CN116484867A - Named entity recognition method and device, storage medium and computer equipment - Google Patents

Named entity recognition method and device, storage medium and computer equipment Download PDF

Info

Publication number
CN116484867A
CN116484867A CN202310466342.1A CN202310466342A CN116484867A CN 116484867 A CN116484867 A CN 116484867A CN 202310466342 A CN202310466342 A CN 202310466342A CN 116484867 A CN116484867 A CN 116484867A
Authority
CN
China
Prior art keywords
named entity
target
category
vector
sequence
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310466342.1A
Other languages
Chinese (zh)
Inventor
周恩国
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Technology Shenzhen Co Ltd
Original Assignee
Ping An Technology Shenzhen Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ping An Technology Shenzhen Co Ltd filed Critical Ping An Technology Shenzhen Co Ltd
Priority to CN202310466342.1A priority Critical patent/CN116484867A/en
Publication of CN116484867A publication Critical patent/CN116484867A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/279—Recognition of textual entities
    • G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
    • G06F40/295—Named entity recognition
    • 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
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/279—Recognition of textual entities
    • G06F40/284—Lexical analysis, e.g. tokenisation or collocates
    • 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
    • G06N3/045—Combinations of networks
    • 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
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Biomedical Technology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Software Systems (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Public Health (AREA)
  • Primary Health Care (AREA)
  • Medical Informatics (AREA)
  • Epidemiology (AREA)
  • Databases & Information Systems (AREA)
  • Machine Translation (AREA)

Abstract

The invention discloses a named entity identification method and device, a storage medium and computer equipment, relates to the technical field of information processing and the field of medical services, and mainly aims to solve the problem of low named entity identification accuracy. The method mainly comprises the steps of obtaining a target text sequence carrying prompt information of a named entity category to be identified, wherein the target text sequence is determined based on the text to be identified; extracting features of the target text sequence by using a feature extraction neural network of the named entity recognition model after training is completed, and obtaining a word vector sequence for representing semantic feature information of the target text sequence; and predicting at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the named entity recognition model after training to obtain the target named entity meeting the named entity class prompt information to be recognized. The method is mainly used for identifying the named entity.

Description

Named entity recognition method and device, storage medium and computer equipment
Technical Field
The present invention relates to the field of information processing and medical services, and in particular, to a named entity recognition method and apparatus, a storage medium, and a computer device.
Background
Named entities are person names, organization names, place names, and all other entities identified by names. Named entity recognition is one of the hot research directions in natural language processing, with the aim of identifying and generalizing named entities in text. Named entity recognition can realize extraction of key information in text information, and is widely applied to various fields. For example, in a medical service scenario, intelligent understanding of the electronic medical record can be achieved based on extraction of the core named entity in the electronic medical record, and the named entity obtained by extraction can also provide more accurate and brief medical information reference for medical workers.
The existing named entity recognition method is to extract features of texts to be recognized based on a pre-training model and classify the texts based on the extracted features, so that named entity categories of different entities in the texts are determined. However, the training process of the pre-training model is completed based on a large amount of labeling resources, and the named entity recognition process is a feature classification process, so that the pre-training process and the downstream fine tuning task are split, the effect of the pre-training model cannot reach expectations in the named entity recognition process, and the named entity recognition accuracy is low.
Disclosure of Invention
In view of the above, the present invention provides a named entity recognition method and apparatus, a storage medium, and a computer device, and is mainly aimed at solving the problem of low accuracy of the existing named entity recognition method.
According to one aspect of the present invention, there is provided a named entity recognition method, including:
acquiring a target text sequence carrying prompt information of a named entity category to be identified, wherein the target text sequence is determined based on the text to be identified;
extracting features of the target text sequence by using a feature extraction neural network of the named entity recognition model after training is completed, and obtaining a word vector sequence for representing semantic feature information of the target text sequence;
and predicting at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the named entity recognition model after training to obtain a target named entity meeting the named entity class prompt information to be recognized, wherein the segment combination vector at least comprises one of a nested named entity vector and a non-nested named entity vector.
Further, the predicting the at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the named entity recognition model after completing training to obtain the target named entity result meeting the named entity class prompt information to be recognized includes:
Determining at least one segment combination vector based on at least one segment start word vector and at least one segment end word vector extracted from the word vector sequence;
predicting each fragment combination vector, and determining the probability of matching the fragment combination vector with the corresponding named entity category prompt information to be identified;
and decoding the segment combination vector with the probability larger than a preset probability threshold value into a target named entity.
Further, the determining at least one segment combination vector based on the at least one segment start word vector and the at least one segment end word vector extracted from the sequence of word vectors comprises:
segment division is carried out on the word vector sequence, and segment start word vectors and segment end word vectors of each segment are extracted;
and carrying out full combination processing on the fragment start word vector and the fragment end word vector to obtain at least one fragment combination vector. Further, before the target text sequence carrying the prompt information of the named entity category to be identified is obtained, the method further includes:
acquiring a text to be recognized and at least one target named entity category, and performing word segmentation on the text to be recognized to obtain the text to be recognized after word segmentation;
Determining at least one named entity class prompt message to be identified based on the target named entity class;
and splicing the to-be-identified named entity category prompt information with the text to be identified after word segmentation processing to obtain at least one target text sequence.
Further, the text to be identified is a medical record text, and before the text to be identified and the at least one target named entity category are obtained, the method further includes:
acquiring a historical diagnosis and treatment information query request of a target object, wherein the historical diagnosis and treatment information query request carries target query information;
determining a named entity category matched with the target query information in a named entity category set as a target named entity category, wherein the named entity category at least comprises one of a disease category, a symptom category, a medicine category, a surgical operation category, an inspection and inspection category and a body part category;
and acquiring the history medical record text of the target object, and determining the history medical record text as a text to be identified.
Further, the method further includes, after predicting at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the named entity recognition model after training to obtain the target named entity meeting the named entity class prompt information to be recognized, performing prediction processing on the segment combination vector, wherein the target named entity includes:
Splicing the target named entity meeting the prompt information of the named entity categories to be identified with the corresponding target named entity category, and taking the spliced result as a historical diagnosis and treatment information query result;
and sending the historical diagnosis and treatment information inquiry result to a terminal sending the historical diagnosis and treatment information inquiry request so that diagnosis and treatment staff evaluate a diagnosis and treatment scheme based on the historical diagnosis and treatment information inquiry result.
Further, before the feature extraction neural network of the trained named entity recognition model is used for carrying out feature extraction on the target text sequence to obtain the word vector sequence for representing the semantic feature information of the target text sequence, the method further comprises:
acquiring a training sample set, wherein each training sample in the training sample set carries named entity category prompt information to be identified;
constructing an initial named entity recognition model, wherein the initial named entity recognition model comprises a feature extraction neural network and an entity category prediction neural network;
and training the initial named entity recognition model by using the training sample set to obtain a named entity recognition model after training.
According to another aspect of the present invention, there is provided a named entity recognition device, including:
the acquisition module is used for acquiring a target text sequence carrying the prompt information of the named entity category to be identified, wherein the target text sequence is determined based on the text to be identified;
the extraction module is used for extracting the characteristics of the target text sequence by utilizing the characteristic extraction neural network of the named entity recognition model after the training is completed, so as to obtain a word vector sequence for representing semantic characteristic information of the target text sequence;
and the prediction module is used for predicting at least one segment combination vector in the word vector sequence by using the entity category prediction neural network of the named entity recognition model after the training to obtain a target named entity meeting the named entity category prompt information to be recognized, wherein the segment combination vector at least comprises one of a nested named entity vector and a non-nested named entity vector.
Further, the prediction module includes:
a first entity extraction unit, configured to determine at least one segment combination vector based on at least one segment start word vector and at least one segment end word vector extracted from the word vector sequence;
The prediction unit is used for carrying out prediction processing on each segment combination vector, determining the probability that the segment combination vector matches the prompt information corresponding to the named entity category to be identified, and decoding the segment combination vector with the probability larger than a preset probability threshold value into a target named entity.
Further, in a specific application scenario, the second entity extraction unit is specifically configured to segment the word vector sequence, and extract a segment start word vector and a segment end word vector of each segment;
and carrying out full combination processing on the fragment start word vector and the fragment end word vector to obtain at least one fragment combination vector. Further, the apparatus further comprises:
the acquisition module is also used for acquiring a text to be identified and at least one target named entity category, and performing word segmentation on the text to be identified to obtain the text to be identified after the word segmentation;
the first determining module is used for determining at least one named entity category prompt message to be identified based on the target named entity category;
and the first splicing module is used for carrying out splicing processing on the prompt information of the named entity category to be identified and the text to be identified after word segmentation processing to obtain at least one target text sequence.
Further, the apparatus further comprises:
the acquisition module is further used for acquiring a historical diagnosis and treatment information query request of the target object, wherein the historical diagnosis and treatment information query request carries target query information;
the second determining module is used for determining a named entity category matched with the target query information in the named entity category set as a target named entity category, wherein the named entity category at least comprises one of a disease category, a symptom category, a medicine category, a surgery operation category, an inspection category and a body part category;
and the third determining module is used for acquiring the history medical record text of the target object and determining the history medical record text as a text to be identified.
Further, the apparatus further comprises:
the second splicing module is used for splicing the target named entities meeting the prompt information of the named entity categories to be identified and the corresponding target named entity categories, and taking the splicing result as a historical diagnosis and treatment information query result;
and the sending module is used for sending the historical diagnosis and treatment information inquiry result to a terminal sending the historical diagnosis and treatment information inquiry request so that diagnosis and treatment staff can evaluate the diagnosis and treatment scheme based on the historical diagnosis and treatment information inquiry result.
Further, the apparatus further comprises:
the acquisition module is further used for acquiring a training sample set, and each training sample in the training sample set carries named entity category prompt information to be identified;
the building module is used for building an initial named entity recognition model, wherein the initial named entity recognition model comprises a feature extraction neural network and an entity category prediction neural network;
and the training module is used for training the initial named entity recognition model by utilizing the training sample set to obtain a named entity recognition model after training.
According to still another aspect of the present invention, there is provided a storage medium having stored therein at least one executable instruction for causing a processor to perform operations corresponding to the named entity recognition method described above.
According to still another aspect of the present invention, there is provided a computer apparatus including: the device comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete communication with each other through the communication bus;
the memory is used for storing at least one executable instruction, and the executable instruction enables the processor to execute the operation corresponding to the named entity identification method.
By means of the technical scheme, the technical scheme provided by the embodiment of the invention has at least the following advantages:
the invention provides a named entity recognition method and device, a storage medium and computer equipment, wherein a target text sequence carrying prompt information of a named entity category to be recognized is firstly obtained, and the target text sequence is determined based on the text to be recognized; extracting features of the target text sequence by using a feature extraction neural network of the named entity recognition model after training is completed, and obtaining a word vector sequence for representing semantic feature information of the target text sequence; and predicting at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the named entity recognition model after training to obtain a target named entity meeting the named entity class prompt information to be recognized, wherein the segment combination vector at least comprises one of a nested named entity vector and a non-nested named entity vector. Compared with the prior art, the method and the device have the advantages that the prompt information is added to the text to be identified, so that the consistency of the fine-tuning training process and the actual application process of the model is improved, the feature extraction capability of the model is guaranteed to be fully exerted, meanwhile, the named entities are extracted and predicted based on the segment splicing mode, all named entities in the nested named entities can be fully extracted, the completeness and the accuracy of named entity extraction are guaranteed, and therefore the accuracy of named entity identification is effectively improved.
The foregoing description is only an overview of the present invention, and is intended to be implemented in accordance with the teachings of the present invention in order that the same may be more clearly understood and to make the same and other objects, features and advantages of the present invention more readily apparent.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to designate like parts throughout the figures. In the drawings:
FIG. 1 shows a flowchart of a named entity recognition method provided by an embodiment of the invention;
FIG. 2 is a flowchart of another named entity recognition method according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a named entity recognition model architecture according to an embodiment of the present invention;
FIG. 4 is a block diagram showing a named entity recognition device according to an embodiment of the present invention;
fig. 5 shows a schematic structural diagram of a computer device according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
The embodiment of the application can acquire and process the related data based on the artificial intelligence technology. Among them, artificial intelligence (AI: artificial Intelligence) is a theory, method, technique and application system that simulates, extends and expands human intelligence using a digital computer or a machine controlled by a digital computer, perceives the environment, acquires knowledge and uses the knowledge to obtain the best result.
Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions.
The method aims at the existing named entity recognition method and is characterized in that the text to be recognized is subjected to feature extraction based on a pre-training model, and classification is carried out based on the extracted features, so that named entity categories of different entities in the text are determined. However, the training process of the pre-training model is completed based on a large amount of labeling resources, and the named entity recognition process is a feature classification process, so that the pre-training process and the downstream fine tuning task are split, the effect of the pre-training model cannot reach expectations in the named entity recognition process, and the problem of low named entity recognition accuracy is caused. The embodiment of the invention provides a named entity identification method, as shown in fig. 1, and the method is applied to computer equipment such as a server and the like for illustration, wherein the server can be an independent server, or can be a cloud server for providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs: content Delivery Network), basic cloud computing services such as big data and artificial intelligent platforms, such as intelligent medical systems, digital medical platforms and the like. The method comprises the following steps:
101. And acquiring a target text sequence carrying the prompt information of the named entity category to be identified.
In the embodiment of the invention, in order to identify the named entity in the text to be identified, the text to be identified is firstly converted into a target text sequence for inputting the trained named entity identification model, namely the target text sequence is determined based on the text to be identified. The text to be identified is a text to be identified by a named entity, and the text can be a text in a medical field, such as an electronic medical record text, an electronic medical academic paper text, a text in an insurance field, such as an electronic insurance check certificate, an electronic health notice, and the like, or an electronic text in other professional and non-professional fields, and the embodiment of the invention is not limited specifically.
It should be noted that, the target text sequence carries the prompt information of the named entity category to be identified, that is, the target text sequence contains the information for prompting the named entity category to be identified currently. By adding the prompt information of the named entity category to be identified in the model input, named entity identification based on prompt learning is realized, so that the subsequent named entity identification process can more fully utilize parameters in the named entity identification model after training, the model prediction effect is improved, and the accuracy of named entity identification is effectively improved.
102. And extracting features of the target text sequence by using a feature extraction neural network of the named entity recognition model after training is completed, so as to obtain a word vector sequence for representing semantic feature information of the target text sequence.
In the embodiment of the invention, the named entity recognition model which is trained comprises a feature extraction neural network for encoding the target text sequence, wherein a basic model of the feature extraction neural network is a pre-training language characterization model, can be a bi-directional encoder (BERT: bidirectional Encoder Representation from Transformers), can be a BERT improved model (Roberta: ARobustly Optimized BERT Pretraining Approach) or can be other pre-training language characterization models, and the embodiment of the invention is not particularly limited. Through the coding of the feature extraction neural network to the target text sequence, feature vectors used for representing semantic information of each word in the target text sequence can be extracted, so that a word vector sequence corresponding to the target text sequence is obtained, and a named entity meeting the category prompt information of the named entity to be identified can be generated based on the represented semantic information in the word vector sequence.
It should be noted that, because the pre-training language characterization model is obtained by training based on a large number of named entity marking resources in the pre-training stage, that is, the pre-training process of the model is a process of extracting semantic information in an entity based on the marked named entity category. In the process of extracting the characteristics of the target text sequence, semantic information extraction is completed on the basis of prompt information learning. The consistency between the pre-training task and the downstream task of the pre-training model is fully ensured, and the feature extraction effect of the pre-training model is fully exerted, so that the accuracy of semantic information extraction is effectively improved, and an accurate feature basis is provided for the subsequent prediction process of the named entity category.
103. And predicting at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the named entity recognition model after training to obtain the target named entity meeting the named entity class prompt information to be recognized.
In the embodiment of the invention, the entity class prediction neural network of the named entity recognition model which is trained is used for extracting entity vectors from the word vector sequence, combining the named entity class prompt information to be recognized, determining whether each extracted entity vector is matched with the named entity class indicated by the named entity class prompt information to be recognized, and if so, determining that the entity corresponding to the current entity vector is the target named entity. The method for extracting the named entity vector may be to predict a segment start position and a segment end position in a word vector sequence based on a span coding mode, so as to determine a segment start word vector of the segment start position and a segment end word vector of the segment end position of at least one of the vector sequences, and determine at least one segment combination vector based on different combinations of the segment start word vector and the segment end word vector.
It should be noted that the segment combination vector at least includes one of a non-nested named entity vector and a nested named entity vector. The nested named entities are named entities of the same entity category or different entity categories. For example, the left ankle fracture, wherein the left ankle fracture and the fracture are named entities of disease category, and the left ankle fracture is a nested named entity. The existing model for named entity recognition is mostly based on a flat layer, and a flattened dimension reduction processing mode based on the flat layer cannot solve the problem of nesting of a plurality of named entities. In the embodiment of the invention, the granularity of entity extraction is refined by extracting the segment start word vector and the segment end word vector in the word vector sequence and combining the segment start word vector and the segment end word vector, so that the full recognition of a plurality of named entities in the nested named entities is realized, the recognition problem of the nested named entities is effectively solved, and the recognition accuracy of the named entities is improved.
In an embodiment of the present invention, for further explanation and limitation, as shown in fig. 2, the predicting, by using the entity class prediction neural network of the trained named entity recognition model, at least one segment combination vector in the word vector sequence to obtain the target named entity result satisfying the named entity class prompt information to be recognized includes:
201. At least one segment combination vector is determined based on at least one segment start word vector and at least one segment end word vector extracted from the sequence of word vectors.
202. And carrying out prediction processing on each fragment combination vector, and determining the probability of matching the fragment combination vector with the corresponding named entity category prompt information to be identified.
203. And decoding the segment combination vector with the probability larger than a preset probability threshold value into a target named entity.
In the embodiment of the invention, in order to completely extract all named entities in the target text sequence, a word vector sequence is divided into a plurality of fragments, word vectors of a starting position and word vectors of an ending position corresponding to each fragment are extracted, and combination is carried out according to the word vectors of all starting positions and the word vectors of the ending positions to obtain a plurality of enumerated fragment combinations. After obtaining the segment combination vectors, in order to confirm the target named entity of the named entity category to be extracted from the segment combination vectors, predicting each segment combination vector as the vector corresponding to the target named entity under the named entity category to be identified, and obtaining the segment combination vector matched with the named entity category prompt information to be identified, thereby obtaining the target named entity by decoding the segment combination vector.
In one embodiment of the present invention, for further explanation and limitation, determining at least one segment combination vector based on at least one segment start word vector and at least one segment end word vector extracted from the word vector sequence includes:
segment division is carried out on the word vector sequence, and segment start word vectors and segment end word vectors of each segment are extracted;
and carrying out full combination processing on the fragment start word vector and the fragment end word vector to obtain at least one fragment combination vector.
In the embodiment of the invention, in order to extract the segment start word vector and the segment end word vector, as shown in fig. 3, the named entity recognition model comprises a feature extraction neural network and an entity category prediction neural network. Wherein the entity class prediction neural network comprises two independent feedforward neural networks (FeedForward Neural Networks) respectively named FFNN-Start and FFNN-End. Wherein FFNN-Start is used for learning semantic representation related to the starting position, and extracting and obtaining segment starting word vectorsFFNN-End is used for learning semantic representation related to End positions, and segment End word vectors are extracted to obtain>Combinations of word vectors of any start position i and any end position j, i.e. full combinations of segment start word vectors and segment end word vectors, are enumerated. In order to realize information interaction between the fragment start word vector and the fragment end word vector, the entity class prediction neural network comprises a double affine Classifier (Biaffine-Classification), and an l×l×c scoring tensor r is created based on the Biaffine-Classification ij The specific formula of the number of named entity categories c indicated by the named entity category prompt information to be identified is as follows:wherein (1)>Fragment combination vector representing (j-i) lengths,> are all learned parameters. And then calculate the current segment combination vector r ij Probability of meeting target named entity category, y ij ＝softmax(r ij ) (2); decoding the segment combination vector with the probability larger than a preset probability threshold into a target named entity, wherein the preset probability threshold can be customized according to specific application scenes and requirements, and the embodiment of the invention is not particularly limited.
It should be noted that, in medical record text, a problem of nesting named entities often occurs, for example, in "using an immunotherapeutic agent against rabies", where both the disease category "rabies" and the drug "immunotherapeutic agent" are involved. The named entity nesting in the medical record text can be effectively extracted based on the span (fragment) encoding and decoding mode, and the entity information is completely extracted. In addition, the semantic information of the starting position and the ending position is interacted based on the double affine attention mechanism, so that more accurate semantic information can be extracted, and the category information of the named entity can be predicted more accurately.
In an embodiment of the present invention, for further explanation and limitation, before the step of obtaining the target text sequence carrying the prompt information of the named entity category to be identified, the method further includes:
acquiring a text to be recognized and at least one target named entity category, and performing word segmentation on the text to be recognized to obtain the text to be recognized after word segmentation;
determining at least one named entity class prompt message to be identified based on the target named entity class;
and splicing the to-be-identified named entity category prompt information with the text to be identified after word segmentation processing to obtain at least one target text sequence.
In the embodiment of the invention, different from the existing named entity recognition method, the categories of all named entities are classified, and each named entity category is marked. And converting the named entity classification problem into a named entity generation problem, namely generating the named entity meeting the target named entity category in the text to be identified. The target named entity category may be one or more than one, for example, in a named entity recognition task, the target named entity category to be recognized may be a plurality of categories such as a disease category, a symptom category, a body part, and the like. And respectively determining a to-be-identified named entity category prompt message based on each target named entity category, and respectively splicing the to-be-identified named entity category prompt message with the text to be identified after word segmentation. For example, the target named entity category is a disease category and a symptom category, the corresponding named entity category prompt information to be identified is disease and symptom in sequence, and then the disease and symptom are respectively spliced with the text to be identified after word segmentation, namely the left ankle fracture for three weeks, and the obtained target text sequence is the left ankle fracture for three weeks of the CLS disease SEP. The specific word segmentation method may be word segmentation based on an open source word segmentation tool jieba, or may be word segmentation based on other word segmentation tools, and the embodiment of the invention is not limited specifically. The downstream named entity task and the upstream pre-training task are integrated better based on the prompt learning method, the characterization capability of the pre-training model is further exerted, and the accuracy of the model is improved.
In an embodiment of the present invention, for further explanation and limitation, before the step of obtaining the text to be identified and the at least one target named entity category, the method further includes:
acquiring a historical diagnosis and treatment information query request of a target object, wherein the historical diagnosis and treatment information query request carries target query information;
determining a named entity category matched with the target query information in the named entity category set as a target named entity category;
and acquiring the history medical record text of the target object, and determining the history medical record text as a text to be identified.
In the embodiment of the invention, the text to be identified is a medical record text, and specifically is an application scene for inquiring historical illness state information and diagnosis and treatment information of a target object (patient). When a doctor diagnoses a patient, the doctor can inquire the historical medical record information of the patient based on the specific requirement on the past medical record information of the patient, and indicates the category of the named entity required to be extracted from the historical medical record of the patient by configuring target inquiry information, so that the named entity content required to be inquired is obtained. For example, if a doctor needs to know whether the left ankle of the patient receives the treatment, the doctor can configure the body part and the named entity category of the operation in the target query content, so that the diagnosis and treatment information of each part of the body of the patient can be obtained quickly. Wherein the named entity category includes at least one of a disease category, a symptom category, a drug category, a surgical procedure category, an inspection category, and a body part category. By matching the target named entity category from the named entity category set based on the target query information, quick and accurate query of the patient information can be realized.
In an embodiment of the present invention, for further explanation and limitation, in the step, the predicting, by using the entity class prediction neural network of the trained named entity recognition model, at least one segment combination vector in the word vector sequence, so as to obtain a target named entity that meets the prompt information of the named entity class to be recognized, the method further includes:
splicing the target named entity meeting the prompt information of the named entity categories to be identified with the corresponding target named entity category, and taking the spliced result as a historical diagnosis and treatment information query result;
and sending the historical diagnosis and treatment information inquiry result to a terminal sending the historical diagnosis and treatment information inquiry request so that diagnosis and treatment staff evaluate a diagnosis and treatment scheme based on the historical diagnosis and treatment information inquiry result.
In the embodiment of the invention, as a plurality of target named entities are available, when the types of the named entities to be queried are more, in order to enable the query result to be more accurately and clearly presented, the target named entities of different types and the corresponding named entity types are spliced. For example, the disease hypertension diabetes left ankle fracture; left ankle of body part blood vessel; surgical operation left ankle fracture repair surgical steel mesh fixation treatment. The method can enable doctors to acquire key information required by diagnosis more quickly, accurately and clearly, and therefore diagnosis evaluation efficiency of the doctors is effectively improved.
In an embodiment of the present invention, for further explanation and limitation, before the step of performing feature extraction on the target text sequence by using the feature extraction neural network of the trained named entity recognition model to obtain the word vector sequence for representing the semantic feature information of the target text sequence, the method further includes:
acquiring a training sample set; constructing an initial named entity recognition model;
and training the initial named entity recognition model by using the training sample set to obtain a named entity recognition model after training.
In the embodiment of the invention, the initial named entity recognition model comprises a feature extraction neural network and an entity category prediction neural network. The feature extraction neural network may be a neural network model constructed based on a pre-training language model, such as Bert and Roberta, and the embodiment of the invention is not particularly limited, and the entity class prediction neural network may be a model structure shown in fig. 3. The training of the model is a fine-tuning training process, in order to ensure the consistency of model training and practical application, the training effect of the model is fully exerted, and each training sample in the training sample set of the model carries the prompt information of the named entity category to be identified, namely the fine-tuning training process of the model is consistent with the practical application process of the model. Meanwhile, by means of the advanced of prompt learning, the task annotation data volume of the named entity can be greatly reduced, the training efficiency of the named entity recognition model is improved, and the labor cost of the model training process is reduced.
The invention provides a named entity recognition method, which comprises the steps of firstly, obtaining a target text sequence carrying prompt information of a named entity category to be recognized, wherein the target text sequence is determined based on the text to be recognized; extracting features of the target text sequence by using a feature extraction neural network of the named entity recognition model after training is completed, and obtaining a word vector sequence for representing semantic feature information of the target text sequence; and predicting at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the named entity recognition model after training to obtain a target named entity meeting the named entity class prompt information to be recognized, wherein the segment combination vector at least comprises one of a nested named entity vector and a non-nested named entity vector. Compared with the prior art, the method and the device have the advantages that the prompt information is added to the text to be identified, so that the consistency of the fine-tuning training process and the actual application process of the model is improved, the feature extraction capability of the model is guaranteed to be fully exerted, meanwhile, the named entities are extracted and predicted based on the segment splicing mode, all named entities in the nested named entities can be fully extracted, the completeness and the accuracy of named entity extraction are guaranteed, and therefore the accuracy of named entity identification is effectively improved.
Further, as an implementation of the method shown in fig. 1, an embodiment of the present invention provides a named entity identifying apparatus, as shown in fig. 4, where the apparatus includes:
the obtaining module 31 is configured to obtain a target text sequence carrying prompt information of a named entity category to be identified, where the target text sequence is determined based on the text to be identified;
the extracting module 32 is configured to perform feature extraction on the target text sequence by using a feature extraction neural network of the trained named entity recognition model, so as to obtain a word vector sequence for representing semantic feature information of the target text sequence;
and the prediction module 33 is configured to perform prediction processing on at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the trained named entity recognition model, so as to obtain a target named entity meeting the named entity class prompt information to be recognized, where the segment combination vector at least includes one of a nested named entity vector and a non-nested named entity vector.
Further, the prediction module 33 includes:
a first entity extraction unit, configured to determine at least one segment combination vector based on at least one segment start word vector and at least one segment end word vector extracted from the word vector sequence;
The prediction unit is used for carrying out prediction processing on each segment combination vector, determining the probability that the segment combination vector matches the prompt information corresponding to the named entity category to be identified, and decoding the segment combination vector with the probability larger than a preset probability threshold value into a target named entity.
Further, in a specific application scenario, the second entity extraction unit is specifically configured to segment the word vector sequence, and extract a segment start word vector and a segment end word vector of each segment;
and carrying out full combination processing on the fragment start word vector and the fragment end word vector to obtain at least one fragment combination vector. Further, the apparatus further comprises:
the acquisition module is also used for acquiring a text to be identified and at least one target named entity category, and performing word segmentation on the text to be identified to obtain the text to be identified after the word segmentation;
the first determining module is used for determining at least one named entity category prompt message to be identified based on the target named entity category;
and the first splicing module is used for carrying out splicing processing on the prompt information of the named entity category to be identified and the text to be identified after word segmentation processing to obtain at least one target text sequence.
Further, the apparatus further comprises:
the obtaining module 31 is further configured to obtain a historical diagnosis and treatment information query request of the target object, where the historical diagnosis and treatment information query request carries target query information;
the second determining module is used for determining a named entity category matched with the target query information in the named entity category set as a target named entity category, wherein the named entity category at least comprises one of a disease category, a symptom category, a medicine category, a surgery operation category, an inspection category and a body part category;
and the third determining module is used for acquiring the history medical record text of the target object and determining the history medical record text as a text to be identified.
Further, the apparatus further comprises:
the second splicing module is used for splicing the target named entities meeting the prompt information of the named entity categories to be identified and the corresponding target named entity categories, and taking the splicing result as a historical diagnosis and treatment information query result;
and the sending module is used for sending the historical diagnosis and treatment information inquiry result to a terminal sending the historical diagnosis and treatment information inquiry request so that diagnosis and treatment staff can evaluate the diagnosis and treatment scheme based on the historical diagnosis and treatment information inquiry result.
Further, the apparatus further comprises:
the obtaining module 31 is further configured to obtain a set of training samples, where each training sample in the set of training samples carries a named entity class prompt message to be identified;
the building module is used for building an initial named entity recognition model, wherein the initial named entity recognition model comprises a feature extraction neural network and an entity category prediction neural network;
and the training module is used for training the initial named entity recognition model by utilizing the training sample set to obtain a named entity recognition model after training.
The invention provides a named entity recognition device, which comprises the steps of firstly, acquiring a target text sequence carrying prompt information of a named entity category to be recognized, wherein the target text sequence is determined based on the text to be recognized; extracting features of the target text sequence by using a feature extraction neural network of the named entity recognition model after training is completed, and obtaining a word vector sequence for representing semantic feature information of the target text sequence; and predicting at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the named entity recognition model after training to obtain a target named entity meeting the named entity class prompt information to be recognized, wherein the segment combination vector at least comprises one of a nested named entity vector and a non-nested named entity vector. Compared with the prior art, the method and the device have the advantages that the prompt information is added to the text to be identified, so that the consistency of the fine-tuning training process and the actual application process of the model is improved, the feature extraction capability of the model is guaranteed to be fully exerted, meanwhile, the named entities are extracted and predicted based on the segment splicing mode, all named entities in the nested named entities can be fully extracted, the completeness and the accuracy of named entity extraction are guaranteed, and therefore the accuracy of named entity identification is effectively improved.
According to one embodiment of the present invention, there is provided a storage medium storing at least one executable instruction for performing the named entity recognition method of any of the above method embodiments.
Fig. 5 is a schematic structural diagram of a computer device according to an embodiment of the present invention, and the specific embodiment of the present invention is not limited to the specific implementation of the computer device.
As shown in fig. 5, the computer device may include: a processor 402, a communication interface (Communications Interface) 404, a memory 406, and a communication bus 408.
Wherein: processor 402, communication interface 404, and memory 406 communicate with each other via communication bus 408.
A communication interface 404 for communicating with network elements of other devices, such as clients or other servers.
The processor 402 is configured to execute the program 410, and may specifically perform relevant steps in the above named entity recognition method embodiment.
In particular, program 410 may include program code including computer-operating instructions.
The processor 402 may be a central processing unit CPU, or a specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement embodiments of the present invention. The one or more processors included in the computer device may be the same type of processor, such as one or more CPUs; but may also be different types of processors such as one or more CPUs and one or more ASICs.
Memory 406 for storing programs 410. Memory 406 may comprise high-speed RAM memory or may also include non-volatile memory (non-volatile memory), such as at least one disk memory.
Program 410 may be specifically operable to cause processor 402 to:
acquiring a target text sequence carrying prompt information of a named entity category to be identified, wherein the target text sequence is determined based on the text to be identified;
extracting features of the target text sequence by using a feature extraction neural network of the named entity recognition model after training is completed, and obtaining a word vector sequence for representing semantic feature information of the target text sequence;
and predicting at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the named entity recognition model after training to obtain a target named entity meeting the named entity class prompt information to be recognized, wherein the segment combination vector at least comprises one of a nested named entity vector and a non-nested named entity vector.
It will be appreciated by those skilled in the art that the modules or steps of the invention described above may be implemented in a general purpose computing device, they may be concentrated on a single computing device, or distributed across a network of computing devices, they may alternatively be implemented in program code executable by computing devices, so that they may be stored in a memory device for execution by computing devices, and in some cases, the steps shown or described may be performed in a different order than that shown or described, or they may be separately fabricated into individual integrated circuit modules, or multiple modules or steps within them may be fabricated into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, but various modifications and variations can be made to the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. A named entity recognition method, comprising:
acquiring a target text sequence carrying prompt information of a named entity category to be identified, wherein the target text sequence is determined based on the text to be identified;
extracting features of the target text sequence by using a feature extraction neural network of the named entity recognition model after training is completed, and obtaining a word vector sequence for representing semantic feature information of the target text sequence;
and predicting at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the named entity recognition model after training to obtain a target named entity meeting the named entity class prompt information to be recognized, wherein the segment combination vector at least comprises one of a nested named entity vector and a non-nested named entity vector.
2. The method according to claim 1, wherein the predicting, by using the trained entity class prediction neural network of the named entity recognition model, at least one segment combination vector in the word vector sequence to obtain the target named entity satisfying the named entity class prompt information to be recognized includes:
determining at least one segment combination vector based on at least one segment start word vector and at least one segment end word vector extracted from the word vector sequence;
predicting each fragment combination vector, and determining the probability of matching the fragment combination vector with the corresponding named entity category prompt information to be identified;
and decoding the segment combination vector with the probability larger than a preset probability threshold value into a target named entity.
3. The method of claim 2, wherein the determining at least one segment combination vector based on at least one segment start word vector and at least one segment end word vector extracted from the sequence of word vectors comprises:
segment division is carried out on the word vector sequence, and segment start word vectors and segment end word vectors of each segment are extracted;
And carrying out full combination processing on the fragment start word vector and the fragment end word vector to obtain at least one fragment combination vector.
4. The method of claim 1, wherein prior to obtaining the target text sequence carrying the named entity category hint information to be identified, the method further comprises:
acquiring a text to be recognized and at least one target named entity category, and performing word segmentation on the text to be recognized to obtain the text to be recognized after word segmentation;
determining at least one named entity class prompt message to be identified based on the target named entity class;
and splicing the to-be-identified named entity category prompt information with the text to be identified after word segmentation processing to obtain at least one target text sequence.
5. The method of claim 4, wherein prior to the obtaining text to be identified and the at least one target named entity category, the method further comprises:
acquiring a historical diagnosis and treatment information query request of a target object, wherein the historical diagnosis and treatment information query request carries target query information;
determining a named entity category matched with the target query information in a named entity category set as a target named entity category, wherein the named entity category at least comprises one of a disease category, a symptom category, a medicine category, a surgical operation category, an inspection and inspection category and a body part category;
And acquiring the history medical record text of the target object, and determining the history medical record text as a text to be identified.
6. The method according to claim 2, wherein the predicting the at least one segment combination vector in the word vector sequence by using the entity class prediction neural network of the trained named entity recognition model, after obtaining the target named entity satisfying the named entity class prompt information to be recognized, further comprises:
splicing the target named entity meeting the prompt information of the named entity categories to be identified with the corresponding target named entity category, and taking the spliced result as a historical diagnosis and treatment information query result;
and sending the historical diagnosis and treatment information inquiry result to a terminal sending the historical diagnosis and treatment information inquiry request so that diagnosis and treatment staff evaluate a diagnosis and treatment scheme based on the historical diagnosis and treatment information inquiry result.
7. The method of claim 1, wherein before feature extraction of the target text sequence using the trained feature extraction neural network of the named entity recognition model to obtain a word vector sequence for characterizing semantic feature information of the target text sequence, the method further comprises:
Acquiring a training sample set, wherein each training sample in the training sample set carries named entity category prompt information to be identified;
constructing an initial named entity recognition model, wherein the initial named entity recognition model comprises a feature extraction neural network and an entity category prediction neural network;
and training the initial named entity recognition model by using the training sample set to obtain a named entity recognition model after training.
8. A named entity recognition device, comprising:
the acquisition module is used for acquiring a target text sequence carrying the prompt information of the named entity category to be identified, wherein the target text sequence is determined based on the text to be identified;
the extraction module is used for extracting the characteristics of the target text sequence by utilizing the characteristic extraction neural network of the named entity recognition model after the training is completed, so as to obtain a word vector sequence for representing semantic characteristic information of the target text sequence;
and the prediction module is used for predicting at least one segment combination vector in the word vector sequence by using the entity category prediction neural network of the named entity recognition model after the training to obtain a target named entity meeting the named entity category prompt information to be recognized, wherein the segment combination vector at least comprises one of a nested named entity vector and a non-nested named entity vector.
9. A storage medium having stored therein at least one executable instruction for causing a processor to perform operations corresponding to the named entity recognition method of any one of claims 1-7.
10. A computer device, comprising: the device comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete communication with each other through the communication bus;
the memory is configured to store at least one executable instruction, where the executable instruction causes the processor to perform operations corresponding to the named entity recognition method according to any one of claims 1 to 7.
CN202310466342.1A 2023-04-19 2023-04-19 Named entity recognition method and device, storage medium and computer equipment Pending CN116484867A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310466342.1A CN116484867A (en) 2023-04-19 2023-04-19 Named entity recognition method and device, storage medium and computer equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310466342.1A CN116484867A (en) 2023-04-19 2023-04-19 Named entity recognition method and device, storage medium and computer equipment

Publications (1)

Publication Number Publication Date
CN116484867A true CN116484867A (en) 2023-07-25

Family

ID=87211547

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310466342.1A Pending CN116484867A (en) 2023-04-19 2023-04-19 Named entity recognition method and device, storage medium and computer equipment

Country Status (1)

Country Link
CN (1) CN116484867A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117077679A (en) * 2023-10-16 2023-11-17 之江实验室 Named entity recognition method and device
CN117540746A (en) * 2023-12-13 2024-02-09 哈尔滨工业大学 Multi-task migration-based crowd-sourced named entity identification personalized prompt fine tuning method and system

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117077679A (en) * 2023-10-16 2023-11-17 之江实验室 Named entity recognition method and device
CN117077679B (en) * 2023-10-16 2024-03-12 之江实验室 Named entity recognition method and device
CN117540746A (en) * 2023-12-13 2024-02-09 哈尔滨工业大学 Multi-task migration-based crowd-sourced named entity identification personalized prompt fine tuning method and system
CN117540746B (en) * 2023-12-13 2024-07-19 哈尔滨工业大学 Multi-task migration-based crowd-sourced named entity identification personalized prompt fine tuning method and system

Similar Documents

Publication Publication Date Title
CN111026842B (en) Natural language processing method, natural language processing device and intelligent question-answering system
CN112016279B (en) Method, device, computer equipment and storage medium for structuring electronic medical record
CN116484867A (en) Named entity recognition method and device, storage medium and computer equipment
US11636936B2 (en) Method and apparatus for verifying medical fact
WO2022068160A1 (en) Artificial intelligence-based critical illness inquiry data identification method and apparatus, device, and medium
CN112908473B (en) Model-based data processing method, device, computer equipment and storage medium
CN112860919B (en) Data labeling method, device, equipment and storage medium based on generation model
CN112667878B (en) Webpage text content extraction method and device, electronic equipment and storage medium
CN113724819B (en) Training method, device, equipment and medium for medical named entity recognition model
CN112906375A (en) Text data labeling method, device, equipment and storage medium
CN111403011B (en) Registration department pushing method, device and system, electronic equipment and storage medium
CN116821373A (en) Map-based prompt recommendation method, device, equipment and medium
CN113821587A (en) Text relevance determination method, model training method, device and storage medium
CN116955646A (en) Knowledge graph generation method and device, storage medium and electronic equipment
CN114676705B (en) Dialogue relation processing method, computer and readable storage medium
CN114780701A (en) Automatic question-answer matching method, device, computer equipment and storage medium
CN111783425B (en) Intention identification method based on syntactic analysis model and related device
CN112182253A (en) Data processing method, data processing equipment and computer readable storage medium
CN116719840A (en) Medical information pushing method based on post-medical-record structured processing
TWM576281U (en) Automatic text labeling system
CN113611427B (en) User portrait generation method, device, equipment and storage medium
CN115859984A (en) Medical named entity recognition model training method, device, equipment and medium
CN113591458B (en) Medical term processing method, device, equipment and storage medium based on neural network
CN114238768A (en) Information pushing method and device, computer equipment and storage medium
CN113408265A (en) Semantic analysis method, device and equipment based on human-computer interaction and storage medium

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