CN112749283A - Entity relationship joint extraction method for legal field - Google Patents
Entity relationship joint extraction method for legal field Download PDFInfo
- Publication number
- CN112749283A CN112749283A CN202011625471.3A CN202011625471A CN112749283A CN 112749283 A CN112749283 A CN 112749283A CN 202011625471 A CN202011625471 A CN 202011625471A CN 112749283 A CN112749283 A CN 112749283A
- Authority
- CN
- China
- Prior art keywords
- legal
- entity
- model
- relationship
- characters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000605 extraction Methods 0.000 title claims abstract description 38
- 238000002372 labelling Methods 0.000 claims abstract description 12
- 238000000034 method Methods 0.000 claims abstract description 12
- 238000012360 testing method Methods 0.000 claims description 21
- 238000011156 evaluation Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 6
- 230000009286 beneficial effect Effects 0.000 abstract description 2
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/18—Legal services
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Business, Economics & Management (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Databases & Information Systems (AREA)
- Tourism & Hospitality (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Technology Law (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Animal Behavior & Ethology (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses a legal field-oriented entity relationship joint extraction method, which comprises the following steps: acquiring a text in the legal field, and carrying out entity relationship triple labeling on the text to construct a corpus in the legal field; vectorizing the text content to obtain the vector representation of the text; building a joint extraction model of the entity relationship of characters and involved objects in the legal field; training the established entity relation joint extraction model; and utilizing the relationship extraction model to extract the relationship between the characters and the involved articles of the unstructured documents in the legal field to be processed to obtain corresponding relationship triples. The invention has the beneficial effects that: the method can more accurately mine the relationship between the entities such as the characters, the involved articles and the like in the legal field.
Description
Technical Field
The invention relates to the field of entity relationship joint extraction, in particular to an entity relationship joint extraction method for the legal field.
Background
The method has the advantages that a large number of entities such as characters and related articles exist in the legal field, the relationships among the entities are complicated, and the knowledge in the documents in the legal field is difficult to integrate efficiently, so that relationship triplets of the entities such as the characters and the related articles in the legal field are mined, a structured knowledge representation is formed for the documents related to the legal field, and accordingly, practitioners related to the law are helped to master cases more accurately and analyze cases, and support is provided for intelligent services such as case pushing, criminal prediction and the like in intelligent law.
At present, most of the relation extraction fields adopt a pipeline type extraction method, namely, entities are extracted first, and then the relationship prediction is carried out on the entities by pairwise combination. The pipeline type extraction method is easy to realize, but has the problem of error propagation, the performance of relation extraction depends on the performance of entity extraction to a certain extent, and the pipeline type extraction can generate a large amount of redundant data, so that the requirement of the legal field, particularly the requirement of the legal field for accurately mastering cases is difficult to meet.
Therefore, it is necessary to provide a new extraction method to solve the above problems.
Disclosure of Invention
To solve the problems set forth in the background art described above. The invention provides a legal-field-oriented entity relationship joint extraction method, which can more accurately mine the relationship among all entities such as characters, case-related objects and the like in the legal field and meet the requirement of accurately mastering case situations in the legal field.
In order to achieve the purpose, the invention provides the following technical scheme: a legal domain-oriented entity relationship joint extraction method comprises the following steps:
step S1: acquiring a text in the legal field, and carrying out entity relationship triple labeling on the text to construct a corpus in the legal field; step S2: vectorizing the text content in step S1 to obtain a vector representation of the text; step S3: building a joint extraction model of the entity relationship of characters and involved objects in the legal field; the method comprises the following specific steps: step S31: for each word vector x obtained in step S2tThe Bi-LSTM of the coding layer is used for respectively obtaining the characteristic information of the documents in the legal field obtained by forward and backward propagation and respectively recording the characteristic informationStep (ii) ofS32: will be provided withSplicing to obtain the characteristic vector of the coding layer at the time t, and recording the characteristic vector asStep S33: inputting the feature vector of the coding layer in the step S13 into a CRF layer for identifying characters and involved object entities in the legal document; outputting the result of entity recognition and coding layerAs the input of the decoding layer Bi-LSTM at the time t, the semantic information of the document in the legal field is obtained by respectively calculating the forward propagation and the backward propagation in the same way and is respectively recorded asStep S34: will be provided withSplicing to obtain final semantic informationThe semantic vector is obtained by analyzing the Bi-LSTM at the decoding layer at the moment t according to the information of the context of the document; step S35: taking the semantic vector obtained in the step 15 as the input of a softmax classifier, and classifying to obtain a relationship label of an entity pair; step S4: training the established entity relation joint extraction model; step S5: and utilizing the relationship extraction model to extract the relationship between the characters and the involved articles of the unstructured documents in the legal field to be processed to obtain corresponding relationship triples.
The step S1 specifically includes:
step S11: downloading legal documents from websites such as a Chinese judge document network and the like according to different legal case types; step S12: labeling entities such as characters, involved articles and the like of the document processed in the step S11, and labeling relations between the characters and the involved articles to form relation triples; step S13: and repeating the steps S11 to S12 until all sentences are labeled, namely constructing a corpus of the legal field.
The step S2 specifically includes: step S21: for each sentence in the legal document, taking a character as a basic unit, and performing one-hot representation on each character by using a one-hot coding mode to obtain one-hot representation of the sentence; step S22: taking the one-hot vector of the sentence as the input of the word2vec model, training the word2vec model, and continuously updating the weight matrix w by using a gradient descent algorithm; step S23: and multiplying the weight matrix obtained by training in the step S22 by the one-hot vector of each word to obtain the word embedding of each word, and finally obtaining the word embedding representation of the whole sentence.
The step S4 specifically includes: step S41: randomly dividing the legal domain corpus to make the proportion of the training set to the test set to be 7:3, train _ x, test _ x, train _ y, and test _ y to train _ test _ split (x, y, and test _ size to be 0.3); step S42: the negative log-likelihood function is selected as a loss function, and the loss function is composed of an entity cost function and a relation cost function because the model realizes the joint extraction of entity relations such as characters, involved articles and the like in the legal field, and the cost function is as follows:
where | S | represents the length of the sentence, ei、riRepresenting entity labels and relation labels of the characters classified by the model, and theta represents a parameter set of the model; step S43: continuously updating the sharing parameter theta by using a random gradient descent algorithm; step S44: and training the model, and storing the trained model.
The step S5 specifically includes: step S51: testing the model by taking the test set obtained in the step S41 as the input of the model; step S22: and (4) performing performance evaluation on the relation triple result obtained in the step (S51), wherein the evaluation indexes adopt accuracy, recall rate and F1 value, and the evaluation method comprises the following steps:
wherein TP indicates the number of correctly classified classes, FP indicates the number of predicting negative classes as positive classes, and FN indicates the number of predicting positive classes as negative classes.
Compared with the prior art, the entity relationship joint extraction method oriented to the legal field has the beneficial effects that: the method can more accurately mine the relationship between the entities such as the characters, the involved articles and the like in the legal field.
Drawings
FIG. 1 is a flow chart of a method for jointly extracting entity relationships from the legal field according to the present invention;
FIG. 2 is a schematic flow chart of the method for building a joint extraction model of the physical relationship between characters and involved objects in the legal field;
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1 to 2, the present invention provides a joint extraction method of entity relationship in the legal domain, which includes the following steps:
step S1: acquiring a text in the legal field, and carrying out entity relationship triple labeling on the text to construct a corpus in the legal field;
specifically, the step S1 specifically includes:
step S11: acquiring related texts in the legal field from websites such as a Chinese referee document network and the like according to different case types, and downloading legal documents;
step S12: labeling entities such as characters, involved articles and the like of the document processed in the step S11, and labeling relations between the characters and the involved articles to form relation triples;
step S13: and repeating the steps S11 to S12 until all sentences are labeled, namely constructing a corpus of the legal field.
Specifically, for the entity designation identification task, the constructed legal domain corpus is required to be utilized for benchmarking, that is, entities such as people and involved articles in the legal document are labeled with BIO by the relation triplet, which is implemented as follows: for any single corpus, firstly, matching out a corresponding entity index in a legal document in a regular matching mode; then labeling the matched entity name according to a BIO labeling method, namely B represents the first character of the entity name, I represents the middle or tail character of the entity name, and O represents other characters; finally, all other characters in the legal document are marked as O.
Step S2: vectorizing the text content in step S1 to obtain a vector representation of the text;
the step S2 specifically includes:
step S21: for each sentence in the legal document, taking a character as a basic unit, and performing one-hot representation on each character by using a one-hot coding mode to obtain one-hot representation of the sentence;
step S22: taking the one-hot vector of the sentence as the input of the word2vec model, training the word2vec model, and continuously updating the weight matrix w by using a gradient descent algorithm;
step S23: and multiplying the weight matrix obtained by training in the step S22 by the one-hot vector of each word to obtain the word embedding of each word, and finally obtaining the word embedding representation of the whole sentence.
Step S3: building a joint extraction model of the entity relationship of characters and involved objects in the legal field; the method comprises the following specific steps:
as shown in fig. 2, a schematic flow chart of the joint extraction model for building the physical relationship between the legal field characters and the involved objects is shown.
Step S31: for each word vector x obtained in step S2tThe Bi-LSTM of the coding layer is used for respectively obtaining the characteristic information of the documents in the legal field obtained by forward and backward propagation and respectively recording the characteristic information
Step S32: will be provided withSplicing to obtain the characteristic vector of the coding layer at the time t, and recording the characteristic vector as
Step S33: inputting the feature vector of the coding layer in the step S32 into a CRF layer for identifying characters and involved object entities in the legal document; outputting the result of entity recognition and coding layerAs the input of the decoding layer Bi-LSTM at the time t, the semantic information of the document in the legal field is obtained by respectively calculating the forward propagation and the backward propagation in the same way and is respectively recorded as
Step S34: will be provided withSplicing to obtain final semantic informationI.e. decoding at time tThe layer Bi-LSTM analyzes the obtained semantic vector according to the information of the context of the document;
step S35: taking the semantic vector obtained in the step 34 as the input of a softmax classifier, and classifying to obtain a relationship label of an entity pair;
step S4: training the established entity relation joint extraction model;
specifically, the step S4 specifically includes:
step S41: randomly dividing the legal domain corpus to make the proportion of the training set to the test set to be 7:3, train _ x, test _ x, train _ y, and test _ y to train _ test _ split (x, y, and test _ size to be 0.3);
step S42: the negative log-likelihood function is selected as a loss function, and the loss function is composed of an entity cost function and a relation cost function because the model realizes the joint extraction of entity relations such as characters, involved articles and the like in the legal field, and the cost function is as follows:
where | S | represents the length of the sentence, ei、riRepresenting entity labels and relation labels of the characters classified by the model, and theta represents a parameter set of the model;
step S43: continuously updating the sharing parameter theta by using a random gradient descent algorithm;
step S44: and training the model, and storing the trained model.
Step S5: and utilizing the relationship extraction model to extract the relationship between the characters and the involved articles of the unstructured documents in the legal field to be processed to obtain corresponding relationship triples.
Specifically, the step S5 specifically includes:
step S51: testing the model by taking the test set obtained in the step S41 as the input of the model;
step S22: and (4) performing performance evaluation on the relation triple result obtained in the step (S51), wherein the evaluation indexes adopt accuracy, recall rate and F1 value, and the evaluation method comprises the following steps:
wherein TP indicates the number of correctly classified classes, FP indicates the number of predicting negative classes as positive classes, and FN indicates the number of predicting positive classes as negative classes.
Claims (5)
1. A legal-field-oriented entity relationship joint extraction method is characterized by comprising the following steps:
step S1: acquiring a text in the legal field, and carrying out entity relationship triple labeling on the text to construct a corpus in the legal field;
step S2: vectorizing the text content in step S1 to obtain a vector representation of the text;
step S3: building a joint extraction model of the entity relationship of characters and involved objects in the legal field; the method comprises the following specific steps:
step S31: for each word vector x obtained in step S2tThe Bi-LSTM of the coding layer is used for respectively obtaining the characteristic information of the documents in the legal field obtained by forward and backward propagation and respectively recording the characteristic information
Step S32: will be provided withSplicing to obtain the characteristic vector of the coding layer at the time t, and recording the characteristic vector as
Step S33: inputting the feature vector of the coding layer in the step S32 into a CRF layer for identifying characters and involved object entities in the legal document; outputting the result of entity recognition and coding layerAs the input of the decoding layer Bi-LSTM at the time t, the semantic information of the document in the legal field is obtained by respectively calculating the forward propagation and the backward propagation in the same way and is respectively recorded as
Step S34: will be provided withSplicing to obtain final semantic information The semantic vector is obtained by analyzing the Bi-LSTM at the decoding layer at the moment t according to the information of the context of the document;
step S35: taking the semantic vector obtained in the step S34 as the input of a softmax classifier, and classifying to obtain a relationship label of an entity pair;
step S4: training the established entity relation joint extraction model;
step S5: and utilizing the relationship extraction model to extract the relationship between the characters and the involved articles of the unstructured documents in the legal field to be processed to obtain corresponding relationship triples.
2. The method for joint extraction of entity relationships in the legal domain according to claim 1, wherein: the step S1 specifically includes:
step S11: downloading legal documents according to different case types;
step S12: labeling entities such as characters, involved articles and the like of the document processed in the step S11, and labeling relations between the characters and the involved articles to form relation triples;
step S13: and repeating the steps S11 to S12 until all sentences are labeled, namely constructing a corpus of the legal field.
3. The method for joint extraction of entity relationships in the legal domain according to claim 1, wherein: the step S2 specifically includes:
step S21: for each sentence in the legal document, taking a character as a basic unit, and performing one-hot representation on each character by using a one-hot coding mode to obtain one-hot representation of the sentence;
step S22: taking the one-hot vector of the sentence as the input of the word2vec model, training the word2vec model, and continuously updating the weight matrix w by using a gradient descent algorithm;
step S23: and multiplying the weight matrix obtained by training in the step S22 by the one-hot vector of each word to obtain the word embedding of each word, and finally obtaining the word embedding representation of the whole sentence.
4. The method for joint extraction of entity relationships in the legal domain according to claim 1, wherein: the step S4 specifically includes:
step S41: randomly dividing the legal domain corpus to make the proportion of the training set to the test set to be 7:3, train _ x, test _ x, train _ y, and test _ y to train _ test _ split (x, y, and test _ size to be 0.3);
step S42: the negative log-likelihood function is selected as a loss function, and the loss function is composed of an entity cost function and a relation cost function because the model realizes the joint extraction of entity relations such as characters, involved articles and the like in the legal field, and the cost function is as follows:
where | S | represents the length of the sentence, ei、riRepresenting entity labels and relation labels of the characters classified by the model, and theta represents a parameter set of the model;
step S43: continuously updating the sharing parameter theta by using a random gradient descent algorithm;
step S44: and training the model, and storing the trained model.
5. The method for joint extraction of entity relationships in the legal domain according to claim 4, wherein: the step S5 specifically includes:
step S51: testing the model by taking the test set obtained in the step S41 as the input of the model;
step S22: and (4) performing performance evaluation on the relation triple result obtained in the step (S51), wherein the evaluation indexes adopt accuracy, recall rate and F1 value, and the evaluation method comprises the following steps:
wherein TP indicates the number of correctly classified classes, FP indicates the number of predicting negative classes as positive classes, and FN indicates the number of predicting positive classes as negative classes.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011625471.3A CN112749283A (en) | 2020-12-31 | 2020-12-31 | Entity relationship joint extraction method for legal field |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011625471.3A CN112749283A (en) | 2020-12-31 | 2020-12-31 | Entity relationship joint extraction method for legal field |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112749283A true CN112749283A (en) | 2021-05-04 |
Family
ID=75650533
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011625471.3A Pending CN112749283A (en) | 2020-12-31 | 2020-12-31 | Entity relationship joint extraction method for legal field |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112749283A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113342929A (en) * | 2021-05-07 | 2021-09-03 | 上海大学 | Material-component-process-performance relation quadruple extraction method for material field |
CN113377916A (en) * | 2021-06-22 | 2021-09-10 | 哈尔滨工业大学 | Extraction method of main relations in multiple relations facing legal text |
CN113553385A (en) * | 2021-07-08 | 2021-10-26 | 北京计算机技术及应用研究所 | Relation extraction method of legal elements in judicial documents |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109165385A (en) * | 2018-08-29 | 2019-01-08 | 中国人民解放军国防科技大学 | Multi-triple extraction method based on entity relationship joint extraction model |
CN110502749A (en) * | 2019-08-02 | 2019-11-26 | 中国电子科技集团公司第二十八研究所 | A kind of text Relation extraction method based on the double-deck attention mechanism Yu two-way GRU |
CN110781683A (en) * | 2019-11-04 | 2020-02-11 | 河海大学 | Entity relation joint extraction method |
CN111046670A (en) * | 2019-12-09 | 2020-04-21 | 大连理工大学 | Entity and relationship combined extraction method based on drug case legal documents |
CN111382575A (en) * | 2020-03-19 | 2020-07-07 | 电子科技大学 | Event extraction method based on joint labeling and entity semantic information |
CN111581387A (en) * | 2020-05-09 | 2020-08-25 | 电子科技大学 | Entity relation joint extraction method based on loss optimization |
CN111783464A (en) * | 2020-06-29 | 2020-10-16 | 中国电力科学研究院有限公司 | Electric power-oriented domain entity identification method, system and storage medium |
-
2020
- 2020-12-31 CN CN202011625471.3A patent/CN112749283A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109165385A (en) * | 2018-08-29 | 2019-01-08 | 中国人民解放军国防科技大学 | Multi-triple extraction method based on entity relationship joint extraction model |
CN110502749A (en) * | 2019-08-02 | 2019-11-26 | 中国电子科技集团公司第二十八研究所 | A kind of text Relation extraction method based on the double-deck attention mechanism Yu two-way GRU |
CN110781683A (en) * | 2019-11-04 | 2020-02-11 | 河海大学 | Entity relation joint extraction method |
CN111046670A (en) * | 2019-12-09 | 2020-04-21 | 大连理工大学 | Entity and relationship combined extraction method based on drug case legal documents |
CN111382575A (en) * | 2020-03-19 | 2020-07-07 | 电子科技大学 | Event extraction method based on joint labeling and entity semantic information |
CN111581387A (en) * | 2020-05-09 | 2020-08-25 | 电子科技大学 | Entity relation joint extraction method based on loss optimization |
CN111783464A (en) * | 2020-06-29 | 2020-10-16 | 中国电力科学研究院有限公司 | Electric power-oriented domain entity identification method, system and storage medium |
Non-Patent Citations (2)
Title |
---|
董哲;康宇佳;: "基于机器学习与模式匹配的食品安全刑事裁判文书关键信息提取方法", 信息技术与信息化, no. 05, 28 May 2020 (2020-05-28) * |
郝志峰;陈培辉;蔡瑞初;温雯;王丽娟;: "基于叠层循环神经网络的语义关系分类模型", 计算机应用研究, no. 01, 30 November 2018 (2018-11-30) * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113342929A (en) * | 2021-05-07 | 2021-09-03 | 上海大学 | Material-component-process-performance relation quadruple extraction method for material field |
CN113377916A (en) * | 2021-06-22 | 2021-09-10 | 哈尔滨工业大学 | Extraction method of main relations in multiple relations facing legal text |
CN113553385A (en) * | 2021-07-08 | 2021-10-26 | 北京计算机技术及应用研究所 | Relation extraction method of legal elements in judicial documents |
CN113553385B (en) * | 2021-07-08 | 2023-08-25 | 北京计算机技术及应用研究所 | Relation extraction method for legal elements in judicial document |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110019839B (en) | Medical knowledge graph construction method and system based on neural network and remote supervision | |
CN107729309B (en) | Deep learning-based Chinese semantic analysis method and device | |
CN111709241A (en) | Named entity identification method oriented to network security field | |
CN112749283A (en) | Entity relationship joint extraction method for legal field | |
CN113051916B (en) | Interactive microblog text emotion mining method based on emotion offset perception in social network | |
CN108664474B (en) | Resume analysis method based on deep learning | |
CN109871452B (en) | Method, apparatus and storage medium for determining crime characteristics | |
CN112818093B (en) | Evidence document retrieval method, system and storage medium based on semantic matching | |
CN113051356B (en) | Open relation extraction method and device, electronic equipment and storage medium | |
CN111783394A (en) | Training method of event extraction model, event extraction method, system and equipment | |
CN111753024A (en) | Public safety field-oriented multi-source heterogeneous data entity alignment method | |
CN106547875B (en) | Microblog online emergency detection method based on emotion analysis and label | |
CN110457585B (en) | Negative text pushing method, device and system and computer equipment | |
CN110909549A (en) | Method, device and storage medium for punctuating ancient Chinese | |
CN113742733B (en) | Method and device for extracting trigger words of reading and understanding vulnerability event and identifying vulnerability type | |
CN110019820B (en) | Method for detecting time consistency of complaints and symptoms of current medical history in medical records | |
CN115587594B (en) | Unstructured text data extraction model training method and system for network security | |
CN113392209A (en) | Text clustering method based on artificial intelligence, related equipment and storage medium | |
CN112257444B (en) | Financial information negative entity discovery method, device, electronic equipment and storage medium | |
CN113011161A (en) | Method for extracting human and pattern association relation based on deep learning and pattern matching | |
CN111930792A (en) | Data resource labeling method and device, storage medium and electronic equipment | |
CN113220768A (en) | Resume information structuring method and system based on deep learning | |
CN114153978A (en) | Model training method, information extraction method, device, equipment and storage medium | |
CN113360654B (en) | Text classification method, apparatus, electronic device and readable storage medium | |
CN113220964B (en) | Viewpoint mining method based on short text in network message field |
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 |