CN111259160A - Knowledge graph construction method, device, equipment and storage medium - Google Patents
Knowledge graph construction method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN111259160A CN111259160A CN201811456359.4A CN201811456359A CN111259160A CN 111259160 A CN111259160 A CN 111259160A CN 201811456359 A CN201811456359 A CN 201811456359A CN 111259160 A CN111259160 A CN 111259160A
- Authority
- CN
- China
- Prior art keywords
- entity
- text
- information
- event
- event information
- 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
Links
Images
Landscapes
- Machine Translation (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention discloses a knowledge graph construction method, a knowledge graph construction device, knowledge graph construction equipment and a storage medium. The method comprises the following steps: acquiring a text for constructing a knowledge graph; extracting entity information and event information corresponding to the entity from the text; and constructing a knowledge graph according to the information of the entity and the event information corresponding to the entity. The knowledge graph construction method provided by the embodiment of the invention not only shows the information of the entity, but also shows the event information corresponding to the entity, and has wide coverage; meanwhile, the method has universality, universality and expansibility.
Description
Technical Field
The embodiment of the invention relates to a data mining technology, in particular to a knowledge graph construction method, a knowledge graph construction device, knowledge graph construction equipment and a storage medium.
Background
The Knowledge map (also called scientific Knowledge map) is a Knowledge domain visualization or Knowledge domain mapping map in the book intelligence world, and is a series of different graphs for displaying the relationship between the Knowledge development process and the structure, describing Knowledge resources and carriers thereof by using a visualization technology, and mining, analyzing, constructing, drawing and displaying Knowledge and the mutual relation among the Knowledge resources and the carriers.
At present, knowledge maps appear in many application fields, and knowledge contents in the fields are displayed in a visual mode. However, most of the existing knowledge maps are used for simply describing the contents of organizations, personnel configurations and the like, and the information that can be displayed is very limited and has a narrow coverage.
Disclosure of Invention
The embodiment of the invention provides a method, a device, equipment and a storage medium for constructing a knowledge graph, which are used for expanding the coverage of the knowledge graph.
In a first aspect, an embodiment of the present invention provides a method for constructing a knowledge graph, including:
acquiring a text for constructing a knowledge graph;
extracting entity information and event information corresponding to the entity from the text;
and constructing a knowledge graph according to the information of the entity and the event information corresponding to the entity.
In a second aspect, an embodiment of the present invention further provides a knowledge graph constructing apparatus, including:
the acquisition module is used for acquiring a text for constructing a knowledge graph;
the extraction module is used for extracting the information of the entity and the event information corresponding to the entity from the text;
and the construction module is used for constructing the knowledge graph according to the information of the entity and the event information corresponding to the entity.
In a third aspect, an embodiment of the present invention further provides an electronic device, where the electronic device includes:
one or more processors;
a memory for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method of knowledge-graph construction of any of the embodiments.
In a fourth aspect, the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to implement the method for constructing a knowledge graph according to any one of the embodiments.
In the embodiment of the invention, the text for constructing the knowledge graph is obtained, and the information of the entity and the event information corresponding to the entity are extracted from the text, so that not only the information of the entity but also the event information corresponding to the entity are obtained; the knowledge graph is constructed according to the information of the entity and the event information corresponding to the entity, so that the knowledge graph not only shows the information of the entity, but also shows the event information corresponding to the entity, and the coverage is wide; meanwhile, the method provided by the embodiment of the invention has strong expansibility, and can be expanded to the construction of a knowledge graph based on a character carrier in the vertical field; moreover, the method provided by the embodiment has universality and universality, and is not limited by the application field and the application scene.
Drawings
FIG. 1a is a flow chart of a knowledge graph construction method according to an embodiment of the present invention;
fig. 1b is a schematic diagram of a webpage of a judicial literature according to an embodiment of the present invention;
FIG. 2a is a flowchart of a knowledge graph construction method according to a second embodiment of the present invention;
FIG. 2b is a sequence annotation diagram according to a second embodiment of the present invention;
FIG. 3a is a flowchart of a knowledge graph construction method according to a third embodiment of the present invention;
FIG. 3b is a schematic diagram of extracting the property information of the residence and the legal representative through the rule template according to the third embodiment of the present invention;
FIG. 4a is a flowchart of a knowledge graph construction method according to a fourth embodiment of the present invention;
fig. 4b is a schematic diagram of a civil judgment book with a current appeal category of "no appeal review" according to the fourth embodiment of the present invention;
FIG. 4c is a schematic diagram of a knowledge-graph according to a fourth embodiment of the present invention;
FIG. 5 is a schematic structural diagram of a knowledge graph constructing apparatus according to a fifth embodiment of the present invention;
fig. 6 is a schematic structural diagram of an electronic device according to a sixth embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Example one
Fig. 1a is a flowchart of a method for constructing a knowledge graph according to an embodiment of the present invention, where this embodiment is applicable to a graph construction situation using text as a carrier in the vertical domain, and this method may be performed by a knowledge graph constructing apparatus, which may be composed of software and/or hardware and is generally integrated in an electronic device. With reference to fig. 1a, the method for constructing a knowledge graph provided by the embodiment of the present invention specifically includes the following operations:
and S110, acquiring a text for constructing the knowledge graph.
Optionally, a document for constructing the knowledge graph is acquired from an open channel or an internal channel, and the document is parsed to obtain a text for constructing the knowledge graph.
In an example, fig. 1b is a schematic diagram of a webpage of a judicial literature provided in an embodiment of the present invention. The webpage divides the judicial literature into five parts, namely a head part, a fact, a reason, a judgment result and a tail part, and then the five parts of texts are analyzed according to the splitting mode of the webpage.
In another example, a formal document typically has a standard format, such as that shown by the judicial document in FIG. 1b, and such as that of a review comment notice, etc. Therefore, the text of the knowledge graph constructed by the user is analyzed according to the document format. For example, the document title, case type, auditor, case occurrence time, case title, auditor court, party, document number, etc. are parsed from the judicial document.
To make the scope of knowledge-graph coverage broader, multiple types of text are optionally acquired. For example, in an application scene of constructing a judicial knowledge graph for an enterprise, six types of judicial documents including criminal cases, civil cases, executive cases, intellectual property cases, administrative cases and the like are obtained, and texts of corresponding types are analyzed from the obtained documents, so that the knowledge graph is constructed, integral graph data display is carried out on the judicial aspect of the enterprise, and a very good data basis is provided for subsequent risk analysis.
And S120, extracting entity information and event information corresponding to the entity from the text.
Most of the texts obtained in S110 are unstructured, and this operation mainly uses the already parsed texts to further obtain more informative entities and event information corresponding to the entities. Optionally, first extracting entity information from the text; and extracting event information corresponding to the entity from the text according to the information of the entity.
In this embodiment, the entities include enterprises and/or individuals, and the text may include information of at least one entity.
In the application scenario of constructing the judicial knowledge graph for the enterprise, the information of the entity in the text comprises the enterprise and other parties, and the other parties can be other enterprises or individuals. The event information corresponding to the entity refers to the event information related to judicial affairs described in the document, such as indemnification event information, trial results and the like.
And S130, constructing a knowledge graph according to the information of the entity and the event information corresponding to the entity.
Optionally, the information of the entity is drawn as nodes, and the event information corresponding to the entity is drawn as a connection line or a label between the nodes to form a knowledge graph.
In the embodiment of the invention, the text for constructing the knowledge graph is obtained, and the information of the entity and the event information corresponding to the entity are extracted from the text, so that not only the information of the entity but also the event information corresponding to the entity are obtained; the knowledge graph is constructed according to the information of the entity and the event information corresponding to the entity, so that the knowledge graph not only shows the information of the entity, but also shows the event information corresponding to the entity, and the coverage is wide; meanwhile, the method provided by the embodiment of the invention has strong expansibility, and can be expanded to the construction of a knowledge graph based on a character carrier in the vertical field; moreover, the method provided by the embodiment has universality and universality, and is not limited by the application field and the application scene.
Example two
Fig. 2a is a flowchart of a knowledge graph construction method according to a second embodiment of the present invention. In this embodiment, each optional implementation manner of the foregoing embodiment is further optimized, and optionally, "extract information of an entity from a text" is optimized to "extract a name of an entity from a text; and extracting role information corresponding to the name of the entity from the text according to the name of the entity, thereby providing the map display of the role dimension of the entity. With reference to fig. 2a, the method provided by the embodiment of the present invention includes the following operations:
and S210, acquiring a text for constructing the knowledge graph.
And S220, extracting the name of the entity from the text.
The name of the entity includes an enterprise name and/or a person name, depending on the entity.
Optionally, the name of the entity is identified by using a named entity identification method, where named entity identification (NER), also called "proper name identification", refers to identifying an entity having a specific meaning in a text, and mainly includes a person name, a place name, an organization name, a proper noun, and the like. Further, in most cases, the document title includes the name of the entity, and the name of the entity is extracted from the document title.
In one example, enterprise name identification and person name identification use the BilSTM + CRF algorithm, and the BilSTM network can take advantage of the two-way, good effect on word segmentation. The CRF algorithm can obtain the output probability of another sequence when a group of input sequences is given, and the BiLSTM + CRF algorithm combined with the two has good effect on the identification of named entities related to NLP. And performing BIOE labeling on the text title, wherein B is the starting position of the entity sequence, I is the internal position, E is the ending position, and O represents others, and performing sequence labeling on the text title by utilizing the memory of the LSTM network structure and combining CRF.
And then, carrying out sample training through a labeling data set, obtaining the label of the prediction sequence, and taking the word between BE labels as an entity so as to correctly identify the name of the entity. Sequence notation As shown in FIG. 2B, the name of the entity "XX, XX city XXXX, Inc" is obtained by the combination of B and E.
And S230, extracting role information corresponding to the name of the entity from the text according to the name of the entity.
In an optional embodiment, after the name of the entity is extracted, the context of the name of the entity is extracted, and the role information corresponding to the name of the entity is extracted from the context.
In another alternative embodiment, a paragraph in which the role information is located, for example, a principal paragraph shown in fig. 3b, is located, and the role information corresponding to the name of the entity is extracted from the paragraph in which the role information is located.
In yet another alternative embodiment, the role information corresponding to the name of the entity is extracted from the entire text.
Alternatively, S230 includes the following two steps:
the first step is as follows: and performing text analysis on the text to obtain keywords of the role information in different events.
Optionally, the words of the text are cut according to a self-defined dictionary and stop words to obtain a plurality of participles. Further, the dictionary and stop words are further adjusted according to the word segmentation effect. And then, carrying out word frequency statistics on the multiple word segments to obtain keywords of the role information in different events. For example, the keywords of the role information "original report and reported report" in different types of cases include: an executives, guarantors, postings, prior review postings, appetitives, etc.
The second step is that: and identifying role information corresponding to the name of the entity according to the name of the entity and the rule template of the keyword.
The rule template of the name and keyword of the entity is, for example, "keyword: name of entity ". And carrying out template matching on the text through a rule template to obtain a keyword corresponding to the name of the entity, and further obtaining role information of the keyword in different events. With reference to fig. 3b, the keywords corresponding to the names of the entities are "first-censored notice, original review applicant, and second-censored appetizer", and the role information corresponding to the keywords is the notice, the review applicant, and the second-censored appetizer.
And S240, extracting event information corresponding to the entity from the text.
And S250, constructing a knowledge graph according to the name of the entity, the role information corresponding to the name of the entity and the event information corresponding to the entity.
Optionally, the information of the entity is drawn as nodes, and the event information corresponding to the entity is drawn as a connection line or a label between the nodes; meanwhile, the role information of the entity is marked around the entity to form a knowledge graph.
In the embodiment, the name of the entity is extracted from the text; extracting role information corresponding to the name of the entity from the text according to the name of the entity, so as to deeply extract the role information of the entity in the event and provide map display of the role dimension of the entity; performing text analysis on the text to obtain keywords of the role information in different events; identifying role information corresponding to the name of the entity according to the name of the entity and a rule template of the keyword, and accurately and efficiently extracting the role information by a method of matching the keyword with the template; moreover, the role information is specific to a certain event type, and the role information is displayed in the knowledge graph, so that the characteristics and the speciality of different events are reserved when the entity overall view is displayed in a combing mode.
EXAMPLE III
Fig. 3a is a flowchart of a method for constructing a knowledge graph according to a third embodiment of the present invention. In this embodiment, each optional implementation manner of the foregoing embodiment is further optimized, and optionally, "information of an entity extracted from a text" is optimized to "attribute information of the entity is identified from the text according to a rule template of the attribute information," so that a map display of attribute dimensions of the entity is provided. With reference to fig. 3a, the method provided by the embodiment of the present invention includes the following operations:
and S310, acquiring a text for constructing the knowledge graph.
And S320, identifying the attribute information of the entity from the text according to the rule template of the attribute information.
In an alternative embodiment, the paragraph in which the attribute information is located, such as the principal paragraph shown in fig. 3b, is located, and the attribute information of the entity is extracted from the paragraph in which the attribute information is located.
In another alternative embodiment, the attribute information of the entity is extracted from the entire text.
In yet another alternative embodiment, the name of the entity is first extracted. After the name of the entity is extracted, the context of the name of the entity is extracted, and attribute information corresponding to the name of the entity is extracted from the context.
Optionally, the attribute information of the individual includes information of a residence, an identification number, a gender, a occupation, and the like. Attribute information for an enterprise includes legal representatives, locations, etc.
The rule template of the attribute information is, for example, "place of residence", "legal representative: *". And carrying out template matching on the text through a rule template to obtain attribute information of the entity. With reference to fig. 3b, for the principal paragraph, the attribute information of the residence and the legal representative is extracted through the rule template.
It should be noted that in some cases, the attribute information of the enterprise and public institution in the text is not described much, even only the name of the enterprise and public institution. Optionally, after identifying the attribute information of the enterprise and public institution from the text, searching the corresponding business information according to the name of the enterprise and public institution, and perfecting the attribute information of the enterprise and public institution.
S330, extracting event information corresponding to the entity from the text.
And S340, constructing a knowledge graph according to the attribute information of the entity and the event information corresponding to the entity.
Optionally, the information of the entity is drawn as nodes, and the event information corresponding to the entity is drawn as a connection line or a label between the nodes; meanwhile, attribute information of the entity is marked around the entity to form a knowledge graph.
In this embodiment, the attribute information of the entity is identified from the text according to the rule template of the attribute information, so that the attribute information of the entity in the event is extracted in depth, and the attribute dimension map display of the entity is provided.
Example four
Fig. 4a is a flowchart of a knowledge graph construction method according to a fourth embodiment of the present invention. The present embodiment is further optimized for each optional implementation manner of the foregoing embodiment, and optionally, the event information at least includes one of current appeal event information, indemnity event information, trial result, event occurrence location, event progress stage, and hierarchy information of the event processing mechanism. Optionally, before the knowledge graph is constructed according to the information of the entity and the event information corresponding to the entity, an additional operation of extracting keywords and an abstract from the text is performed, and further optionally, the construction of the knowledge graph according to the information of the entity and the event information corresponding to the entity is optimized to the construction of the knowledge graph according to the information of the entity, the event information corresponding to the entity, and the keywords and the abstract extracted from the text. With reference to fig. 4a, the method provided by the embodiment of the present invention includes the following operations:
and S410, acquiring a text for constructing the knowledge graph.
And S420, extracting entity information and event information corresponding to the entity from the text, wherein the event information at least comprises one of current appeal event information, indemnity event information, trial result, event occurrence place, event progress stage and event processing mechanism level information.
The following describes the extraction process of event information in detail for each type of event information.
For the current appeal event information, optionally, first, keywords and templates corresponding to multiple current appeal categories are obtained, for example, the current appeal categories covering all documents in the judicial field include applying for withdrawal, applying for execution, setting up a case for acceptance, not judging to go to prose, proposing a debating opinion, and the like, and different templates and keywords corresponding to different current appeal categories. Fig. 4b shows a civil decision book with the current appeal category "do not obey requisition", corresponding to a template such as "do not obey", "apply to the home court to reexamine", and a keyword such as "reexamine". Next, according to the keywords corresponding to the various current appeal categories, the sentence drop stating the current appeal is positioned, according to the templates corresponding to the various current appeal categories, template matching is performed on the sentence drop stating the current appeal, current appeal event information is extracted, namely the current appeal event information is matched with the template corresponding to which current appeal category, extraction is performed according to the matched template, and for example, the content underlined in fig. 4b is extracted.
For the indemnity event information, optionally, first, performing word segmentation and part-of-speech analysis on the text to obtain an indemnity item list in which nouns and nouns are added. For example, the words of the text "disability claims and reimbursements" are segmented, stop words are filtered, and part-of-speech analysis is performed, wherein the part-of-speech after segmentation is as follows: "[" disability "," n "], [" indemnity "," n "]", wherein "n" represents a noun attribute. The expense name obtained by analysis is the combination of the addition of the noun and the noun, namely the disability and the indemnity can be combined to obtain the indemnity of the disability indemnity, the operation basically achieves 98% of accuracy, and then the expense money can be correctly obtained by manual screening. Further, the frequency of each of the compensation items appearing in the plurality of texts is counted, and the compensation items with the frequency of appearance greater than M are retained in the list of the compensation items, so as to remove the wrong compensation items, for example, M is 2000, 3000, etc. Then, based on the normative of the text, a regular expression is superimposed on the claim compensation list, and the regular expression is used for matching the figures after the claim compensation, so as to obtain the claim compensation and the corresponding claim compensation amount, for example, the medical fee: 39,138.62 yuan, the hospitalization food subsidy fee: 2,200.00 yuan, etc.
For the trial results, optionally, first, keywords corresponding to a plurality of trial result categories are obtained, for example, keywords such as the following judgment, the following adjudication, the following protocol, the following execution, and the like. And then, positioning sentence sets which state the trial results according to the keywords which respectively correspond to the various types of the trial results. Then, from the sentence drop stating the trial result, the key sentence corresponding to the trial result is screened out, optionally, matching is performed according to the rule template "rejections" and "maintenance" of the trial result, and the key sentence is screened out, for example, the key sentence is screened out: "other litigation requests of somebody refuted against original plum, and somebody advised of the defendant against the original plum. And then, performing word segmentation and part-of-speech analysis on the key sentence to obtain information of the dynamic name word combination and the entity, and obtaining a judgment result corresponding to the entity according to the information of the dynamic name word combination and the entity. For example, the information of the entity includes information of a certain li and an original role, and if the combination of the action names is a rejectional original, the corresponding trial result of the certain li is rejectional. In some cases, when the action-name phrase is the refusal of a certain plum person, the corresponding trial result of the certain plum person is directly obtained as refusal.
For the event occurrence location, optionally, a plurality of preset locations are first acquired, for example, information of provinces across the country is acquired. Then, word segmentation is carried out on the text to obtain a plurality of word segments; and matching the multiple word segmentations in multiple places to obtain matched event occurrence places. For example, the segmentation is sequentially matched with three dimensions of national province, city and district, and information of the other two dimensions (for example, province and city) is further deduced according to the result of one dimension (for example, district) of matching.
Optionally, keywords corresponding to different progress stages of the event are obtained for the progress stages of the event. Taking the judicial field as an example, the keywords corresponding to different stages of progress include: final review decision, first review decision, etc. In general, keywords corresponding to different stages of progress will appear in the title, and possibly in the content. Therefore, according to the keywords corresponding to different progress stages of the event, the progress stage of the event corresponding to the entity is identified from the title and/or the content of the text.
Extracting the name of the event processing mechanism from the text optionally according to the hierarchy information of the event processing mechanism; hierarchy information corresponding to a name of an event handling mechanism is acquired. The name of the event processing mechanism is generally structured, and the event processing mechanism can be directly extracted without carrying out structured operation. In general, if the hierarchical information of the event processing means is not included in the text, the hierarchical information corresponding to the name of the event processing means can be collected from the network. Taking the judicial field as an example, the hierarchical information of the event processing agency includes the first, middle and high levels of the court.
And S430, extracting keywords and abstracts from the text.
Optionally, keywords and abstracts are extracted from the text using a TextRank algorithm, which is based on PageRank and is used to generate keywords and abstracts for the text.
S440, constructing a knowledge graph according to the information of the entity, the event information corresponding to the entity, and the keywords and the abstract extracted from the text.
Optionally, the information of the entity is drawn as nodes, and the event information corresponding to the entity is drawn as a connection line or a label between the nodes; meanwhile, the keywords and the abstracts extracted from the text are marked around the corresponding entities to form a knowledge graph.
In the embodiment of the invention, the event information of the current appeal is extracted, the event key points of the multiple document complaints of one case are extracted, the time sequence level display of the event development situation is provided, and the case change of one case is displayed; by extracting the indemnity event information, the atlas display of a statistical level is provided, and the indemnity event information reflects the case-involved severity of the entity from the side; by extracting the judging result, the comprehensive strength of the entity is shown, and the business and operation conditions of the entity are influenced; the method helps to widen the entity overall appearance, such as the judicial overall appearance of an enterprise, by extracting the hierarchical information of the event occurrence place, the event progress stage and the event processing mechanism; by extracting the key words and the abstract, the analysis, the comparison and the overall grasp of the entity information and the events are facilitated.
In each embodiment, at least one of a relationship between entities, a time sequence relationship of events, and an entity risk assessment is determined according to information of the entities and event information corresponding to the entities; and constructing a knowledge graph according to at least one of the relationship among the entities, the time sequence relationship of the events and the risk assessment of the entities.
Optionally, the relationship between the entities and the time sequence relationship of the events are determined according to the information of the entities and the event information corresponding to the entities. With the above embodiments, in the judicial field, the appellation relationship between entities and the time sequence relationship of cases are obtained by combining the role information of the entities and the case development (current appeal). Optionally, the entity risk assessment is performed according to the information of the entity and the event information corresponding to the entity. By combining the embodiments, in the judicial field, information such as case types, role information, case starting principles, case development (current appeal), annual case times, compensation amount, case keyword distribution and the like is integrated, and the judicial overall view of an enterprise is visually displayed; and simultaneously, applying the information to the risk monitoring construction of the enterprise to obtain the enterprise risk assessment.
Optionally, after obtaining the relationship between the entities, the time sequence relationship of the event, and the risk assessment of the entities, drawing a connecting line between the entities according to the relationship between the entities and the time sequence relationship of the event, representing the relationship between the entities by using an arrow, and labeling the time sequence relationship of the event; and simultaneously marking enterprise risk assessment around the entity. Fig. 4c is a schematic diagram of a knowledge graph according to a fourth embodiment of the present invention, where fig. 4c shows a relationship between a plurality of entities, where an entity role pointed by an arrow is a defendant, and an entity role pointed by the arrow is a defendant.
Taking the application to the judicial field as an example, the knowledge graph construction method provided by the embodiments of the invention can achieve the following technical effects:
firstly, the judicial atlas construction provided by the embodiment of the invention contains six types of judicial cases, including criminal cases, civil cases, executive cases, intellectual property cases, administrative cases and others, namely, the common level of the six types of judicial cases, such as the level of 'case progress stage, court level, compensation amount' and the like, can be considered, the characteristics of each case type can be grasped at a slight place, the characteristics of each large type of case are kept, the extraction of 'current appeal', namely the division of current appeal categories such as 'non-complaint, application withdrawal' and the like with universality, and the category of 'putting forward criminal reduction opinions' specific to the criminal cases is also taken into consideration. By the method, when the corporate judicial landscape is displayed in a combing mode, the characteristics and the specialties of different types of cases are still maintained by the map.
Then, the embodiment of the invention is used as a core module of the enterprise related intelligent project, and the main innovation after the judicial atlas is constructed is to provide the 'comprehensive' risk monitoring of the enterprise in the judicial field. According to fields such as 'court level, indemnity amount' and the like obtained by extraction, case quantity and indemnity amount are counted based on dimensions such as enterprise industry, region, time and the like, and then quantification of enterprise judicial risks is obtained, and analysis and evaluation of the enterprise in all aspects are obtained in an auxiliary manner by combining public opinion monitoring and industrial and commercial characteristics.
In addition, the judicial construction process of the embodiment of the invention relates to the general event extraction, named entity identification and other related machine learning and deep learning technologies, the applied rule template is simple without losing the accuracy and efficiency, and the construction thought of the whole judicial atlas has universality and popularization.
Finally, the method provided by the invention has strong expansibility, and can be easily expanded to other similar scenes, especially map construction in the professional field by using characters as carriers, such as 'patents, bidding documents' and the like.
EXAMPLE five
Fig. 5 is a schematic structural diagram of a knowledge graph constructing apparatus according to a fifth embodiment of the present invention, and the fifth embodiment of the present invention is suitable for a graph constructing situation in a vertical field using characters as carriers. With reference to fig. 5, the apparatus specifically includes: an acquisition module 51, an extraction module 52 and a construction module 53.
An obtaining module 51, configured to obtain a text for constructing a knowledge graph;
an extracting module 52, configured to extract information of the entity and event information corresponding to the entity from the text;
and the constructing module 53 is configured to construct a knowledge graph according to the information of the entity and the event information corresponding to the entity.
In the embodiment of the invention, the text for constructing the knowledge graph is obtained, and the information of the entity and the event information corresponding to the entity are extracted from the text, so that not only the information of the entity but also the event information corresponding to the entity are obtained; the knowledge graph is constructed according to the information of the entity and the event information corresponding to the entity, so that the knowledge graph not only shows the information of the entity, but also shows the event information corresponding to the entity, and the coverage is wide; meanwhile, the method provided by the embodiment of the invention has strong expansibility, and can be expanded to the construction of a knowledge graph based on a character carrier in the vertical field; moreover, the method provided by the embodiment has universality and universality, and is not limited by the application field and the application scene.
Optionally, when extracting the entity information from the text, the extracting module 52 is specifically configured to: extracting the name of the entity from the text; and extracting role information corresponding to the name of the entity from the text according to the name of the entity.
Optionally, when extracting the role information corresponding to the name of the entity from the text according to the name of the entity, the extraction module 52 is specifically configured to: performing text analysis on the text to obtain keywords of the role information in different events; and identifying role information corresponding to the name of the entity according to the name of the entity and the rule template of the keyword.
Optionally, when extracting the entity information from the text, the extracting module 52 is specifically configured to: and identifying the attribute information of the entity from the text according to the rule template of the attribute information.
Optionally, the event information includes: current appeal event information; when extracting the event information corresponding to the entity from the text, the extracting module 52 is specifically configured to: acquiring keywords and templates corresponding to various current appeal categories respectively; according to the keywords respectively corresponding to the various current appeal categories, positioning and stating the sentence drop of the current appeal; and according to the templates corresponding to the various current appeal categories, template matching is carried out on the sentence drops stating the current appeal, and current appeal event information is extracted.
Optionally, the event information includes: indemnity event information; when extracting the event information corresponding to the entity from the text, the extracting module 52 is specifically configured to: performing word segmentation and part-of-speech analysis on the text to obtain a list of compensation terms formed by adding nouns and nouns; and superposing the regular expression on the compensation item list to obtain the compensation items and the corresponding compensation amount.
Optionally, the event information includes: judging results; when extracting the event information corresponding to the entity from the text, the extracting module 52 is specifically configured to: obtaining keywords respectively corresponding to various judging result types; positioning sentence drops for stating the judging results according to keywords respectively corresponding to the various judging result categories; screening out key sentences corresponding to the trial results from the sentence drops which state the trial results; performing word segmentation and part-of-speech analysis on the key sentences to obtain information of dynamic name word combinations and entities; and obtaining a judging result corresponding to the entity according to the mobile name word combination and the information of the entity.
Optionally, the event information includes: a location of occurrence of the event; when extracting the event information corresponding to the entity from the text, the extracting module 52 is specifically configured to: acquiring a plurality of preset places; performing word segmentation on the text to obtain a plurality of word segments; and matching the multiple word segmentations in multiple places to obtain matched event occurrence places.
Optionally, the event information includes: an event progression phase; when extracting the event information corresponding to the entity from the text, the extracting module 52 is specifically configured to: acquiring keywords corresponding to different progress stages of an event; and identifying the event progress stages corresponding to the entities from the titles and/or contents of the texts according to the keywords corresponding to the different progress stages of the events.
Optionally, the event information includes: hierarchy information of the event handling mechanism; when extracting the event information corresponding to the entity from the text, the extracting module 52 is specifically configured to: extracting the name of the event processing mechanism from the text; hierarchy information corresponding to a name of an event handling mechanism is acquired.
Optionally, the apparatus further includes a keyword and abstract extraction module, configured to extract keywords and abstract from the text before constructing the knowledge graph according to the information of the entity and the event information corresponding to the entity. Based on this, when constructing the knowledge graph according to the information of the entity and the event information corresponding to the entity, the constructing module 53 is specifically configured to: and constructing a knowledge graph according to the information of the entity, the event information corresponding to the entity, and the keywords and the abstract extracted from the text.
Optionally, when the construction module 53 constructs the knowledge graph according to the information of the entity and the event information corresponding to the entity, it is specifically configured to: determining at least one of a relationship between entities, a time sequence relationship of events and entity risk assessment according to the information of the entities and event information corresponding to the entities; and constructing a knowledge graph according to at least one of the relationship among the entities, the time sequence relationship of the events and the risk assessment of the entities.
The knowledge graph constructing device provided by the embodiment of the invention can execute the knowledge graph constructing method provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the executing method.
EXAMPLE six
Fig. 6 is a schematic structural diagram of an electronic device according to a sixth embodiment of the present invention. FIG. 6 illustrates a block diagram of an exemplary electronic device 12 suitable for use in implementing embodiments of the present invention. The electronic device 12 shown in fig. 6 is only an example and should not bring any limitation to the function and the scope of use of the embodiment of the present invention.
As shown in FIG. 6, electronic device 12 is embodied in the form of a general purpose computing device. The components of electronic device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including the system memory 28 and the processing unit 16.
The system memory 28 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM)30 and/or cache memory 32. The electronic device 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 34 may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 6, and commonly referred to as a "hard drive"). Although not shown in FIG. 6, a magnetic disk drive for reading from and writing to a removable, nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 by one or more data media interfaces. Memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the invention.
A program/utility 40 having a set (at least one) of program modules 42 may be stored, for example, in memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may comprise an implementation of a network environment. Program modules 42 generally carry out the functions and/or methodologies of the described embodiments of the invention.
The processing unit 16 executes various functional applications and data processing, such as implementing the knowledge graph construction method provided by the embodiments of the present invention, by executing programs stored in the system memory 28.
EXAMPLE seven
The seventh embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method for constructing a knowledge graph according to any embodiment of the present invention. The knowledge graph construction method comprises the following steps: acquiring a text for constructing a knowledge graph; extracting entity information and event information corresponding to the entity from the text; and constructing a knowledge graph according to the information of the entity and the event information corresponding to the entity.
Computer storage media for embodiments of the invention may employ any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + +, or the like, as well as conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments illustrated herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (15)
1. A knowledge graph construction method is characterized by comprising the following steps:
acquiring a text for constructing a knowledge graph;
extracting entity information and event information corresponding to the entity from the text;
and constructing a knowledge graph according to the information of the entity and the event information corresponding to the entity.
2. The method of claim 1, wherein extracting entity information from the text comprises:
extracting the name of the entity from the text;
and extracting role information corresponding to the name of the entity from the text according to the name of the entity.
3. The method of claim 2, wherein extracting role information corresponding to the name of the entity from the text according to the name of the entity comprises:
performing text analysis on the text to obtain keywords of the role information in different events;
and identifying role information corresponding to the name of the entity according to the name of the entity and the rule template of the keyword.
4. The method of claim 1, wherein extracting entity information from the text comprises:
and identifying the attribute information of the entity from the text according to the rule template of the attribute information.
5. The method of claim 1, wherein the event information comprises: current appeal event information;
the extracting event information corresponding to the entity from the text includes:
acquiring keywords and templates corresponding to various current appeal categories respectively;
according to the keywords respectively corresponding to the various current appeal categories, positioning and stating the sentence drop of the current appeal;
and according to the templates corresponding to the various current appeal categories, template matching is carried out on the sentence drops stating the current appeal, and current appeal event information is extracted.
6. The method of claim 1, wherein the event information comprises: indemnity event information;
the extracting event information corresponding to the entity from the text includes:
performing word segmentation and part-of-speech analysis on the text to obtain a list of compensation terms formed by adding nouns and nouns;
and superposing the regular expression on the compensation item list to obtain the compensation items and the corresponding compensation amount.
7. The method of claim 1, wherein the event information comprises: judging results;
the extracting event information corresponding to the entity from the text includes:
obtaining keywords respectively corresponding to various judging result types;
positioning sentence drops for stating the judging results according to keywords respectively corresponding to the various judging result categories;
screening out key sentences corresponding to the trial results from the sentence drops which state the trial results;
performing word segmentation and part-of-speech analysis on the key sentences to obtain information of dynamic name word combinations and entities;
and obtaining a judging result corresponding to the entity according to the mobile name word combination and the information of the entity.
8. The method of claim 1, wherein the event information comprises: a location of occurrence of the event;
the extracting event information corresponding to the entity from the text includes:
acquiring a plurality of preset places;
performing word segmentation on the text to obtain a plurality of word segments;
and matching the multiple word segmentations in multiple places to obtain matched event occurrence places.
9. The method of claim 1, wherein the event information comprises: an event progression phase;
the extracting event information corresponding to the entity from the text includes:
acquiring keywords corresponding to different progress stages of an event;
and identifying the event progress stages corresponding to the entities from the titles and/or contents of the texts according to the keywords corresponding to the different progress stages of the events.
10. The method of claim 1, wherein the event information comprises: hierarchy information of the event handling mechanism;
the extracting event information corresponding to the entity from the text includes:
extracting the name of the event processing mechanism from the text;
hierarchy information corresponding to a name of an event handling mechanism is acquired.
11. The method of claim 1, further comprising, before constructing the knowledge graph according to the information of the entity and the event information corresponding to the entity:
extracting key words and abstracts from the text;
the constructing of the knowledge graph according to the information of the entity and the event information corresponding to the entity comprises the following steps:
and constructing a knowledge graph according to the information of the entity, the event information corresponding to the entity, and the keywords and the abstract extracted from the text.
12. The method according to any one of claims 1 to 11, wherein the constructing a knowledge graph according to the information of the entity and the event information corresponding to the entity comprises:
determining at least one of a relationship between entities, a time sequence relationship of events and entity risk assessment according to the information of the entities and event information corresponding to the entities;
and constructing a knowledge graph according to at least one of the relationship among the entities, the time sequence relationship of the events and the risk assessment of the entities.
13. A knowledge-graph building apparatus, comprising:
the acquisition module is used for acquiring a text for constructing a knowledge graph;
the extraction module is used for extracting the information of the entity and the event information corresponding to the entity from the text;
and the construction module is used for constructing the knowledge graph according to the information of the entity and the event information corresponding to the entity.
14. An electronic device, characterized in that the electronic device comprises:
one or more processors;
a memory for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method of knowledge-graph construction as claimed in any one of claims 1-12.
15. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out a method of constructing a knowledge graph according to any one of claims 1-12.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811456359.4A CN111259160B (en) | 2018-11-30 | 2018-11-30 | Knowledge graph construction method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811456359.4A CN111259160B (en) | 2018-11-30 | 2018-11-30 | Knowledge graph construction method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111259160A true CN111259160A (en) | 2020-06-09 |
| CN111259160B CN111259160B (en) | 2023-08-29 |
Family
ID=70953875
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811456359.4A Active CN111259160B (en) | 2018-11-30 | 2018-11-30 | Knowledge graph construction method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111259160B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112486919A (en) * | 2020-11-13 | 2021-03-12 | 北京北大千方科技有限公司 | Document management method, system and storage medium |
| CN112528660A (en) * | 2020-12-04 | 2021-03-19 | 北京百度网讯科技有限公司 | Method, apparatus, device, storage medium and program product for processing text |
| CN112632223A (en) * | 2020-12-29 | 2021-04-09 | 天津汇智星源信息技术有限公司 | Case and event knowledge graph construction method and related equipment |
| CN113010684A (en) * | 2020-12-31 | 2021-06-22 | 北京法意科技有限公司 | Construction method and system of civil complaint and judgment map |
| CN115269879A (en) * | 2022-09-05 | 2022-11-01 | 北京百度网讯科技有限公司 | Knowledge structure data generation method, data search method and risk warning method |
| CN115730078A (en) * | 2022-11-04 | 2023-03-03 | 南京擎盾信息科技有限公司 | Event knowledge graph construction method and device for class case retrieval and electronic equipment |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050071217A1 (en) * | 2003-09-30 | 2005-03-31 | General Electric Company | Method, system and computer product for analyzing business risk using event information extracted from natural language sources |
| CN106156365A (en) * | 2016-08-03 | 2016-11-23 | 北京智能管家科技有限公司 | A kind of generation method and device of knowledge mapping |
| WO2016196320A1 (en) * | 2015-05-29 | 2016-12-08 | Microsoft Technology Licensing, Llc | Language modeling for speech recognition leveraging knowledge graph |
| CN106991092A (en) * | 2016-01-20 | 2017-07-28 | 阿里巴巴集团控股有限公司 | The method and apparatus that similar judgement document is excavated based on big data |
| CN107122444A (en) * | 2017-04-24 | 2017-09-01 | 北京科技大学 | A kind of legal knowledge collection of illustrative plates method for auto constructing |
| CN107633044A (en) * | 2017-09-14 | 2018-01-26 | 国家计算机网络与信息安全管理中心 | A kind of public sentiment knowledge mapping construction method based on focus incident |
| CN108153901A (en) * | 2018-01-16 | 2018-06-12 | 北京百度网讯科技有限公司 | The information-pushing method and device of knowledge based collection of illustrative plates |
| CN108596439A (en) * | 2018-03-29 | 2018-09-28 | 北京中兴通网络科技股份有限公司 | A kind of the business risk prediction technique and system of knowledge based collection of illustrative plates |
-
2018
- 2018-11-30 CN CN201811456359.4A patent/CN111259160B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050071217A1 (en) * | 2003-09-30 | 2005-03-31 | General Electric Company | Method, system and computer product for analyzing business risk using event information extracted from natural language sources |
| WO2016196320A1 (en) * | 2015-05-29 | 2016-12-08 | Microsoft Technology Licensing, Llc | Language modeling for speech recognition leveraging knowledge graph |
| CN106991092A (en) * | 2016-01-20 | 2017-07-28 | 阿里巴巴集团控股有限公司 | The method and apparatus that similar judgement document is excavated based on big data |
| CN106156365A (en) * | 2016-08-03 | 2016-11-23 | 北京智能管家科技有限公司 | A kind of generation method and device of knowledge mapping |
| CN107122444A (en) * | 2017-04-24 | 2017-09-01 | 北京科技大学 | A kind of legal knowledge collection of illustrative plates method for auto constructing |
| CN107633044A (en) * | 2017-09-14 | 2018-01-26 | 国家计算机网络与信息安全管理中心 | A kind of public sentiment knowledge mapping construction method based on focus incident |
| CN108153901A (en) * | 2018-01-16 | 2018-06-12 | 北京百度网讯科技有限公司 | The information-pushing method and device of knowledge based collection of illustrative plates |
| CN108596439A (en) * | 2018-03-29 | 2018-09-28 | 北京中兴通网络科技股份有限公司 | A kind of the business risk prediction technique and system of knowledge based collection of illustrative plates |
Non-Patent Citations (1)
| Title |
|---|
| 郑好: "基于微博中的人物图谱的构建方法研究", 《优秀硕士论文》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112486919A (en) * | 2020-11-13 | 2021-03-12 | 北京北大千方科技有限公司 | Document management method, system and storage medium |
| CN112528660A (en) * | 2020-12-04 | 2021-03-19 | 北京百度网讯科技有限公司 | Method, apparatus, device, storage medium and program product for processing text |
| CN112528660B (en) * | 2020-12-04 | 2023-10-24 | 北京百度网讯科技有限公司 | Method, apparatus, device, storage medium and program product for processing text |
| CN112632223A (en) * | 2020-12-29 | 2021-04-09 | 天津汇智星源信息技术有限公司 | Case and event knowledge graph construction method and related equipment |
| CN112632223B (en) * | 2020-12-29 | 2023-01-20 | 天津汇智星源信息技术有限公司 | Case and event knowledge graph construction method and related equipment |
| CN113010684A (en) * | 2020-12-31 | 2021-06-22 | 北京法意科技有限公司 | Construction method and system of civil complaint and judgment map |
| CN113010684B (en) * | 2020-12-31 | 2024-02-09 | 北京法意科技有限公司 | Construction method and system of civil complaint judging map |
| CN115269879A (en) * | 2022-09-05 | 2022-11-01 | 北京百度网讯科技有限公司 | Knowledge structure data generation method, data search method and risk warning method |
| CN115730078A (en) * | 2022-11-04 | 2023-03-03 | 南京擎盾信息科技有限公司 | Event knowledge graph construction method and device for class case retrieval and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111259160B (en) | 2023-08-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109189942B (en) | Construction method and device of patent data knowledge graph | |
| CN111259160A (en) | Knowledge graph construction method, device, equipment and storage medium | |
| US7933843B1 (en) | Media-based computational influencer network analysis | |
| US9881037B2 (en) | Method for systematic mass normalization of titles | |
| US9922383B2 (en) | Patent claims analysis system and method | |
| CN111597351A (en) | Visual document map construction method | |
| CN113807098A (en) | Model training method and device, electronic equipment and storage medium | |
| US20170075983A1 (en) | Subject-matter analysis of tabular data | |
| US11163806B2 (en) | Obtaining candidates for a relationship type and its label | |
| CN108932218B (en) | Instance extension method, device, equipment and medium | |
| US10282467B2 (en) | Mining product aspects from opinion text | |
| Al Qundus et al. | Exploring the impact of short-text complexity and structure on its quality in social media | |
| CN112836018A (en) | Method and device for processing emergency plan | |
| Vadapalli et al. | Twitterosint: automated cybersecurity threat intelligence collection and analysis using twitter data | |
| CN114692628A (en) | Sample generation method, model training method, text extraction method and text extraction device | |
| Yang et al. | Tour: Dynamic topic and sentiment analysis of user reviews for assisting app release | |
| Rosa et al. | A visual approach for identification and annotation of business process elements in process descriptions | |
| CN111177401A (en) | Power grid free text knowledge extraction method | |
| CN110737770B (en) | Text data sensitivity identification method and device, electronic equipment and storage medium | |
| Pham et al. | Information extraction for Vietnamese real estate advertisements | |
| Qumsiyeh et al. | Searching web documents using a summarization approach | |
| CN114036921A (en) | Policy information matching method and device | |
| Al-Ayyoub et al. | Framework for Affective News Analysis of Arabic News: 2014 Gaza Attacks Case Study. | |
| Liang et al. | Clustering help-seeking behaviors in LGBT online communities: a prospective trial | |
| CN112163072B (en) | Data processing method and device based on multiple data sources |
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 |