CN111026877A - Knowledge verification model construction and analysis method based on probability soft logic - Google Patents

Knowledge verification model construction and analysis method based on probability soft logic Download PDF

Info

Publication number
CN111026877A
CN111026877A CN201911236858.7A CN201911236858A CN111026877A CN 111026877 A CN111026877 A CN 111026877A CN 201911236858 A CN201911236858 A CN 201911236858A CN 111026877 A CN111026877 A CN 111026877A
Authority
CN
China
Prior art keywords
entity
knowledge
logic
rule
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.)
Pending
Application number
CN201911236858.7A
Other languages
Chinese (zh)
Inventor
韩伟红
宫云宝
陈雷霆
林长海
刘健威
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.)
Guangdong Electronic Information Engineering Research Institute of UESTC
Original Assignee
Guangdong Electronic Information Engineering Research Institute of UESTC
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 Guangdong Electronic Information Engineering Research Institute of UESTC filed Critical Guangdong Electronic Information Engineering Research Institute of UESTC
Priority to CN201911236858.7A priority Critical patent/CN111026877A/en
Publication of CN111026877A publication Critical patent/CN111026877A/en
Pending legal-status Critical Current

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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/02—Knowledge representation; Symbolic representation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G06N7/01—Probabilistic graphical models, e.g. probabilistic networks

Abstract

The invention belongs to the technical field of information extraction, and particularly relates to a knowledge verification model construction and analysis method based on probability soft logic, which comprises the following steps: a. forming a candidate knowledge set by the knowledge extracted from the web texts of the web pages of a plurality of data sources by the information extraction system; b. carrying out reliability calculation on the candidate knowledge set; c. performing logic predicate expression on each entity in the candidate knowledge set; d. the method comprises the steps of constructing a first-order logic rule of a knowledge verification model based on entity analysis and body constraint respectively, and generating the first-order logic rule in a probability soft logic model through the constructed logic rule to realize entity relation and entity label verification in a candidate knowledge set; e. and setting probability distribution of the knowledge verification model and selecting corresponding knowledge to be updated through calculation of an inference algorithm. According to the invention, the accuracy of the candidate knowledge set is greatly improved by verifying the candidate knowledge set.

Description

Knowledge verification model construction and analysis method based on probability soft logic
Technical Field
The invention belongs to the technical field of information extraction, and particularly relates to a knowledge verification model construction and analysis method based on probability soft logic.
Background
With the rapid popularization of the internet and the continuous deepening of the Web3.0 concept, the establishment of a large-scale knowledge base becomes an urgent necessity of various industries, and meanwhile, search engines and research institutions also release various large-scale knowledge bases. For example, large-scale multi-user online editing knowledge bases such as wikipedia, Baidu encyclopedia, etc., Freebase knowledge base of Google, KnowledgeGraph, etc.
The knowledge graph construction is a dynamic process, new knowledge needs to be added and dynamic knowledge needs to be updated in time, and the existing knowledge graph is perfected and supplemented. The new knowledge acquisition method is often used for extracting from different data sources by adopting information extraction technology, for example, NELL, TextRunner continuously extracts new knowledge from the web to construct a knowledge base. The problem of inconsistency often exists when the information extraction system extracts new knowledge, for example, entity labels in the new knowledge are not unique, multiple representation forms of entities, uncertain entity relationships and the like. The accuracy of the new knowledge determines the overall quality of the constructed knowledge graph, so that the verification of the new knowledge is an important process for constructing the knowledge graph, and the new knowledge with high reliability can be selected as a candidate set to construct the high-quality knowledge graph through knowledge verification. Under the background, the method for verifying knowledge constructed by researching the subject and facing the knowledge map has deep theoretical and practical significance. At present, knowledge verification research methods for knowledge graph construction are few, and most of researches are knowledge fusion technologies for data sources.
Therefore, there is a need for an improved solution with respect to knowledge validation model construction.
Disclosure of Invention
The invention aims to: aiming at the defects of the prior art, the invention provides the knowledge verification model construction and analysis method based on the probability soft logic, and the accuracy of the candidate knowledge set is greatly improved by verifying the candidate knowledge set.
In order to achieve the purpose, the invention adopts the following technical scheme:
a knowledge verification model construction and analysis method based on probability soft logic comprises the following steps:
a. forming a candidate knowledge set by the knowledge extracted from the web texts of the web pages of a plurality of data sources by the information extraction system;
b. carrying out reliability calculation on the candidate knowledge set;
c. performing logic predicate expression on each entity in the candidate knowledge set;
d. the method comprises the steps of constructing a first-order logic rule of a knowledge verification model based on entity analysis and body constraint respectively, and generating the first-order logic rule in a probability soft logic model through the constructed logic rule to realize entity relation and entity label verification in a candidate knowledge set;
e. and setting probability distribution of the knowledge verification model and selecting corresponding knowledge to be updated through calculation of an inference algorithm.
As an improvement of the method for constructing and analyzing the knowledge verification model based on the probabilistic soft logic, in the step a, the knowledge of the candidate knowledge set is represented by an RDF triple, and the candidate knowledge set is set as a triple < S > < P > < O > with the confidence level set as F (S > < P > < O >), wherein the value of F is a continuous value between [0 and 1 ].
As an improvement on the knowledge verification model construction and analysis method based on the probabilistic soft logic in the present invention, the step b specifically includes the following steps:
b1, dividing the extracted knowledge into subjective knowledge and objective knowledge according to the extracted knowledge source, and setting an objectivity score of each knowledge, wherein the score belongs to {0,1}, and is 0 when the knowledge is subjective knowledge and 1 when the knowledge is objective knowledge;
b2, calculating the credibility of the candidate knowledge set according to the formula (1),
Figure BDA0002305124830000031
wherein D issFor knowledge source web document collections, O (d)i) Objectivity score for document d containing candidate knowledge i, FiIs the credibility score of the candidate knowledge i.
As an improvement to the probabilistic soft logic-based knowledge verification model building and analyzing method in the present invention, in the step c, the logic predicate of each entity in the candidate knowledge set is represented as canent (E), and the logic predicate of the entity relationship is represented as CanRel (E)1,E2R), the logical predicate of the entity label is represented as CanLbl (E, L).
As an improvement to the knowledge verification model construction and analysis method based on probabilistic soft logic in the present invention, the step c further includes respectively adopting CanRel for different extraction technologiesT() And CanlblT() Expressing, and establishing an inference logic rule according to the entity, the entity relation, the logic predicate of the entity label and the target knowledge:
Figure BDA0002305124830000032
Figure BDA0002305124830000041
where T represents the different decimation techniques.
As an improvement of the knowledge verification model construction and analysis method based on the probabilistic soft logic in the present invention, the logic rule constructed based on the entity analysis in step d is:
Figure BDA0002305124830000042
Figure BDA0002305124830000043
Figure BDA0002305124830000044
Figure BDA0002305124830000045
wherein the SameEnt () logical predicate represents the similarity between the entities, and the logical rules (d1) and (d2) represent when the entity E is1And E2Similarly, if one of the entities E1Or E2Is L, then another entity E2Or E1Class is also L, and the weight is respectively
Figure BDA0002305124830000046
And
Figure BDA0002305124830000047
the logic rules (d3) and (d4) indicate when entity E is1And E2Similarly, if one of the entities E1And E3The relationship being R or entity E2And E3The relationship is R, then entity E2And E3The relationship being R or entity E1And E3The relationship is R, the weight of which is respectively
Figure BDA0002305124830000048
And
Figure BDA0002305124830000049
as an improvement of the knowledge verification model construction and analysis method based on the probabilistic soft logic in the present invention, the logic rule constructed based on the ontology constraint in step d is:
Figure BDA00023051248300000410
Figure BDA00023051248300000411
Figure BDA00023051248300000412
Figure BDA00023051248300000413
Figure BDA00023051248300000414
Figure BDA00023051248300000415
Figure BDA0002305124830000051
wherein, the logic rule (d5) indicates that under the ontology constraint Dom, if the entity E1And entity E2The relationship between is R, then entity E1Is L, the logical rule weight size is WO-Dom(ii) a The logical rule (d6) indicates that under the ontology constraint Rng, if entity E1And entity E2The relationship between is R, then entity E2Is L, the logical rule weight size is WO-Rng(ii) a The logical rule (d7) indicates that the entity relationships R and S are reciprocal if entity E1And entity E2The relationship between is R, then entity E2And entity E1The relation between is S, the logical rule weight is WO-Inv(ii) a The logical rule (d8) indicates that the entity label L is P subset, if the category of the entity E is L, the category of the entity E is P, and the weight of the logical rule is WO-Sub(ii) a The logical rule (d9) indicates that the entity relationship R is a subset of S if the entity E1And entity E2The relationship between is R, then entity E2And entity E1The relation between is S, the logical rule weight is WO-RSub(ii) a The logical rule (d10) represents an entity label L1And L2Is mutually exclusiveIf the label L of entity E1Then entity E must not have a label L2The logical rule weight is WO-Mut(ii) a The logical rule (d11) indicates that the entity relationships R and S are mutually exclusive if entity E1And entity E2The relationship between is R, then entity E1And entity E2The relation between is not S necessarily, the weight of the logic rule is WO-RMut。
As an improvement to the method for constructing and analyzing the knowledge verification model based on the probabilistic soft logic in the present invention, the knowledge verification model in step e is configured as follows:
Figure BDA0002305124830000052
wherein R is a set of logic rules in the probabilistic soft logic model, λrDenotes the weight of the logic rule r, Z denotes the planning factor, d (r) denotes the distance satisfaction of the logic rule r, and p ═ 1 denotes the first order logic rule.
As an improvement of the knowledge verification model construction and analysis method based on the probabilistic soft logic in the present invention, the inference algorithm in the step e is an MPE inference algorithm or a marginal inference algorithm.
The invention has the beneficial effects that: compared with the prior art, the reliability of the candidate knowledge set is evaluated by providing a reliability calculation model; the knowledge verification model solves the problems of errors and the like of entity relations and entity labels, and then obtains a knowledge set to be updated of the high-quality knowledge graph, so that the accuracy of the candidate knowledge set is greatly improved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention and not to limit the invention. In the drawings:
FIG. 1 is a schematic diagram of a process implementation of the present invention;
FIG. 2 is a basic framework diagram of the knowledge verification model in the present invention.
Detailed Description
As used in the specification and in the claims, certain terms are used to refer to particular components. As one skilled in the art will appreciate, manufacturers may refer to a component by different names. This specification and claims do not intend to distinguish between components that differ in name but not function. In the following description and in the claims, the terms "include" and "comprise" are used in an open-ended fashion, and thus should be interpreted to mean "include, but not limited to. "substantially" means within an acceptable error range, within which a person skilled in the art can solve the technical problem to substantially achieve the technical result.
In the description of the present invention, it is to be understood that the terms "upper", "lower", "front", "rear", "left", "right", horizontal ", and the like indicate orientations or positional relationships based on those shown in the drawings, and are only for convenience in describing the present invention and simplifying the description, but do not indicate or imply that the referred device or element must have a specific orientation, be constructed in a specific orientation, and be operated, and thus, should not be construed as limiting the present invention.
In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "secured," and the like are to be construed broadly and can, for example, be fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
The present invention will be described in further detail below with reference to the accompanying drawings, but the present invention is not limited thereto.
As shown in fig. 1-2, the knowledge verification model construction and analysis method based on the probability soft logic includes the following steps:
a. forming a candidate knowledge set by the knowledge extracted from the web texts of the web pages of a plurality of data sources by the information extraction system;
b. carrying out reliability calculation on the candidate knowledge set;
c. performing logic predicate expression on each entity in the candidate knowledge set;
d. the method comprises the steps of constructing a first-order logic rule of a knowledge verification model based on entity analysis and body constraint respectively, and generating the first-order logic rule in a probability soft logic model through the constructed logic rule to realize entity relation and entity label verification in a candidate knowledge set;
e. and setting probability distribution of the knowledge verification model and selecting corresponding knowledge to be updated through calculation of an inference algorithm.
Preferably, in step a, the knowledge of the candidate knowledge set is represented by an RDF triple, and the candidate knowledge set is set as a triple < S > < P > < O > with a confidence level of F (< S > < P > < O >), wherein F takes a continuous value between [0 and 1 ].
Preferably, step b specifically comprises the following steps:
b1, dividing the extracted knowledge into subjective knowledge and objective knowledge according to the extracted knowledge source, and setting an objectivity score of each knowledge, wherein the score belongs to {0,1}, and is 0 when the knowledge is subjective knowledge and 1 when the knowledge is objective knowledge;
b2, calculating the credibility of the candidate knowledge set according to the formula (1),
Figure BDA0002305124830000081
wherein D issFor knowledge source web document collections, O (d)i) Objectivity score for document d containing candidate knowledge i, FiIs the credibility score of the candidate knowledge i.
In the invention, most of objective knowledge is sourced from encyclopedias, scientific publications, textbooks and the like, such as news reports, hundred-degree encyclopedia knowledge and the like, most of the knowledge is objective and real, and personal real emotion does not exist; the subjective knowledge sources such as forums, posts and the like are all individual thought statements, and have individual prejudices and low authenticity.
Preferably, the logical predicate of each entity in the candidate knowledge set in step c is represented as canent (E), and the logical predicate of the entity relationship is represented as CanRel (E)1,E2R), the logical predicate of the entity label is represented as CanLbl (E, L).
Preferably, step c further comprises using CanRel for different extraction techniquesT() And CanlblT() Expressing, and establishing an inference logic rule according to the entity, the entity relation, the logic predicate of the entity label and the target knowledge:
Figure BDA0002305124830000091
Figure BDA0002305124830000092
where T represents the different decimation techniques.
Entity resolution essentially computes entity E1With entity E2According to the calculated similarity and the entity E therein1To deduce a further entity E2To correct the correct knowledge of. Therefore, in the invention, the similarity between the entities is represented by adopting the Sameent () logic predicate, and the logic rule constructed based on entity analysis in the step d is as follows:
Figure BDA0002305124830000093
Figure BDA0002305124830000094
Figure BDA0002305124830000095
Figure BDA0002305124830000096
wherein the logic rules (d1) and (d2) indicate when the entity E is1And E2Similarly, if one of the entities E1Or E2Is L, then another entity E2Or E1Class is also L, and the weight is respectively
Figure BDA0002305124830000097
And
Figure BDA0002305124830000098
the logic rules (d3) and (d4) indicate when entity E is1And E2Similarly, if one of the entities E1And E3The relationship being R or entity E2And E3The relationship is R, then entity E2And E3The relationship being R or entity E1And E3The relationship is R, the weight of which is respectively
Figure BDA0002305124830000099
And
Figure BDA00023051248300000910
the same entities in the above logic rules must have the same entity labels and relationships, and when SameEnt () has a high true value, the two entities are very similar, so any label assigned by the first entity will also be assigned to the second entity; on the other hand, if the similarity score of two entities is low, the truth of their respective labels and relationships will not be strongly constrained.
In the process of building the knowledge-graph architecture, an ontology mode needs to be defined first, and then an entity is added into the knowledge graph. In the process of updating the knowledge map, newly added knowledge must obey the ontology constraint defined in advance, so that the invention constructs a first-order logic rule based on the existing ontology constraint and tests the verification of the candidate knowledge set. Wherein the logical predicate representation of the common ontology constraint is shown in Table 1,
table 1 ontology predicates
Type(E,L) The label representing entity E is L
Dom(R,L) Representing associations of entity relationships to entity labels
INV(R,S) Representing the entity relationship R as reflexive to S
RNG(R,L) Representing associations of entity labels to entity relationships
SUB(L,P) Indicating that entity label L is a subset of entity label P
RSUB(R,S) Indicating that entity relationship R is a subset of entity relationship S
MUT(L1,L2) Indicating that entity tags L1 and L2 have mutually exclusive relationships
RMUT(R,S) Indicating that the entity relationships R and S have mutually exclusive relationships
Therefore, the logic rule constructed based on the ontology constraint in step d is:
Figure BDA0002305124830000101
Figure BDA0002305124830000102
Figure BDA0002305124830000103
Figure BDA0002305124830000104
Figure BDA0002305124830000105
Figure BDA0002305124830000106
Figure BDA0002305124830000107
wherein, the logic rule (d5) indicates that under the ontology constraint Dom, if the entity E1And entity E2The relationship between is R, then entity E1Is L, the logical rule weight size is WO-Dom(ii) a The logical rule (d6) indicates that under the ontology constraint Rng, if entity E1And entity E2The relationship between is R, then entity E2Is L, the logical rule weight size is WO-Rng(ii) a The logical rule (d7) indicates that the entity relationships R and S are reciprocal if entity E1And entity E2The relationship between is R, then entity E2And entity E1The relation between is S, the logical rule weight is WO-Inv(ii) a The logical rule (d8) indicates that the entity label L is P subset, if the category of the entity E is L, the category of the entity E is P, and the weight of the logical rule is WO-Sub(ii) a The logical rule (d9) indicates that the entity relationship R is a subset of S if the entity E1And entity E2The relationship between is R, then entity E2And entity E1The relation between is S, the logical rule weight is WO-RSub(ii) a The logical rule (d10) represents an entity label L1And L2Is mutually exclusive if the label L of entity E1Then entity E must not have a label L2The logical rule weight is WO-Mut(ii) a The logical rule (d11) indicates that the entity relationships R and S are mutually exclusive if entity E1And entity E2The relationship between is R, then entity E1And entity E2The relation between is not S necessarily, the weight of the logic rule is WO-RMut。
And constructing a weighted first-order logic rule based on entity analysis and body constraint, and generating a first-order logic rule in a probability logic model through the constructed logic rule to realize entity relation and entity label verification in the candidate knowledge set.
Preferably, the knowledge verification model in step e is set as:
Figure BDA0002305124830000111
wherein R is a set of logic rules in the probabilistic soft logic model, λrDenotes the weight of the logic rule r, Z denotes the planning factor, d (r) denotes the distance satisfaction of the logic rule r, and p ═ 1 denotes the first order logic rule.
Preferably, the inference algorithm in step e is an MPE inference algorithm or a marginal inference algorithm. The two algorithms are the MPE reasoning mechanism which is most widely used in the probability soft logic model, and because the MPE algorithm is more suitable for solving the probability value calculation of the uncertainty problem, the MPE reasoning algorithm is adopted for calculation in the embodiment, the correct maximum probability value of each candidate knowledge is calculated through the MPE reasoning algorithm, and the high-quality knowledge is selected as the knowledge to be updated through setting a reasonable threshold.
The foregoing description shows and describes several preferred embodiments of the invention, but as aforementioned, it is to be understood that the invention is not limited to the forms disclosed herein, but is not to be construed as excluding other embodiments and is capable of use in various other combinations, modifications, and environments and is capable of changes within the scope of the inventive concept as expressed herein, commensurate with the above teachings, or the skill or knowledge of the relevant art. And that modifications and variations may be effected by those skilled in the art without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (9)

1. The knowledge verification model construction and analysis method based on the probability soft logic is characterized by comprising the following steps of:
a. forming a candidate knowledge set by the knowledge extracted from the web texts of the web pages of a plurality of data sources by the information extraction system;
b. carrying out reliability calculation on the candidate knowledge set;
c. performing logic predicate expression on each entity in the candidate knowledge set;
d. the method comprises the steps of constructing a first-order logic rule of a knowledge verification model based on entity analysis and body constraint respectively, and generating the first-order logic rule in a probability soft logic model through the constructed logic rule to realize entity relation and entity label verification in a candidate knowledge set;
e. and setting probability distribution of the knowledge verification model and selecting corresponding knowledge to be updated through calculation of an inference algorithm.
2. The probabilistic soft logic-based knowledge verification model building and analyzing method of claim 1, wherein: in the step a, knowledge of the candidate knowledge set is represented by an RDF triple, and the candidate knowledge set is set as a triple < S > < P > < O >, and the reliability of the triple is set as F (< S > < P > < O >), wherein the value of F is a continuous value between [0 and 1 ].
3. The probabilistic soft logic-based knowledge verification model construction and analysis method according to claim 1, wherein the step b specifically comprises the steps of:
b1, dividing the extracted knowledge into subjective knowledge and objective knowledge according to the extracted knowledge source, and setting an objectivity score of each knowledge, wherein the score belongs to {0,1}, and is 0 when the knowledge is subjective knowledge and 1 when the knowledge is objective knowledge;
b2, calculating the credibility of the candidate knowledge set according to the formula (1),
Figure FDA0002305124820000021
wherein D issFor knowledge source web document collections, O (d)i) Objectivity score for document d containing candidate knowledge i, FiIs the credibility score of the candidate knowledge i.
4. The probabilistic soft logic-based knowledge verification model building and analyzing method of claim 1, wherein: the logic predicate of each entity in the candidate knowledge set in the step c is represented as CanEnt (E), and the logic predicate of the entity relationship is represented as CanRel (E)1,E2R), the logical predicate of the entity label is represented as CanLbl (E, L).
5. The probabilistic soft logic-based knowledge verification model construction and analysis method according to claim 4, wherein the step c further comprises adopting CanRel for different extraction technologies respectivelyT() And CanlblT() Expressing, and establishing an inference logic rule according to the entity, the entity relation, the logic predicate of the entity label and the target knowledge:
Figure FDA0002305124820000022
Figure FDA0002305124820000023
where T represents the different decimation techniques.
6. The method for constructing and analyzing a knowledge verification model based on probabilistic soft logic according to claim 1, wherein the logic rules constructed based on entity parsing in the step d are:
Figure FDA0002305124820000031
Figure FDA0002305124820000032
Figure FDA0002305124820000033
Figure FDA0002305124820000034
wherein the SameEnt () logical predicate represents the similarity between the entities, and the logical rules (d1) and (d2) represent when the entity E is1And E2Similarly, if one of the entities E1Or E2Is L, then another entity E2Or E1Class is also L, and the weight is respectively
Figure FDA0002305124820000035
And
Figure FDA0002305124820000036
the logic rules (d3) and (d4) indicate when entity E is1And E2Similarly, if one of the entities E1And E3The relationship being R or entity E2And E3The relationship is R, then entity E2And E3The relationship being R or entity E1And E3The relationship is R, the weight of which is respectively
Figure FDA0002305124820000037
And
Figure FDA0002305124820000038
7. the method for constructing and analyzing a knowledge verification model based on probabilistic soft logic according to claim 1, wherein the logic rules constructed based on ontology constraint in the step d are:
Figure FDA0002305124820000039
Figure FDA00023051248200000310
Figure FDA00023051248200000311
Figure FDA00023051248200000312
Figure FDA00023051248200000313
Figure FDA0002305124820000041
Figure FDA0002305124820000042
wherein, the logic rule (d5) indicates that under the ontology constraint Dom, if the entity E1And entity E2The relationship between is R, then entity E1Is L, the logical rule weight size is WO-Dom(ii) a The logical rule (d6) indicates that under the ontology constraint Rng, if entity E1And entity E2The relationship between is R, then entity E2Is L, the logical rule weight size is WO-Rng(ii) a The logical rule (d7) indicates that the entity relationships R and S are reciprocal if entity E1And entity E2The relationship between is R, then entity E2And entity E1The relationship between is S, the logicThe weight of the edit rule is WO-Inv(ii) a The logical rule (d8) indicates that the entity label L is P subset, if the category of the entity E is L, the category of the entity E is P, and the weight of the logical rule is WO-Sub(ii) a The logical rule (d9) indicates that the entity relationship R is a subset of S if the entity E1And entity E2The relationship between is R, then entity E2And entity E1The relation between is S, the logical rule weight is WO-RSub(ii) a The logical rule (d10) represents an entity label L1And L2Is mutually exclusive if the label L of entity E1Then entity E must not have a label L2The logical rule weight is WO-Mut(ii) a The logical rule (d11) indicates that the entity relationships R and S are mutually exclusive if entity E1And entity E2The relationship between is R, then entity E1And entity E2The relation between is not S necessarily, the weight of the logic rule is WO-RMut。
8. The probabilistic soft logic-based knowledge verification model construction and analysis method according to claim 1, wherein the knowledge verification model in the step e is configured as:
Figure FDA0002305124820000043
wherein R is a set of logic rules in the probabilistic soft logic model, λrDenotes the weight of the logic rule r, Z denotes the planning factor, d (r) denotes the distance satisfaction of the logic rule r, and p ═ 1 denotes the first order logic rule.
9. The probabilistic soft logic-based knowledge verification model building and analyzing method of claim 1, wherein: and e, the inference algorithm in the step e is an MPE inference algorithm or a marginal inference algorithm.
CN201911236858.7A 2019-12-05 2019-12-05 Knowledge verification model construction and analysis method based on probability soft logic Pending CN111026877A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911236858.7A CN111026877A (en) 2019-12-05 2019-12-05 Knowledge verification model construction and analysis method based on probability soft logic

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911236858.7A CN111026877A (en) 2019-12-05 2019-12-05 Knowledge verification model construction and analysis method based on probability soft logic

Publications (1)

Publication Number Publication Date
CN111026877A true CN111026877A (en) 2020-04-17

Family

ID=70204419

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911236858.7A Pending CN111026877A (en) 2019-12-05 2019-12-05 Knowledge verification model construction and analysis method based on probability soft logic

Country Status (1)

Country Link
CN (1) CN111026877A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112529186A (en) * 2021-02-18 2021-03-19 中国科学院自动化研究所 Industrial process optimization decision knowledge reasoning method embedded with domain rules
CN112529184A (en) * 2021-02-18 2021-03-19 中国科学院自动化研究所 Industrial process optimization decision method fusing domain knowledge and multi-source data
CN112529185A (en) * 2021-02-18 2021-03-19 中国科学院自动化研究所 Industrial process field rule knowledge acquisition method
CN113220973A (en) * 2021-05-31 2021-08-06 北京海纳数聚科技有限公司 Public opinion truth testing method based on knowledge reasoning technology
CN114880406A (en) * 2022-05-05 2022-08-09 国网智能电网研究院有限公司 Data management method and device

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
宫云宝: "面向知识图谱构建的知识验证方法研究" *

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112529186A (en) * 2021-02-18 2021-03-19 中国科学院自动化研究所 Industrial process optimization decision knowledge reasoning method embedded with domain rules
CN112529184A (en) * 2021-02-18 2021-03-19 中国科学院自动化研究所 Industrial process optimization decision method fusing domain knowledge and multi-source data
CN112529185A (en) * 2021-02-18 2021-03-19 中国科学院自动化研究所 Industrial process field rule knowledge acquisition method
CN112529184B (en) * 2021-02-18 2021-07-02 中国科学院自动化研究所 Industrial process optimization decision method fusing domain knowledge and multi-source data
US11409270B1 (en) * 2021-02-18 2022-08-09 Institute Of Automation, Chinese Academy Of Sciences Optimization decision-making method of industrial process fusing domain knowledge and multi-source data
US20220260981A1 (en) * 2021-02-18 2022-08-18 Institute Of Automation, Chinese Academy Of Sciences Optimization decision-making method of industrial process fusing domain knowledge and multi-source data
US11455548B2 (en) 2021-02-18 2022-09-27 Institute Of Automation, Chinese Academy Of Sciences Acquisition method for domain rule knowledge of industrial process
CN113220973A (en) * 2021-05-31 2021-08-06 北京海纳数聚科技有限公司 Public opinion truth testing method based on knowledge reasoning technology
CN113220973B (en) * 2021-05-31 2023-10-24 北京海纳数聚科技有限公司 Public opinion verification method based on knowledge reasoning technology
CN114880406A (en) * 2022-05-05 2022-08-09 国网智能电网研究院有限公司 Data management method and device

Similar Documents

Publication Publication Date Title
CN109472033B (en) Method and system for extracting entity relationship in text, storage medium and electronic equipment
CN111414393B (en) Semantic similar case retrieval method and equipment based on medical knowledge graph
CN111026877A (en) Knowledge verification model construction and analysis method based on probability soft logic
Matci et al. Address standardization using the natural language process for improving geocoding results
CN109036577B (en) Diabetes complication analysis method and device
CN112256828B (en) Medical entity relation extraction method, device, computer equipment and readable storage medium
CN113535974B (en) Diagnostic recommendation method and related device, electronic equipment and storage medium
CN108681557B (en) Short text topic discovery method and system based on self-expansion representation and similar bidirectional constraint
US8239349B2 (en) Extracting data
US20090192954A1 (en) Semantic Relationship Extraction, Text Categorization and Hypothesis Generation
CN112667794A (en) Intelligent question-answer matching method and system based on twin network BERT model
CN104699730A (en) Identifying and displaying relationships between candidate answers
CN111949759A (en) Method and system for retrieving medical record text similarity and computer equipment
CN113707297A (en) Medical data processing method, device, equipment and storage medium
WO2020010834A1 (en) Faq question and answer library generalization method, apparatus, and device
CN102663129A (en) Medical field deep question and answer method and medical retrieval system
CN105138864B (en) Protein interactive relation data base construction method based on Biomedical literature
CN112214335B (en) Web service discovery method based on knowledge graph and similarity network
WO2023178971A1 (en) Internet registration method, apparatus and device for seeking medical advice, and storage medium
CN111222330A (en) Chinese event detection method and system
Yan et al. A knowledge-driven generative model for multi-implication chinese medical procedure entity normalization
EP4030355A1 (en) Neural reasoning path retrieval for multi-hop text comprehension
Wang et al. An improved clustering method for detection system of public security events based on genetic algorithm and semisupervised learning
CN109144999B (en) Data positioning method, device, storage medium and program product
CN117235275A (en) Medical disease coding mapping method and device based on large language model reasoning

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