CN113191156A - Medical examination item standardization system and method based on medical knowledge graph and pre-training model - Google Patents
Medical examination item standardization system and method based on medical knowledge graph and pre-training model Download PDFInfo
- Publication number
- CN113191156A CN113191156A CN202110475162.0A CN202110475162A CN113191156A CN 113191156 A CN113191156 A CN 113191156A CN 202110475162 A CN202110475162 A CN 202110475162A CN 113191156 A CN113191156 A CN 113191156A
- Authority
- CN
- China
- Prior art keywords
- medical
- standard
- text
- module
- item
- 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
Images
Classifications
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2455—Query execution
- G06F16/24564—Applying rules; Deductive queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2457—Query processing with adaptation to user needs
- G06F16/24578—Query processing with adaptation to user needs using ranking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
- G06F16/355—Class or cluster creation or modification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
Abstract
The invention provides a medical examination item standardization system and method based on a medical knowledge map and a pre-training model, wherein the system comprises a medical text preprocessing module, a candidate standard item recall module and a semantic similarity evaluation module; in a medical text preprocessing module, preprocessing a medical Chinese text; in the candidate standard item recall module, n standard items with the highest matching degree are recalled by evaluating the text characteristics between the text to be standardized and the standard items; in the semantic similarity evaluation module, a pre-training deep learning model is utilized to sequentially calculate the semantic similarity of the recalled candidate standard items and the items to be standardized; the knowledge graph module is an external data source of the system and provides entity information for text preprocessing and semantic matching. According to the invention, through standardization processing, medical examination project names from different medical institutions can be mapped to a unified standard examination project, so that support is provided for subsequent information processing.
Description
Technical Field
The invention relates to the field of natural language processing and the field of machine learning application, in particular to a medical examination item standardization system and method based on a medical knowledge map and a pre-training model.
Background
With the development of society and the improvement of health consciousness of people, more and more medical data need to be processed uniformly, especially a great amount of medical text data, such as electronic medical record texts, physical examination report texts and imaging examination report texts. However, due to the development status of the medical industry and the complexity of medical knowledge, the application specifications of different medical institutions for medical terms are quite different, and sometimes, a medical entity with the same meaning may have more than ten or even dozens of different expressions, which brings great inconvenience to the processing of medical texts. In hospitals, the most important medical entity includes the type of examination items, the examination items in different hospitals or medical institutions are approximately in the same range, but the expressions are different, and the text names can be standardized through standardization of the medical items, so that great convenience is provided for subsequent data processing statistics and index abnormality identification.
In order to solve the above problems, workers in the field have conducted various studies on medical examination items:
chinese patent application 202011415694.7 discloses a processing method for medical knowledge graph, which relates to the field of artificial intelligence and can be used in the fields of knowledge graph, deep learning, natural language processing and the like. The specific implementation scheme is as follows: extracting a medical entity from the medical text; identifying the medical entity by using the medical entity model obtained by the transfer learning to obtain a corresponding identification result; and in response to the recognition result characterizing the medical entity as an alias of an existing standard entity, adding the medical entity in the medical knowledge-graph and adding corresponding alias attribute information for the medical entity. However, the invention is a method in the knowledge graph construction stage, mainly aims to expand entities in the knowledge graph and update attributes, and has the main tasks of naming entity identification and entity connection, the recall stage in the entity connection stage is rough, only edit distance is used as a judgment basis, the names of index items and parent names contain less information, and misjudgment and omission of matching are easily caused.
The Chinese patent application 201910520186.6 discloses a medical text named entity recognition method based on a pre-training model and a fine-tuning technology, and the method comprises the steps of pre-training a BERT pre-training model by utilizing medical texts such as large-scale unstructured electronic medical records and the like to train the pre-training model containing semantic representation information in the texts. And fine-tuning the generated pre-training model by utilizing the stacked expanded convolutional neural network to obtain a deep neural network model capable of automatically identifying the named entities in the medical field. The pre-training model provided by the invention can more accurately capture semantic information in a text, can be more effectively transferred to a specific task, and improves the accuracy of the model for identifying the named entity; according to the method, the stack expansion convolutional neural network is combined with the pre-training model to finely adjust the model, and finally, the medical text named entity is identified, so that not only can semantic information in the text be well captured, but also parallel calculation can be performed, and the model training speed is improved. However, the task of the patent of the invention is named entity recognition, the main purpose is to recognize the position of the named entity in a section of text, S1-S4 are common steps of a natural language processing model, and the preprocessing utilizes the traditional method to perform word segmentation and stop word filtering on the text, so that a good effect is not achieved.
Therefore, there is a need to invent a system and method for standardizing medical examination items based on medical knowledge maps and pre-trained models to solve the above-mentioned disadvantages.
Disclosure of Invention
The purpose of the application is: the system and the method for standardizing the medical examination items based on the medical knowledge graph and the pre-training model are provided, and medical examination item names from different medical institutions can be mapped to a unified standard examination item through standardization processing, so that support is provided for subsequent information processing.
The purpose of the application is achieved by the following technical scheme, the system for standardizing the medical examination items based on the medical knowledge graph and the pre-training model comprises the following steps: the system comprises a medical text preprocessing module, a candidate standard project recalling module and a semantic similarity evaluation module;
the medical text preprocessing module: the method comprises the following steps of preprocessing the medical texts, wherein the preprocessing module can be configured according to the range, the quantity and the type of the processed texts;
the candidate standard item recall module: the text characteristics between the text to be standardized and the standard items are evaluated, the n standard items with the highest matching degree are recalled, the standardized screening range is quickly reduced, and the difference between the standard items and the standard items to be standardized is evaluated from the coarse granularity;
the semantic similarity evaluation module: and sequentially performing semantic similarity calculation on the recalled candidate standard items and the items to be standardized by using the pre-training deep learning model, outputting a standardized result if the standard item with the highest semantic similarity exceeds a preset threshold, and outputting a non-matching item if the standard item does not exceed the preset threshold.
Preferably, in the training of the deep learning model, the constructed knowledge-map attributes can be utilized to enrich the medical information of the standard item in addition to the information of the medical text itself.
A medical examination project standardization system based on a medical knowledge map and a pre-training model further comprises a network service, wherein three modules are respectively deployed on a server, the three modules are respectively three independent web applications, and information communication and data transmission among the three modules are processed through restful API.
A method for medical examination item standardization based on a medical knowledge-graph and a pre-trained model, comprising:
the method comprises the following steps: the inspection items generally comprise three parts, namely parent names, inspection item names and other attribute information, the two names are preprocessed through a medical text preprocessing module, wherein texts are simply sorted by utilizing regular expressions and other text processing rules, and noise data and irrelevant service information are deleted, so that subsequent standardized calculation is more accurate and faster;
step two: sequencing massive standardized items by using an information retrieval means, wherein the sequencing standard and indexes can be configured according to specific service requirements, generally are coarse-grained text matching indexes, and are not subjected to calculation-intensive operation, and the top n bits of a recall result are selected as a candidate queue for semantic recognition;
step three: semantic similarity evaluation is carried out on the text and standard inspection items in the candidate queue, the standard inspection items contain all information of related entities through retrieval of a knowledge graph module, and semantic information comparison can be comprehensively and comprehensively carried out;
step four: and taking the standard inspection item with the highest semantic similarity score in the queue as a final matching result, and if the score of the final matching result is too low, the final matching result has no referential property.
Preferably, the knowledge-graph module is an external data source of the system, providing entity information for text preprocessing and semantic matching, and the original data source of the knowledge-graph can be from a structured database or a manually written data document.
Preferably, semantic matching is performed on the standardized items and the candidate standard check items in sequence, the result of each group of semantic matching is a floating point numerical value between 0 and 1, the larger the number is, the larger the semantic relevance between the two groups of standard items is, and the smaller the semantic relevance is otherwise.
Preferably, the knowledge graph module comprises a knowledge graph network for storing information of all standard examination items, the other modules carry out inference query through a specific SPARQL query statement, and the other modules and the knowledge graph module utilize restful API to carry out interaction with the knowledge graph module, so that entity information is provided for text preprocessing and semantic matching.
Compared with the prior art, the application has the following obvious advantages and effects:
1. in the present invention: by combining the knowledge map and the deep learning natural language processing technology, medical examination items which are expressed differently in different medical institutions are mapped to a uniform standard item, so that support is provided for subsequent information processing.
2. In the present invention: based on actual service data, a complete preprocessing flow and a more perfect recall strategy are provided, the method is more suitable for actual application scenes, more medical knowledge can be given to the model for decision making according to attribute information in the knowledge graph, for example, units and brief introduction in the knowledge graph can describe assay means, test objects and detection parts more carefully, and the matching accuracy is improved.
3. In the present invention: training and fine tuning are carried out on the deep learning pre-training model, the existing data set containing the label is utilized to train the current semantic recognition task, the semantic matching capability of the pre-training model is optimized, the accuracy and the generalization of the model are improved, and the model has higher practicability on the task.
4. The task of the invention is text matching, the main purpose is to judge whether the semantics of two text segments are the same, the text is subjected to special preprocessing based on medical knowledge based on the category and the text mode of an index item, a large-scale pre-training model based on a Transformer model is used, fine-tuning training is carried out according to the labeled data of the text matching, the parameter scale is larger, the expression effect is better, and the prediction result is more reasonable.
Drawings
Fig. 1 is an overall architecture diagram of the system of the present application.
Fig. 2 is an overall architecture diagram of the system medical text pre-processing module in the present application.
FIG. 3 is a block diagram of the matching method of the system pre-training deep learning model in the present application.
Reference numerals:
the method comprises the following steps of 1 item inspection information, 2 medical text preprocessing module, 21 to-be-processed text processing, 22 text cleaning, 23 inspection item error correction, 24 imaging inspection part matching, 25 abbreviation expansion, 26 special rules, 27 term replacement, 28 flow configuration file, 29 processed text, 3 knowledge map, 4 candidate standard item recall module, 5 deep learning semantic matching module, 6 standard inspection item, 7 pre-training model, 71 to-be-normalized item information, 72 standard item information, 73 combined input text information, 74 pre-training deep learning model, 75 text characterization vector, 76 multi-layer perceptron classifier and 77 semantic similarity.
Detailed Description
The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It will be appreciated that the data so used may be interchanged under appropriate circumstances such that the embodiments described herein may be practiced otherwise than as specifically illustrated or described herein. Furthermore, the terms "comprises," "comprising," or "having," and any variations thereof, are intended to cover non-exclusive inclusions, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Fig. 1-3 illustrate one embodiment of the system and method for medical examination items of the present application.
The invention provides a medical examination item standardization system and method based on a medical knowledge map and a pre-training model, wherein the system comprises a medical text preprocessing module 2, a candidate standard item recall module and a semantic similarity 77 evaluation module; the inspection items generally comprise three parts, namely parent names, inspection item names and other attribute information, the two names are preprocessed through the medical text preprocessing module 2, texts are simply sorted by using regular expressions and other text processing rules, and noise data and irrelevant service information are deleted, so that subsequent standardized calculation is more accurate and faster; sequencing massive standardized items by using an information retrieval means, wherein the sequencing standard and indexes can be configured according to specific service requirements, generally are coarse-grained text matching indexes, and are not subjected to calculation-intensive operation, and the top n bits of a recall result are selected as a candidate queue for semantic recognition; semantic similarity 77 evaluation is carried out on the text and a standard inspection item 6 in the candidate queue, the standard inspection item 6 contains all information of related entities through retrieval of a knowledge graph 3 module, and semantic information comparison can be comprehensively and comprehensively carried out; and taking the standard check item 6 with the highest semantic similarity 77 score in the queue as a final matching result, wherein if the score of the final matching result is too low, the final matching result has no referential property. According to the invention, through standardization processing, medical examination item names from different medical institutions can be mapped into the unified standard examination item 6, so that support is provided for subsequent information processing.
As shown in fig. 1 to 3, in the present embodiment,
firstly, a medical knowledge map 3 module is started on a server, the knowledge map 3 module comprises a knowledge map 3 network for storing all standard item inspection information 1, each standard item inspection information 1 entity comprises the name of an inspection item, the name of a father class, a unit, an index range, a brief introduction and other related attributes, such as an index ' thyroid stimulating hormone ', the name of the father class is ' hormone detection ', the unit is ' mIU/L ', the index range is ' 0.2-7 ', the brief introduction is ' thyroid stimulating hormone is a hormone for promoting synthesis and release of thyroid hormone, and if the hypothalamus and anterior pituitary function normally, the concentration of the thyroid hormone reflects the state of the thyroid hormone in the tissue. Other modules perform inference queries through a specific SPARQL query statement, and other modules interact with the knowledgegraph 3 module using restful apis. The knowledge-graph 3 module is an external data source of the system and provides entity information for text preprocessing and semantic matching. The original data source of the knowledge-graph 3 may be from a structured database or a manually written data document.
And secondly, training and fine tuning the deep learning pre-training model 7, training a current semantic recognition task by utilizing the existing data set containing the label, optimizing the semantic matching capability of the pre-training model 7, improving the accuracy and the generalization of the model, and ensuring that the model has higher practicability on the task. The matched input is a text composed of project information 71 to be standardized and standard project information 72, and the structure of the text is that the step of 'index name < SEP1> index parent class name < SEP2> standard index name < SEP3> standard index parent class name < SEP3> standard index knowledge graph 3 attribute information' is generally completed on a computer with GPU parallel computing capability.
And thirdly, respectively deploying network services of three modules on the server, wherein the three modules are respectively three independent web application programs, and the information communication and data transmission between the three modules are processed through restful API. The input of the medical text preprocessing module 2 is original user index item information, noise information of the input information is filtered through a plurality of independent preprocessing steps, and processed medical texts, such as 'pituitary Prolactin (PRL) and six sex hormones' (delivery) 'delivery' in the index can be filtered. The processed medical text and other attribute information are used as input information and transmitted to a candidate standard item recall module 4, which outputs a candidate standard item list, wherein the list generally only comprises ID information of standard items, and the length of the list is determined by user configuration parameters. Generally, as the length increases, the accuracy of semantic matching is improved to a certain extent, and when a certain peak value is reached, the accuracy is not improved, but the response speed of semantic matching is affected. The specific user configuration parameters need to be determined by performing multiple business actual data experiments. The candidate list is input into a deep learning semantic matching module 5 for semantic matching, and in order to improve the computation speed of the deep learning pre-training model 7, the module is generally deployed on a platform with GPU parallel computation capability. And semantic matching is sequentially carried out on the item to be normalized and the candidate standard check item 6, the result of each group of semantic matching is a floating point numerical value between 0 and 1, the larger the number is, the larger the semantic relevance between the two groups of standard items is, and the smaller the semantic relevance is otherwise. And taking the item with the highest semantic matching value in the candidate standard check item 6 list as a final output result. In particular, if the semantic matching degree of the final output result is lower than a certain acceptable degree, which means that it is possible that this check item does not exist in the standard check item 6 list, the knowledge-graph 3 does not contain the corresponding entity, and a null value is required to be output, indicating that no result is matched.
Specifically, as shown in fig. 2, in the embodiment of the present application,
and (3) cleaning the text 21: including three parts of parent class name, inspection item name and other attribute information
Text washing 22: the text is simply sorted by regular expressions and other text processing rules.
Process configuration file 28: checking item error correction 23, imaging checking part matching 24, abbreviation expansion 25, special rules 26, term replacement 27, deleting noise data and irrelevant service information, and finally processing to complete text 29.
It should be noted that, as shown in fig. 2, in the embodiment of the present application,
the existing data set containing the label is utilized to train the current semantic recognition task, the semantic matching capability of the pre-training model 7 is optimized, the accuracy and the generalization of the model are improved, and the model has higher practicability on the task. The matched input is a text composed of the item information 71 to be standardized and the standard item information 72, and the step of combining the input text information 73 with the structure of 'index name < SEP1> index parent class name < SEP2> standard index name < SEP3> standard index parent class name < SEP3> standard index knowledge map 3 attribute information' is generally completed on a computer with GPU parallel computing capability.
The network service of three modules is respectively deployed on the server, the three modules are respectively three independent web application programs, and information communication and data transmission between the three modules are processed through restful API. Pre-trained deep learning model 74: the result of each group of semantic matching is a floating point number between 0 and 1, the larger the number is, the larger the semantic relevance between the two groups of standard items is, and the smaller the semantic relevance is otherwise
Text characterization vector 75: the text is processed using the text characterization vectors 75.
Multi-layer perceptron classifier 76: the classification is performed using a multi-level perceptron classifier 76.
Semantic similarity 77: if the semantic matching degree of the final output result is lower than a certain acceptable degree, which means that the check item may not exist in the standard check item 6 directory, the knowledge-graph 3 does not contain the corresponding entity, and a null value is required to be output to indicate that any result is not matched.
Since any modifications, equivalents, improvements, etc. made within the spirit and principles of the application may readily occur to those skilled in the art, it is intended to be included within the scope of the claims of this application.
Claims (6)
1. A system for medical examination item standardization based on a medical knowledge-graph and a pre-trained model, comprising: the system comprises a medical text preprocessing module, a candidate standard project recalling module and a semantic similarity evaluation module;
the medical text preprocessing module: the method comprises the following steps of preprocessing the medical texts, wherein the preprocessing module can be configured according to the range, the quantity and the type of the processed texts;
the candidate standard item recall module: the text characteristics between the text to be standardized and the standard items are evaluated, the n standard items with the highest matching degree are recalled, the standardized screening range is quickly reduced, and the difference between the standard items and the standard items to be standardized is evaluated from the coarse granularity;
the semantic similarity evaluation module: and sequentially performing semantic similarity calculation on the recalled candidate standard items and the items to be standardized by using the pre-training deep learning model, outputting a standardized result if the standard item with the highest semantic similarity exceeds a preset threshold, and outputting a non-matching item if the standard item does not exceed the preset threshold.
2. The system of claim 1, wherein the medical knowledge-graph is normalized with respect to a pre-trained model of the medical examination item, and wherein: in the training of the deep learning model, the constructed knowledge map attributes can be utilized to enrich the medical information of standard items besides the information of the medical text.
3. The system of claim 2, wherein the medical knowledge-graph is normalized with respect to a pre-trained model of the medical examination item, and wherein: the knowledge graph module is an external data source of the system and provides entity information for text preprocessing and semantic matching, and the original data source of the knowledge graph can be from a structured database or a manually written data document.
4. A method for medical examination item standardization based on a medical knowledge-graph and a pre-trained model, comprising:
the method comprises the following steps: the inspection items generally comprise three parts, namely parent names, inspection item names and other attribute information, the two names are preprocessed through a medical text preprocessing module, wherein texts are simply sorted by utilizing regular expressions and other text processing rules, and noise data and irrelevant service information are deleted, so that subsequent standardized calculation is more accurate and faster;
step two: sequencing massive standardized items by using an information retrieval means, wherein the sequencing standard and indexes can be configured according to specific service requirements, generally are coarse-grained text matching indexes, and are not subjected to calculation-intensive operation, and the top n bits of a recall result are selected as a candidate queue for semantic recognition;
step three: semantic similarity evaluation is carried out on the text and standard inspection items in the candidate queue, the standard inspection items contain all information of related entities through retrieval of a knowledge graph module, and semantic information comparison can be comprehensively and comprehensively carried out;
step four: and taking the standard inspection item with the highest semantic similarity score in the queue as a final matching result, and if the score of the final matching result is too low, the final matching result has no referential property.
5. The method of claim 4, wherein the medical knowledge-graph is normalized with respect to a pre-trained model of the medical examination item, and the method comprises: and performing semantic matching on the item to be normalized and the candidate standard check item in sequence, wherein the result of each group of semantic matching is a floating point numerical value between 0 and 1, the larger the number is, the larger the semantic relevance between the two groups of standard items is, and the smaller the semantic relevance is otherwise.
6. The method of claim 4, wherein the medical knowledge-graph is normalized with respect to a pre-trained model of the medical examination item, and the method comprises: the knowledge graph module comprises a knowledge graph network for storing information of all standard checking items, other modules carry out inference query through specific SPARQL query sentences, and other modules and the knowledge graph module utilize restful API to carry out interaction of the knowledge graph module, are external data sources of the system and provide entity information for text preprocessing and semantic matching.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110475162.0A CN113191156A (en) | 2021-04-29 | 2021-04-29 | Medical examination item standardization system and method based on medical knowledge graph and pre-training model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110475162.0A CN113191156A (en) | 2021-04-29 | 2021-04-29 | Medical examination item standardization system and method based on medical knowledge graph and pre-training model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113191156A true CN113191156A (en) | 2021-07-30 |
Family
ID=76980792
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110475162.0A Pending CN113191156A (en) | 2021-04-29 | 2021-04-29 | Medical examination item standardization system and method based on medical knowledge graph and pre-training model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113191156A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113590842A (en) * | 2021-08-05 | 2021-11-02 | 思必驰科技股份有限公司 | Medical term standardization method and system |
| CN113657086A (en) * | 2021-08-09 | 2021-11-16 | 腾讯科技（深圳）有限公司 | Word processing method, device, equipment and storage medium |
| CN113742494A (en) * | 2021-09-06 | 2021-12-03 | 湘潭大学 | Domain text similarity calculation method and system based on label graph conversion |
| CN114153995A (en) * | 2022-02-09 | 2022-03-08 | 杭州太美星程医药科技有限公司 | Medical term processing method, apparatus, computer device and storage medium |
| WO2023065858A1 (en) * | 2021-10-19 | 2023-04-27 | 之江实验室 | Medical term standardization system and method based on heterogeneous graph neural network |
| CN116821712A (en) * | 2023-08-25 | 2023-09-29 | 中电科大数据研究院有限公司 | Semantic matching method and device for unstructured text and knowledge graph |
| CN116992294A (en) * | 2023-09-26 | 2023-11-03 | 成都国恒空间技术工程股份有限公司 | Satellite measurement and control training evaluation method, device, equipment and storage medium |
-
2021
- 2021-04-29 CN CN202110475162.0A patent/CN113191156A/en active Pending
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113590842A (en) * | 2021-08-05 | 2021-11-02 | 思必驰科技股份有限公司 | Medical term standardization method and system |
| CN113657086A (en) * | 2021-08-09 | 2021-11-16 | 腾讯科技（深圳）有限公司 | Word processing method, device, equipment and storage medium |
| CN113657086B (en) * | 2021-08-09 | 2023-08-15 | 腾讯科技（深圳）有限公司 | Word processing method, device, equipment and storage medium |
| CN113742494A (en) * | 2021-09-06 | 2021-12-03 | 湘潭大学 | Domain text similarity calculation method and system based on label graph conversion |
| CN113742494B (en) * | 2021-09-06 | 2024-03-15 | 湘潭大学 | Domain text similarity calculation method and system based on label graph conversion |
| WO2023065858A1 (en) * | 2021-10-19 | 2023-04-27 | 之江实验室 | Medical term standardization system and method based on heterogeneous graph neural network |
| CN114153995A (en) * | 2022-02-09 | 2022-03-08 | 杭州太美星程医药科技有限公司 | Medical term processing method, apparatus, computer device and storage medium |
| CN116821712A (en) * | 2023-08-25 | 2023-09-29 | 中电科大数据研究院有限公司 | Semantic matching method and device for unstructured text and knowledge graph |
| CN116821712B (en) * | 2023-08-25 | 2023-12-19 | 中电科大数据研究院有限公司 | Semantic matching method and device for unstructured text and knowledge graph |
| CN116992294A (en) * | 2023-09-26 | 2023-11-03 | 成都国恒空间技术工程股份有限公司 | Satellite measurement and control training evaluation method, device, equipment and storage medium |
| CN116992294B (en) * | 2023-09-26 | 2023-12-19 | 成都国恒空间技术工程股份有限公司 | Satellite measurement and control training evaluation method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113191156A (en) | Medical examination item standardization system and method based on medical knowledge graph and pre-training model | |
| CN104008395A (en) | Intelligent bad video detection method based on face retrieval | |
| CN112015721A (en) | E-commerce platform storage database optimization method based on big data | |
| CN109284374A (en) | For determining the method, apparatus, equipment and computer readable storage medium of entity class | |
| CN116089873A (en) | Model training method, data classification and classification method, device, equipment and medium | |
| CN110134777A (en) | Problem De-weight method, device, electronic equipment and computer readable storage medium | |
| CN112632226A (en) | Semantic search method and device based on legal knowledge graph and electronic equipment | |
| CN115098706A (en) | Network information extraction method and device | |
| CN114297140A (en) | Archive management system based on artificial intelligence | |
| CN113486670B (en) | Text classification method, device, equipment and storage medium based on target semantics | |
| CN114398473A (en) | Enterprise portrait generation method, device, server and storage medium | |
| KR102280490B1 (en) | Training data construction method for automatically generating training data for artificial intelligence model for counseling intention classification | |
| CN114328800A (en) | Text processing method and device, electronic equipment and computer readable storage medium | |
| CN111259223B (en) | News recommendation and text classification method based on emotion analysis model | |
| CN112685374B (en) | Log classification method and device and electronic equipment | |
| CN116976321A (en) | Text processing method, apparatus, computer device, storage medium, and program product | |
| CN114610882A (en) | Abnormal equipment code detection method and system based on electric power short text classification | |
| CN113722484A (en) | Rumor detection method, device, equipment and storage medium based on deep learning | |
| CN113722421A (en) | Contract auditing method and system and computer readable storage medium | |
| CN112837771B (en) | Abnormal physical examination item normalization method integrating text classification and lexical analysis | |
| CN116860977B (en) | Abnormality detection system and method for contradiction dispute mediation | |
| US20240070175A1 (en) | Method for Determining Company Related to News Based on Scoring and Apparatus for Performing the Method | |
| US20240046039A1 (en) | Method for News Mapping and Apparatus for Performing the Method | |
| Палій et al. | Data recognition in documents and classification algorithm | |
| Palii et al. | Data recognition in documents and classification algorithm |
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 |