether the grammar of the front sentence is a main predicate object and the grammar of the rear sentence is an object supplement, if so, performing a step 8, and if not, performing a step 9;
and 8: directly matching the front sentence entity with the object, and directly matching the rear sentence keyword with the object entity to generate an entity relationship triple and store the entity relationship triple;
and step 9: calculating the vocabulary relevancy, outputting the entity-entity relationship with the relevancy exceeding a threshold value, generating an entity-relationship triple, and storing the entity-relationship triple;
step 10: and (4) outputting the entity relationship triples generated in the steps 4, 8 and 9 to obtain a complete semantic structure entity specification, namely completing the automatic completion of the track traffic specification entity relationship.
The present invention is also technically characterized in that,
the entity-relationship triplets are "entity-verb-entity" or "entity-degree-attribute".
The specific operation steps of step 1 are as follows:
step 1.1: acquiring rail transit standard textual data, preprocessing and training the acquired rail transit standard textual data, and generating a dictionary;
step 1.2: processing the dictionary, mining the missing features, and extracting entity completion rules and methods;
step 1.3: and constructing an entity relation completion model by adopting the extracted entity completion rule and method.
In the step 2, a Bi-LSTM + CRF part-of-speech tagging model is adopted to tag the part-of-speech of the rail transit standard, adjective part-of-speech participles are divided into attribute words, and verb part-of-speech participles are used for judging the relation between the entity and the entity.
In step 3, the simple sentence is a sentence containing only a pause and a period.
The specific operation steps of step 4 are as follows:
step 4.1: searching attributes and action relations related to the entities in the rail transit specification;
step 4.2: extracting verbs, judging the relation between the entities, and analyzing the part of speech to extract attributes;
step 4.4: and generating an entity relationship triple according to the entity-entity relationship and the extracted attributes, and storing the entity relationship.
The specific operation steps of step 9 are as follows:
step 9.1: judging whether the front sentence is in a non-parallel structure or not and the rear sentence has a substitute word or not according to the rail transit standard, if so, calculating the vocabulary relevancy between the attribute word behind the substitute word and all entities of the front sentence, otherwise, naming the entity recognition phrase and calculating the vocabulary relevancy between all the participles;
step 9.2: and outputting the entity and entity relationship with the vocabulary relevancy exceeding the threshold value, generating an entity relationship triple and storing the entity relationship triple.
In step 9, the vocabulary relevancy is calculated by using a vocabulary relevancy calculation algorithm based on the word bank of the known network.
The vocabulary relevancy calculation algorithm based on the word bank of the known network is as follows:
rel(w1，w2)＝max{a1*sim(s1，s2)+(1-a1)*asso(s1，s2)}
Figure BDA0002475839530000051
ssso(s1，s2)＝∑ri*asso(p1，p2)
in the above formula, rel (w)1，w2) The expression vocabulary w1And the word w2Correlation of (c), sim(s)1，s2) The expression vocabulary w1And the word w2Similarity of (a), asso(s)1，s2) Representing an entity s1And s2The semantic relevance of (2); alpha1Expressing adjustable parameters for linearly harmonizing the similarity and the semantic association degree, wherein the value range of the adjustable parameters is [0,1 ]]；sliThe expression vocabulary w1N denotes the vocabulary w1Having n sense items; s2jThe expression vocabulary w2Is 1, m represents the vocabulary w2Having m sense items; gamma rayiThe semantic relevance coefficient representing different parts of the entity concept is a fit to each part of the two concepts, and must satisfy ∑ ri＝1，p1Is an item of sense s1Of (2) a sense atom, p2Is an item of sense s2The sense of (1).
The method has the advantages that the rail transit normative relation is obtained through deep learning, then the relation among entity type nouns is supplemented according to the semantic method of the known network, the relation completion in the construction process of the knowledge map is completed, the workload of manually constructing the map and the relation database is greatly reduced, the accuracy of the rail transit normative entity relation completion is improved, the structural accuracy of the rail transit normative knowledge map is improved, and a foundation is laid for intelligent rail transit query, reasoning and question and answer based on the knowledge map;
through deep learning, the recognition degree of the entities and the relations thereof is improved, and the semantic similarity of the entity words in the knowledge network is adopted for automatic judgment, so that the supplement of the entity relations is realized, and a insist foundation is provided for constructing a knowledge graph.
Drawings
FIG. 1 is a flow chart of an automatic completion method for track traffic regulation entity relationship based on artificial intelligence;
FIG. 2 is a schematic diagram of a process of performing part-of-speech tagging on a rail transit specification by using a BI + LSTM + CRF part-of-speech tagging model in the automatic completion method for a rail transit specification entity relationship according to the present invention;
FIG. 3 is a schematic diagram of a process of calculating vocabulary relevancy by using a vocabulary relevancy calculation algorithm based on a known network lexicon in the rail transit normative entity relationship automatic completion method of the present invention.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
The invention discloses an artificial intelligence-based automatic completion method for a track traffic regulation entity relationship, which comprises the following steps of:
step 1: building entity relation completion model according to rail transit standard
Step 1.1: acquiring rail transit standard original text data from urban rail transit technical specifications, carrying out format check on the acquired original text data, deleting unnecessary information such as blank spaces and the like, acquiring preprocessed data, and then training the preprocessed data to generate a dictionary;
step 1.2: processing data in the dictionary, mining the missing features, and extracting entity completion rules and methods;
step 1.3: and constructing an entity relation completion model by adopting the extracted entity completion rule and method.
Step 2: referring to FIG. 2, a Bi-LSTM + CRF part-of-speech tagging model is adopted to tag the part of speech of the rail transit standard, data in a dictionary is converted into a format of hdf5, then the data are input into the Bi-LSTM + CRF part-of-speech tagging model, extracted noun part-of-speech participles in the rail transit standard are output, the adjective part-of-speech participles are divided into attribute words, and verb part-of-speech participles are used for judging the relation between an entity and the entity.
The specific process of performing part-of-speech tagging on the rail transit specification by adopting the Bi-LSTM + CRF part-of-speech tagging model is shown as the following table:
Figure BDA0002475839530000071
and step 3: inputting all rail transit standard texts and the extracted noun part-of-speech participles into an entity relationship completion model, wherein the extracted noun part-of-speech participles serve as entities to be completed; then, judging whether each input rail traffic specification is a simple sentence or not by using a symbol detection method, wherein the simple sentence is a sentence only containing a pause sign and a period sign and comprises a subject, a predicate or an object;
if the simple sentence is the simple sentence, performing the step 4, if the simple sentence is not the simple sentence, performing the step 5, wherein the complex sentence usually comprises a plurality of subjects, a plurality of predicates and a plurality of objects;
and 4, step 4: searching entity related attributes in the rail transit specification, judging the relationship between the entity and the entity, generating entity relationship triples and storing the entity relationship triples; the method specifically comprises the following operation steps:
step 4.1: searching attributes and action relations related to the entities in the rail transit specification;
step 4.2: extracting verbs, judging the relation between the entities, and analyzing the part of speech to extract attributes;
step 4.4: and generating an entity relationship triple according to the entity-entity relationship and the extracted attributes, and storing the entity relationship.
And 5: performing dependency syntax analysis based on deep learning on the rail transit standard, if the previous sentence is an NP phrase with a parallel structure, performing step 6, and if not, performing step 7;
step 6: extracting the attribute words and entities of the later sentence to ensure that the attribute words n of the earlier sentence and the attribute words n of the later sentence are: n, matching, generating entity relation triples, namely 'entity-verb-entity' or 'entity-degree-attribute', and storing;
and 7: judging whether the grammar of the front sentence is a main predicate object and the grammar of the rear sentence is an object supplement, if so, performing a step 8, and if not, performing a step 9;
and 8: directly matching the former sentence entity with the object, and directly matching the latter sentence keyword with the object entity to generate an entity relationship triple, namely 'entity-verb-entity' or 'entity-degree-attribute', and storing the entity relationship triple;
and step 9: referring to fig. 3, the vocabulary relevancy is calculated by using a vocabulary relevancy calculation algorithm based on the known network lexicon, the entity and entity relationship with the relevancy exceeding a threshold value is output, and an entity relationship triple, namely an entity-verb-entity or an entity-degree-attribute, is generated and stored;
the specific operation steps of step 9 are as follows:
step 9.1: judging whether the front sentence is in a non-parallel structure or not and the rear sentence has a substitute word or not, if so, calculating the vocabulary relevancy between the attribute word after the substitute word and all entities of the front sentence, if not, naming the entity recognition phrase, then calculating the vocabulary relevancy between all the participles, and calculating the relevancy between all the participles;
the vocabulary relevancy calculation algorithm based on the word bank of the known network is as follows:
Figure BDA0002475839530000091
Figure BDA0002475839530000101
the book of understanding adopts the minimal unit that the sememe is the most basic and is not suitable for being divided, adopts 1618 sememes in total, and describes 62174 conceptual entities. In "Zhi Li", an entity-class-semantic-primitive phrase w is set1And w2If w is1There are different concepts (semantic terms): s11，s12，……,s1n，w2There are different concepts (semantic terms): s21，s22，……,s2mThe vocabulary relevancy calculation algorithm based on the word bank of the known network is as follows:
rel(wl，w2)＝max{a1*sim(s1，s2)-(1-a1)*asso(s1，s2)}
Figure BDA0002475839530000102
asso(s1，s2)＝Σri*asso(p1，p2)
in the above formula, rel (w)1，w2) The expression vocabulary w1And the word w2Correlation of (c), sim(s)1,s2) The expression vocabulary w1And the word w2Similarity of (a), asso(s)1,s2) Representing an entity s1And s2The semantic relevance of (2); alpha1Expressing adjustable parameters for linearly harmonizing the similarity and the semantic association degree, wherein the value range of the adjustable parameters is [0,1 ]]；s1iThe expression vocabulary w1Meaning term of (i) ═ 1, …, n denotes the word w1Having n sense items; s2jThe expression vocabulary w2J is 1, …, m represents the word w2Having m sense items; gamma rayiThe semantic relevance coefficient representing different parts of the entity concept is a fit to each part of the two concepts, and must satisfy ∑ ri＝1，p1Is an item of sense s1Of (2) a sense atom, p2Is an item of sense s2The sense of (1).
According to the three formulas, the semantic similarity between the entities of the two words is calculated. If the similarity of the entities is higher, the correlation degree between the entities is higher; the greater the degree of association between the entity senses of two words, the greater their similarity. And then, calculating after linear combination adjustment is carried out on the similarity and the relevance to obtain the final semantic similarity.
Step 9.2: and outputting the entity and entity relationship with the vocabulary relevancy exceeding the threshold value, generating an entity relationship triple, namely 'entity-verb-entity' or 'entity-degree-attribute', and storing. Wherein the threshold value is determined by preliminary experiments in the rail transit specification entry.
Step 10: and (4) outputting the entity relationship triples generated in the steps 4, 8 and 9 to obtain a complete semantic structure entity specification, namely completing the automatic completion of the track traffic specification entity relationship.
For example: the running speed of the train on the plane curve is calculated according to the radius of the curve, and the unbalanced transverse acceleration of the train is not suitable to exceed 0.4m/s2. Firstly, the statement structure is analyzed, and according to the rule, the attribute that the radius is a plane curve can be obtained, and the running speed and the balance transverse acceleration are both the attributes of a train. And for another example, the performance and the main size of the bogie are matched with those of a vehicle body and a line, and related parts are ensured to be within an allowable abrasion limit, so that the train can safely and smoothly run at the highest allowable speed. Firstly, the complex sentences are divided, the first sentence "performance" and "size" of the complex sentences are parallel attributes, the "bogie" is the entity of the complex sentences, the "performance" and the "size" can be distinguished as the attribute words of the "bogie" according to the word correlation by matching, the "train" and the "speed" are in attribute relation, and the "running" is in action relation.
The automatic completion method for the track traffic regulation entity relationship reasonably and effectively solves the problem of entity relationship error completion caused by the fact that no clear semantics exist among entities, attributes and relationships when the entities are in a missing relationship. After the model processing according to the invention is carried out, the semantic relation of the missing of the completion entity can be improved, and the relevance among the entities and the accuracy of the entity relation can be improved.

Claims (9)

1. An automatic track traffic regulation entity relationship completion method based on artificial intelligence is characterized by comprising the following steps:
step 1: constructing an entity relation completion model according to the rail transit standard;
step 2: performing part-of-speech tagging on the rail transit specification, and extracting noun part-of-speech participles in the rail transit specification;
and step 3: inputting all rail transit specifications and extracted noun part-of-speech participles into an entity relationship completion model, wherein the extracted noun part-of-speech participles serve as entities to be completed; judging whether each input rail transit standard is a simple sentence or not by using a symbol detection method, if so, performing a step 4, and if not, performing a step 5;
and 4, step 4: searching entity related attributes in the rail transit specification, judging the relationship between the entity and the entity, generating entity relationship triples and storing the entity relationship triples;
and 5: performing dependency syntax analysis based on deep learning on the rail transit standard, if the previous sentence is an NP phrase with a parallel structure, performing step 6, and if not, performing step 7;
step 6: extracting the attribute words and entities of the later sentence to ensure that the attribute words n of the earlier sentence and the attribute words n of the later sentence are: n, matching, generating entity relation triples and storing the entity relation triples;
and 7: judging whether the grammar of the front sentence is a main predicate object and the grammar of the rear sentence is an object supplement, if so, performing a step 8, and if not, performing a step 9;
and 8: directly matching the front sentence entity with the object, and directly matching the rear sentence keyword with the object entity to generate an entity relationship triple and store the entity relationship triple;
and step 9: calculating the vocabulary relevancy, outputting the entity-entity relationship with the relevancy exceeding a threshold value, generating an entity-relationship triple, and storing the entity-relationship triple;
step 10: and (4) outputting the entity relationship triples generated in the steps 4, 8 and 9 to obtain a complete semantic structure entity specification, namely completing the automatic completion of the track traffic specification entity relationship.
2. The rail transit regulation entity relationship automatic completion method based on artificial intelligence of claim 1, wherein the entity relationship triple is "entity-verb-entity" or "entity-degree-attribute".
3. The method for automatically completing the track traffic regulation entity relationship based on artificial intelligence according to claim 1, wherein the specific operation steps of the step 1 are as follows:
step 1.1: acquiring rail transit standard textual data, preprocessing and training the acquired rail transit standard textual data, and generating a dictionary;
step 1.2: processing the dictionary, mining the missing features, and extracting entity completion rules and methods;
step 1.3: and constructing an entity relation completion model by adopting the extracted entity completion rule and method.
4. The method as claimed in claim 1, wherein in step 2, a Bi-LSTM + CRF part-of-speech tagging model is used to tag part-of-speech of the rail transit specification, wherein the adjective part-of-speech is divided into attribute words, and the verb part-of-speech is used to determine the relationship between the entity and the entity.
5. The method as claimed in claim 4, wherein in the step 3, the simple sentence is a sentence with a pause and a period.
6. The method for automatically completing the track traffic regulation entity relationship based on artificial intelligence according to claim 5, wherein the specific operation steps of the step 4 are as follows:
step 4.1: searching attributes and action relations related to the entities in the rail transit specification;
step 4.2: extracting verbs, judging the relation between the entities, and analyzing the part of speech to extract attributes;
step 4.4: and generating an entity relationship triple according to the entity-entity relationship and the extracted attributes, and storing the entity relationship.
7. The method for automatically completing the track traffic regulation entity relationship based on artificial intelligence according to claim 1, wherein the specific operation steps of the step 9 are as follows:
step 9.1: judging whether the front sentence is in a non-parallel structure or not and the rear sentence has a substitute word or not, if so, calculating the vocabulary relevancy between the attribute word after the substitute word and all entities of the front sentence, otherwise, naming the entity recognition phrase and calculating the vocabulary relevancy between all the participles;
step 9.2: and outputting the entity and entity relationship with the vocabulary relevancy exceeding the threshold value, generating an entity relationship triple and storing the entity relationship triple.
8. The method for automatically completing rail transit regulation entity relationship based on artificial intelligence as claimed in claim 7, wherein in the step 9, the vocabulary relevancy is calculated by using a vocabulary relevancy calculation algorithm based on a word bank of a known network.
9. The method for automatically completing the rail transit regulation entity relationship based on artificial intelligence as claimed in claim 8, wherein the calculation algorithm of the vocabulary relevancy based on the word bank of the known network is as follows:
rel(w1，w2)＝max{a1*sim(s1，s2)+(1-a1)*asso(s1，s2)}
Figure FDA0002475839520000031
asso(s1，s2)＝∑ri*asso(p1，p2)
in the above formula, rel (w)1，w2) The expression vocabulary w1And the word w2Correlation of (c), sim(s)1,s2) The expression vocabulary w1And the word w2Similarity of (a), asso(s)1,s2) Representing an entity s1And s2The semantic relevance of (2); alpha1Expressing adjustable parameters for linearly harmonizing the similarity and the semantic association degree, wherein the value range of the adjustable parameters is [0,1 ]]；s1iTo representWord w1Meaning term of (i) ═ 1, …, n denotes the word w1Having n sense items; s2jThe expression vocabulary w2J is 1, …, m represents the word w2Having m sense items; gamma rayiThe semantic relevance coefficient representing different parts of the entity concept is a fit to each part of the two concepts, and must satisfy ∑ ri＝1，p1Is an item of sense s1Of (2) a sense atom, p2Is an item of sense s2The sense of (1).
CN202010363261.5A 2020-04-30 2020-04-30 Rail transit standard entity relation automatic completion method based on artificial intelligence Active CN111597349B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010363261.5A CN111597349B (en) 2020-04-30 2020-04-30 Rail transit standard entity relation automatic completion method based on artificial intelligence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010363261.5A CN111597349B (en) 2020-04-30 2020-04-30 Rail transit standard entity relation automatic completion method based on artificial intelligence

Publications (2)

Publication Number Publication Date
CN111597349A true CN111597349A (en) 2020-08-28
CN111597349B CN111597349B (en) 2022-10-11

Family

ID=72185606

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010363261.5A Active CN111597349B (en) 2020-04-30 2020-04-30 Rail transit standard entity relation automatic completion method based on artificial intelligence

Country Status (1)

Country Link
CN (1) CN111597349B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112463960A (en) * 2020-10-30 2021-03-09 完美世界控股集团有限公司 Entity relationship determination method and device, computing equipment and storage medium
CN112949280A (en) * 2021-03-02 2021-06-11 中国联合网络通信集团有限公司 Data processing method and device
CN116049148A (en) * 2023-04-03 2023-05-02 中国科学院成都文献情报中心 Construction method of domain meta knowledge engine in meta publishing environment

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170032275A1 (en) * 2015-07-28 2017-02-02 Linkedin Corporation Entity matching for ingested profile data
CN106815293A (en) * 2016-12-08 2017-06-09 中国电子科技集团公司第三十二研究所 System and method for constructing knowledge graph for information analysis
CN108763353A (en) * 2018-05-14 2018-11-06 中山大学 Rule-based and remote supervisory Baidupedia relationship triple abstracting method
CN109241538A (en) * 2018-09-26 2019-01-18 上海德拓信息技术股份有限公司 Based on the interdependent Chinese entity relation extraction method of keyword and verb
CN109614501A (en) * 2018-12-13 2019-04-12 浙江工商大学 A kind of industrial hidden danger standardization report method and system of knowledge based map
CN110263019A (en) * 2019-06-18 2019-09-20 中南民族大学 Construction method, device and the storage medium of entity relation extraction model
CN110287704A (en) * 2019-06-25 2019-09-27 北京中科微澜科技有限公司 A kind of loophole software dependence construction method based on loophole map

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170032275A1 (en) * 2015-07-28 2017-02-02 Linkedin Corporation Entity matching for ingested profile data
CN106815293A (en) * 2016-12-08 2017-06-09 中国电子科技集团公司第三十二研究所 System and method for constructing knowledge graph for information analysis
CN108763353A (en) * 2018-05-14 2018-11-06 中山大学 Rule-based and remote supervisory Baidupedia relationship triple abstracting method
CN109241538A (en) * 2018-09-26 2019-01-18 上海德拓信息技术股份有限公司 Based on the interdependent Chinese entity relation extraction method of keyword and verb
CN109614501A (en) * 2018-12-13 2019-04-12 浙江工商大学 A kind of industrial hidden danger standardization report method and system of knowledge based map
CN110263019A (en) * 2019-06-18 2019-09-20 中南民族大学 Construction method, device and the storage medium of entity relation extraction model
CN110287704A (en) * 2019-06-25 2019-09-27 北京中科微澜科技有限公司 A kind of loophole software dependence construction method based on loophole map

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
JIE SUN ET AL.: ""Knowledge Map Completion Method Based on Metric Space and Relational Path"", 《INTERNATIONAL CONFERENCE ON COMPUTER SCIENCE & EDUCATION》 *
乔晶晶: ""多源知识库实体对齐与属性补全"", 《万方数据知识服务平台》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112463960A (en) * 2020-10-30 2021-03-09 完美世界控股集团有限公司 Entity relationship determination method and device, computing equipment and storage medium
CN112463960B (en) * 2020-10-30 2021-07-27 完美世界控股集团有限公司 Entity relationship determination method and device, computing equipment and storage medium
CN112949280A (en) * 2021-03-02 2021-06-11 中国联合网络通信集团有限公司 Data processing method and device
CN112949280B (en) * 2021-03-02 2023-07-07 中国联合网络通信集团有限公司 Data processing method and device
CN116049148A (en) * 2023-04-03 2023-05-02 中国科学院成都文献情报中心 Construction method of domain meta knowledge engine in meta publishing environment
CN116049148B (en) * 2023-04-03 2023-07-18 中国科学院成都文献情报中心 Construction method of domain meta knowledge engine in meta publishing environment

Also Published As

Publication number Publication date
CN111597349B (en) 2022-10-11

Similar Documents

Publication Publication Date Title
CN110442760B (en) Synonym mining method and device for question-answer retrieval system
CN110059311B (en) Judicial text data-oriented keyword extraction method and system
CN109271529B (en) Method for constructing bilingual knowledge graph of Xilier Mongolian and traditional Mongolian
Rahman et al. Narrowing the modeling gap: A cluster-ranking approach to coreference resolution
CN111597349B (en) Rail transit standard entity relation automatic completion method based on artificial intelligence
US20150227505A1 (en) Word meaning relationship extraction device
CN111325029B (en) Text similarity calculation method based on deep learning integrated model
Xie et al. Topic enhanced deep structured semantic models for knowledge base question answering
CN110502744B (en) Text emotion recognition method and device for historical park evaluation
Sahu et al. Prashnottar: a Hindi question answering system
EP2430568A1 (en) Methods and systems for knowledge discovery
CN111444704B (en) Network safety keyword extraction method based on deep neural network
CN113196277A (en) System for retrieving natural language documents
CN109522396B (en) Knowledge processing method and system for national defense science and technology field
CN110675962A (en) Traditional Chinese medicine pharmacological action identification method and system based on machine learning and text rules
CN116775874B (en) Information intelligent classification method and system based on multiple semantic information
CN113196278A (en) Method for training a natural language search system, search system and corresponding use
CN112183059A (en) Chinese structured event extraction method
CN116775812A (en) Traditional Chinese medicine patent analysis and excavation tool based on natural voice processing
CN110399603A (en) A kind of text-processing technical method and system based on sense-group division
CN109446313B (en) Sequencing system and method based on natural language analysis
CN111241299A (en) Knowledge graph automatic construction method for legal consultation and retrieval system thereof
CN114219248A (en) Man-sentry matching method based on LDA model, dependency syntax and deep learning
CN111597330A (en) Intelligent expert recommendation-oriented user image drawing method based on support vector machine
Kessler et al. Extraction of terminology in the field of construction

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