WO2021213314A1

WO2021213314A1 - Data processing method and device, and computer readable storage medium

Info

Publication number: WO2021213314A1
Application number: PCT/CN2021/088090
Authority: WO
Inventors: 冷莹
Original assignee: 北京京东拓先科技有限公司
Priority date: 2020-04-20
Filing date: 2021-04-19
Publication date: 2021-10-28
Also published as: CN112307215B; CN112307215A

Abstract

The present invention relates to the field of information technology. Provided are a data processing method and device, and a computer readable storage medium. The data processing method comprises: performing analysis on a query sentence input by a user, and obtaining a medical entity to be searched and a medical relationship to be searched; querying a pre-established medical knowledge graph using the medical entity to be searched and the medical relationship to be searched, and obtaining a target medical entity, wherein nodes of the medical knowledge graph are respective medical entities, sides of the medical knowledge graph are medical relationships between the respective medical entities, the respective medical entities include the medical entity to be searched and the target medical entity, and the medical relationships between the respective medical entities include the medical relationship to be searched; and outputting an answer sentence according to the target medical entity. The method of the present invention enables automatic answering for medical knowledge, thereby automatically providing a user with a scientifically reasonable medical suggestion and medical recommendation.

Description

Data processing method, device and computer readable storage medium

Cross-references to related applications

This application is based on the application with the CN application number 202010311319.1 and the filing date of April 20, 2020, and claims its priority. The disclosure of the CN application is hereby incorporated into this application as a whole.

Technical field

The present disclosure relates to the field of information technology, and in particular to a data processing method, device, and computer-readable storage medium.

Background technique

Question answering systems usually integrate different research fields such as natural language processing, information retrieval, and databases. After the user enters the question to be asked, the question and answer system analyzes and processes the question entered by the user, performs various operations such as information retrieval or database query, and returns the answer that the user needs. Question answering systems include automatic question answering systems based on search engines and community-based question answering systems.

Summary of the invention

A technical problem solved by the present disclosure is how to automatically provide users with scientific and reasonable medical advice and medical recommendations.

According to one aspect of the embodiments of the present disclosure, a computer-executed data processing method is provided, including: parsing an inquiry sentence input by a user to obtain a medical entity to be queried and a medical relationship to be queried; using the medical care to be queried Entity and medical relationship to be queried query the pre-created medical knowledge graph to obtain the target medical entity. Among them, the nodes of the medical knowledge graph are each medical entity, and the edges of the medical knowledge graph are the medical relationship between each medical entity, and each medical entity Including the medical entity to be queried and the target medical entity. The medical relationship between each medical entity includes the medical relationship to be queried; according to the target medical entity, a response sentence is output.

In some embodiments, parsing the query sentence input by the user to obtain the medical entity to be queried and the medical relationship to be queried includes: identifying the medical entity to be queried included in the query sentence input by the user; and identifying the query input by the user At least one keyword contained in the query sentence; the medical relationship to be queried is determined according to the medical entity to be queried and at least one keyword of the query sentence.

In some embodiments, the query sentence includes multiple keywords, and the medical relationship to be queried includes the initial medical relationship and the association relationship between each keyword. According to the medical entity to be queried and at least one keyword of the query sentence, Determining the medical relationship to be queried includes: determining the initial medical relationship to be queried according to the medical entity to be queried and the first keyword of the query sentence; determining the association between each keyword according to the keywords of the query sentence relation.

In some embodiments, identifying the medical entity to be queried included in the query sentence input by the user includes: using a two-way long and short-term memory network and a conditional random field to recognize the medical entity to be queried included in the query sentence input by the user.

In some embodiments, the parsing of the query sentence input by the user to obtain the medical entity to be queried and the medical relationship to be queried includes: extracting the medical characteristic vocabulary contained in the query sentence to identify what the query sentence contains The medical entity to be queried.

In some embodiments, parsing the query sentence input by the user to obtain the medical entity to be queried and the medical relationship to be queried includes: understanding the semantics expressed by the query sentence through a relationship recognition ruler, and matching medical knowledge The relationship type in the graph, so as to obtain the medical relationship corresponding to the medical knowledge graph.

In some embodiments, using the medical entity to be queried and the medical relationship to be queried to query the pre-created medical knowledge graph includes: using the medical entity to be queried and the medical relationship to be queried to generate a graph database query statement; using the graph database to query Sentence queries the pre-created medical knowledge graph.

In some embodiments, using the medical entity to be queried and the medical relationship to be queried to generate a graph database query sentence includes: combining the medical entity to be queried and the medical relationship to be queried to form a logical expression of the query sentence Formula; convert the logical expression into a graph database query statement corresponding to the query statement based on the medical knowledge graph.

In some embodiments, the query statement of the graph database includes a one-degree query statement or a multiple-degree query statement.

In some embodiments, the graph database query statement includes a Cypher graph database query statement.

In some embodiments, outputting the response sentence according to the target medical entity includes: determining the sentence template of the response sentence according to the medical relationship to be queried; filling the sentence template of the response sentence with the medical entity to be queried and the target medical entity to generate the response Sentence; output the response sentence.

In some embodiments, filling the medical entity to be queried and the target medical entity into the sentence template of the response sentence, and generating the response sentence includes: according to the query result from the medical knowledge graph, combining the sentence template of the medical knowledge and the response sentence And natural language that conforms to the user’s expression habits to generate response sentences.

In some embodiments, the data processing method further includes: using each medical entity and the medical relationship between each medical entity to create a medical knowledge graph.

According to another aspect of the embodiments of the present disclosure, there is provided a data processing device, including: a sentence parsing module configured to parse a query sentence input by a user to obtain a medical entity to be queried and a medical relationship to be queried; The knowledge graph query module is configured to use the medical entity to be queried and the medical relationship to be queried to query the pre-created medical knowledge graph to obtain the target medical entity. Among them, the nodes of the medical knowledge graph are each medical entity, and the edges of the medical knowledge graph For the medical relationship between various medical entities, each medical entity includes the medical entity to be queried and the target medical entity, and the medical relationship between each medical entity includes the medical relationship to be queried; the sentence output module is configured to be based on the target medical entity , Output the response sentence.

In some embodiments, the sentence analysis module is configured to: identify the medical entity to be queried included in the query sentence entered by the user; identify at least one keyword included in the query sentence entered by the user; and according to the medical entity to be queried and the query At least one keyword of the query sentence determines the medical relationship to be queried.

In some embodiments, the query sentence includes multiple keywords, and the medical relationship to be queried includes the initial medical relationship and the association relationship between each keyword; the sentence parsing module is configured to: according to the medical entity to be queried and the query The first keyword of the sentence determines the initial medical relationship to be queried; according to each keyword of the query sentence, the association relationship between each keyword is determined.

In some embodiments, the sentence parsing module is configured to use a two-way long and short-term memory network and a conditional random field to identify the medical entity to be queried included in the query sentence input by the user.

In some embodiments, the knowledge graph query module is configured to: use the medical entity to be queried and the medical relationship to be queried to generate a graph database query statement; use the graph database query statement to query a pre-created medical knowledge graph.

In some embodiments, the sentence output module is configured to: determine the sentence template of the response sentence according to the medical relationship to be queried; fill the sentence template of the response sentence with the medical entity to be queried and the target medical entity to generate the response sentence; output Response sentence.

In some embodiments, the data processing device further includes a graph creation module configured to use each medical entity and the medical relationship between each medical entity to create a medical knowledge graph.

According to another aspect of the embodiments of the present disclosure, there is provided a data processing device, including: a memory; and a processor coupled to the memory, the processor being configured to execute the aforementioned data processing method based on instructions stored in the memory .

According to another aspect of the embodiments of the present disclosure, there is provided a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the instructions are executed by a processor to implement the aforementioned data processing method.

The present disclosure uses the knowledge graph as the answer source of the automatic question and answer, realizes the automatic question and answer of medical knowledge, can more accurately understand the user's question described in natural language, analyze the user's true intention, and return to the user more accurate and professional answers, thereby Provide users with scientific and reasonable medical advice and medical recommendations.

Through the following detailed description of exemplary embodiments of the present disclosure with reference to the accompanying drawings, other features and advantages of the present disclosure will become clear.

Description of the drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only These are some embodiments of the present disclosure. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative labor.

Fig. 1 shows a schematic flowchart of a data processing method according to some embodiments of the present disclosure.

Fig. 2 shows a schematic diagram of the process of parsing the query sentence input by the user.

Figure 3 shows a network model composed of a two-way long and short-term memory network and a conditional random field.

Fig. 4 exemplarily shows an application example of the automated medical question answering system.

Figure 5 shows a schematic structural diagram of a data processing device according to some embodiments of the present disclosure.

Fig. 6 shows a schematic structural diagram of a data processing device according to other embodiments of the present disclosure.

Detailed ways

The technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all the embodiments. The following description of at least one exemplary embodiment is actually only illustrative, and in no way serves as any limitation to the present disclosure and its application or use. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present disclosure.

The inventor has conducted in-depth analysis and research on various automatic question answering systems.

In a question and answer system based on a search engine, users can enter keywords or sentences, and the search engine will return a sorted collection of related documents. Nowadays, this kind of question answering system has been difficult to meet the needs of users. Automatic question answering systems based on search engines have two major shortcomings: on the one hand, it is difficult to guarantee the data quality of the documents feedback to users, because the search engine itself cannot control the data source and data. By itself, search engines cannot guarantee the accuracy and comprehensiveness of the data; on the other hand, users ask questions in order to get an accurate and direct answer, but a series of documents given by search engines require users to read carefully to find the answers they need. .

The community-based question answering system is mainly based on the community platform on the Internet. Community platforms often ask users to ask questions, other users give answers to the questions, and other users can choose which answer they agree with by voting or liking. Therefore, the main technology adopted by the community-based question answering system is to match the new question with all existing questions in the community, and select the existing question and answer pair with the highest similarity, which contains a list of possible answers to the new question. The disadvantage of community-based question answering systems is that they cannot judge and verify the accuracy of the data, because the answers provided by users are often not accurate and complete. At the same time, community-based question and answer systems cannot solve long-tail questions, that is, questions that no one has asked cannot match possible answers.

In order to realize a more intelligent medical question-and-answer system, the present disclosure provides a data processing method, which automatically provides users with scientific and reasonable medical advice and medical recommendations based on the medical knowledge graph.

First introduce the related concepts of the knowledge graph.

Knowledge graph: Knowledge graph is a technical method that uses graph models to describe the relationship between knowledge and modeling things. It visually displays complex domain knowledge through data mining, information processing, knowledge measurement and graph drawing , Reveals the dynamic development law of the knowledge field, and can provide practical and valuable reference for field research. In essence, the knowledge graph is to systemize and relational industry knowledge, and visually display the knowledge in the form of graphs.

Named entity: A named entity generally refers to an entity with a specific meaning or strong referentiality in the text, which usually includes the name of a person, place name, organization name, date and time, proper nouns, etc.

Named entity recognition: Named entity recognition is to extract named entities from unstructured input text, and can identify more categories of named entities according to business needs, such as product names, models, prices, etc.

Entity alignment: Entity alignment is also called entity matching, which refers to finding the same entity in the real world for each entity in the knowledge base of heterogeneous data sources.

Knowledge fusion: that is, merging two knowledge graphs, fusing description information about the same entity or concept from multiple sources.

Before using the data processing method provided by the present disclosure to realize automatic question and answer of medical knowledge, each medical entity and the medical relationship between each medical entity can be used to create a medical knowledge graph. Among them, the nodes of the medical knowledge graph are various medical entities, and the edges of the medical knowledge graph are the medical relationships between various medical entities. Each medical entity includes the medical entity to be queried and the target medical entity. The medical relationship between each medical entity includes The medical relationship to be queried.

For example, the medical entity can be specific disease names such as cold, fever, emphysema, etc., can be specific drug names such as Contec, or specific examination names such as B-ultrasound and fluoroscopy, or specific Food name or symptom name. The medical relationship can specifically be the corresponding relationship between the disease and the medicine, the corresponding relationship between the disease and the examination, the corresponding relationship between the disease and the food, the corresponding relationship between the disease and the symptom, and so on.

The medical knowledge graph can be used as an accurate knowledge base to provide business knowledge in the medical field. Compared with related technologies, the medical knowledge map can be greatly improved in three aspects. First, the results are more accurate. Due to linguistic phenomena such as polysemous words, multiple words and one meaning, and multiple data sources, the knowledge graph ensures the accuracy and comprehensiveness of its own data through entity alignment and knowledge fusion technology. Second, the data relevance in the knowledge graph is stronger. Due to the characteristics of the graph data structure, the knowledge graph can be easily related from one entity to other entities, and can more deeply understand the information users want to search. Third, the knowledge graph can provide search results that contain complete knowledge and relationships, so users can discover a wealth of unknown information through the knowledge graph.

The following describes some embodiments of the data processing method of the present disclosure with reference to FIG. 1.

Fig. 1 shows a schematic flowchart of a data processing method according to some embodiments of the present disclosure. As shown in Fig. 1, this embodiment includes steps S101 to S103.

In step S101, the query sentence input by the user is parsed to obtain the medical entity to be queried and the medical relationship to be queried.

For example, if the query sentence entered by the user is "What medicine should you take for a cold", the query sentence entered by the user can be parsed through entity alignment, and the medical entity to be queried "Cold" can be obtained, and the medical relationship to be queried can be obtained. Correspondence between diseases and drugs".

In step S102, use the medical entity to be queried and the medical relationship to be queried to query the pre-created medical knowledge graph to obtain the target medical entity.

When querying the medical knowledge graph, the medical entity to be queried and the medical relationship to be queried can be used to generate the query statement of the graph database. Then, use the graph database query sentence to query the pre-created medical knowledge graph.

In step S103, a response sentence is output according to the target medical entity.

When outputting the response sentence, the simpler way is to directly output the target medical entity, such as "Contec". In some embodiments, the sentence template of the response sentence may be determined first according to the medical relationship to be queried. Then fill in the sentence template of the response sentence with the medical entity to be queried and the target medical entity to generate the response sentence, and finally output the response sentence. The latter way of outputting response sentences can provide users with a more complete user experience.

For example, the pre-set sentence template "___吃___" can be determined based on the "correspondence between diseases and drugs". Then, fill in the sentence template according to the position correspondence of the medical entity in the sentence template and generate and output the response sentence "Cold cold eat Kangtaike".

In this embodiment, the knowledge graph is used as the answer source of the automatic question and answer, which realizes the automatic question and answer of medical knowledge, can more accurately understand the user's question described in natural language, analyze the user's true intention, and return to the user more accurate and professional answers. So as to provide users with scientific and reasonable medical advice and medical recommendations. This embodiment helps to standardize medical question-and-answer behaviors, improve the service efficiency and service quality of medical question-and-answer, and adjust the supply relationship of the medical industry.

The following describes how to parse the query sentence input by the user in conjunction with Figure 2.

Fig. 2 shows a schematic diagram of the process of parsing the query sentence input by the user. As shown in FIG. 2, this embodiment includes steps S2011 to S2013.

In step S2011, the medical entity to be queried included in the query sentence input by the user is identified.

In some embodiments, a two-way long-term short-term memory network and a conditional random field can be used to identify the medical entity to be queried included in the query sentence input by the user. Long short-term memory network (LSTM for short) is a kind of recurrent neural network, first proposed by Hochreiter and Schmidhuber in 1997. The original design was to solve the long-term dependence problem in recurrent neural network and make remembering long-term information become the default of neural network. Behavior, and the bidirectional long-term short-term memory network (Bi-LSTM for short) can not only learn the above information, but also the following information. Conditional random fields (CRF for short) can be used to construct a conditional probability distribution model for a given set of input random variables and another set of output random variables.

Figure 3 shows a network model composed of a two-way long and short-term memory network and a conditional random field. Adding a CRF layer to Bi-LSTM is to obtain the global optimal output sequence, which is equivalent to reusing Bi-LSTM information. If only the Bi-LSTM network is used, for the labeling sequence of a sentence, the network model will select the sequence with the most occurrences of each word, that is, it will not consider whether the context is reasonable, and with a layer of CRF filtering, the network model will consider the In this context, the label with the greatest probability of a word appears, so that a higher accuracy rate can be obtained. The advantage of Bi-LSTM is that it can learn the dependencies between input sequences in both directions. In the training process, Bi-LSTM can automatically extract the features of the input sequence according to the entity. The advantage of CRF is that it can model hidden states to learn the characteristics of state sequences. Therefore, adding a layer of CRF after Bi-LSTM can obtain the advantages of both at the same time, thereby making the task of named entity recognition more accurate and comprehensive, with good robustness and field openness, and targeting multiple fields and different entities. The category recognition task has good portability. Using the combination of two-way long and short-term memory network and conditional random field, it can mine the deep semantic information and hidden information that cannot be expressed on the surface of the text, and more closely represent the semantic vector of the text, which greatly improves the performance of named entity recognition.

In the process of medical entity identification, the name of the disease itself may appear in the query sentence, or the common name of the disease may appear. Regardless of whether it is the name of the disease itself or the common name of the disease, the identification result of the medical entity can ultimately correspond to the same medical entity in the medical knowledge graph.

In step S2012, at least one keyword contained in the query sentence input by the user is identified.

The keywords can be collective names of medical entities, such as "disease", "food", "check", "medicine", "symptom" and so on. Those skilled in the art should understand that the number of keywords may be one or more. For example, in the query sentence "What medicine should you take for a cold", the keyword is "medicine" and the number of keywords is one; while in the query sentence "How much do you take for a cold?", the keyword is "medicine" , "Money", the number of keywords is two.

In step S2013, the medical relationship to be queried is determined according to the medical entity to be queried and at least one keyword of the query sentence.

For example, the keyword "disease" belonging to the medical entity "cold" to be queried is "disease", and the keyword "disease" belonging to the medical entity "cold" to be queried and the keyword "medicine" in the query sentence can be determined The medical relationship to be queried is "correspondence between disease and drug".

In some embodiments, the query sentence includes multiple keywords. In this case, the medical relationship to be queried includes the initial medical relationship and the association relationship between each keyword. Therefore, the initial medical relationship to be queried can be determined according to the medical entity to be queried and the first keyword of the query sentence; the association relationship between the keywords can be determined according to the keywords of the query sentence.

For example, in the query sentence "how much does it cost to take medicine for a cold", the keywords are "medicine" and "money", and the number of keywords is two. Then, according to the medical entity to be queried "cold" and the first keyword "medicine", it can be determined that the initial medical relationship to be queried is "correspondence between disease and drug". According to the first keyword "medicine" and the first keyword For the two keywords "money", the correlation between the keywords can be determined as the "correspondence between drugs and amount".

The following describes an application example of an automated medical question-and-answer system generated by using the data processing method provided by the present disclosure with reference to FIG. 4.

Fig. 4 exemplarily shows an application example of the automated medical question answering system. The automated medical question and answer system is mainly composed of three modules: semantic analysis (including medical named entity recognition and medical relationship recognition), query sentence construction and conversion, and answer reply ruler.

The semantic analysis module includes two parts: medical named entity recognition and medical relationship recognition. Medical named entity recognition aims to extract the medical characteristic vocabulary contained in the user's query sentence, and can extract more than 20 kinds of diseases, symptoms, drugs, examinations, etc. Type of medical entity. That is, by extracting the medical characteristic vocabulary contained in the questioning sentence, the medical entity to be queried contained in the questioning sentence is identified.

Medical relationship recognition uses the relationship recognition ruler to understand the semantics of query sentences, match the relationship types in the medical knowledge graph, and obtain the medical relationship corresponding to the medical knowledge graph, that is, the medical relationship to be queried. Through medical named entity recognition and medical relationship recognition, the medical entity and medical relationship information of the user's question is obtained, and the entity and relationship information is sent to the query sentence construction and transformation module.

The query sentence construction and transformation module, based on the medical entity and medical relationship of the user query sentence obtained by medical entity recognition and medical relationship recognition, combines the medical entity and medical relationship to form the logical expression of the user question, and then express the logic The formula is converted into a graph database query statement corresponding to the user question based on the medical knowledge graph, such as a Cypher graph database query statement. The established medical knowledge graph is queried through the query sentence of the graph database (specifically including one-degree or multiple-degree query, etc.). The medical knowledge graph gives the result of the query, that is, the corresponding medical knowledge, which is sent to the answer reply ruler module.

The answer reply ruler module receives the results of the medical knowledge graph query, and combines medical knowledge, answer templates and natural language that conforms to the user's expression habits to obtain the final answer and reply to the user. That is, according to the results of the query from the medical knowledge graph, combined with the medical knowledge, the sentence template of the response sentence, and the natural language that conforms to the user's expression habits, the response sentence is generated and output to the user.

In the following, some embodiments of the data processing device of the present disclosure will be described with reference to FIG. 5.

Figure 5 shows a schematic structural diagram of a data processing device according to some embodiments of the present disclosure. As shown in FIG. 5, the data processing device 50 in this embodiment includes: a sentence parsing module 501 configured to parse the query sentence input by the user to obtain the medical entity to be queried and the medical relationship to be queried; a knowledge graph The query module 502 is configured to use the medical entity to be queried and the medical relationship to be queried to query the pre-created medical knowledge graph to obtain the target medical entity. The nodes of the medical knowledge graph are each medical entity, and the edges of the medical knowledge graph are The medical relationship between various medical entities, each medical entity includes the medical entity to be queried and the target medical entity, and the medical relationship between each medical entity includes the medical relationship to be queried; the sentence output module 503 is configured to be based on the target medical entity , Output the response sentence.

In some embodiments, the sentence parsing module 501 is configured to: identify the medical entity to be queried included in the query sentence input by the user; recognize at least one keyword contained in the query sentence entered by the user; according to the medical entity to be queried and At least one keyword of the query sentence determines the medical relationship to be queried.

In some embodiments, the query sentence includes multiple keywords, and the medical relationship to be queried includes the initial medical relationship and the association relationship between each keyword; the sentence parsing module 501 is configured to: according to the medical entity to be queried and the query The first keyword of the query sentence determines the initial medical relationship to be queried; according to each keyword of the query sentence, the association relationship between each keyword is determined.

In some embodiments, the sentence parsing module 501 is configured to use a two-way long and short-term memory network and a conditional random field to identify the medical entity to be queried included in the query sentence input by the user.

In some embodiments, the knowledge graph query module 502 is configured to: use the medical entity to be queried and the medical relationship to be queried to generate a graph database query statement; use the graph database query statement to query a pre-created medical knowledge graph.

In some embodiments, the sentence output module 503 is configured to: determine the sentence template of the response sentence according to the medical relationship to be queried; fill the sentence template of the response sentence with the medical entity to be queried and the target medical entity to generate the response sentence; The response sentence is output.

In some embodiments, the data processing device 50 further includes a graph creating module 500 configured to use each medical entity and the medical relationship between each medical entity to create a medical knowledge graph.

In the following, other embodiments of the data processing device of the present disclosure will be described with reference to FIG. 6.

Fig. 6 shows a schematic structural diagram of a data processing device according to other embodiments of the present disclosure. As shown in FIG. 6, the data processing device 60 of this embodiment includes a memory 610 and a processor 620 coupled to the memory 610. The processor 620 is configured to execute any of the foregoing implementations based on instructions stored in the memory 610. The data processing method in the example.

Among them, the memory 610 may include, for example, a system memory, a fixed non-volatile storage medium, and the like. The system memory stores, for example, an operating system, an application program, a boot loader (Boot Loader), and other programs.

The data processing device 60 may also include an input and output interface 630, a network interface 640, a storage interface 650, and the like. These

interfaces

630, 640, 650, and the memory 610 and the processor 620 may be connected via a bus 660, for example. Among them, the input and output interface 630 provides a connection interface for input and output devices such as a display, a mouse, a keyboard, and a touch screen. The network interface 640 provides a connection interface for various networked devices. The storage interface 650 provides a connection interface for external storage devices such as SD cards and U disks.

The present disclosure also includes a computer-readable storage medium on which computer instructions are stored, and when the instructions are executed by a processor, the data processing method in any of the foregoing embodiments is implemented.

The present disclosure is described with reference to flowcharts and/or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present disclosure. It should be understood that each process and/or block in the flowchart and/or block diagram, and the combination of processes and/or blocks in the flowchart and/or block diagram can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing equipment to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing equipment are generated It is a device that realizes the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.

These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing equipment to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction device. The device implements the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.

These computer program instructions can also be loaded on a computer or other programmable data processing equipment, so that a series of operation steps are executed on the computer or other programmable equipment to produce computer-implemented processing, so as to execute on the computer or other programmable equipment. The instructions provide steps for implementing the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.

The above descriptions are only preferred embodiments of the present disclosure and are not intended to limit the present disclosure. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present disclosure shall be included in the protection of the present disclosure. Within range.

Claims

A computer-executed data processing method includes:

Analyze the query sentence entered by the user to obtain the medical entity to be queried and the medical relationship to be queried;

Use the medical entity to be queried and the medical relationship to be queried to query the pre-created medical knowledge graph to obtain the target medical entity. The node of the medical knowledge graph is each medical entity, and the edge of the medical knowledge graph is the medical treatment between each medical entity. Relationship, each medical entity includes the medical entity to be queried and the target medical entity, and the medical relationship between each medical entity includes the medical relationship to be queried;

According to the target medical entity, a response sentence is output.
The data processing method according to claim 1, wherein the parsing the query sentence input by the user to obtain the medical entity to be queried and the medical relationship to be queried comprises:

Identify the medical entity to be queried contained in the query sentence entered by the user;

Identify at least one keyword contained in the query sentence entered by the user;

According to the medical entity to be queried and at least one keyword of the query sentence, the medical relationship to be queried is determined.
The data processing method according to claim 2, wherein the query sentence includes a plurality of keywords, the medical relationship to be queried includes an initial medical relationship and an association relationship between each keyword, and the medical entity to be queried and At least one key word of the query sentence to determine the medical relationship to be queried includes:

Determine the initial medical relationship to be queried according to the medical entity to be queried and the first keyword of the query sentence;

According to each keyword of the query sentence, determine the association relationship between each keyword.
The data processing method according to claim 2, wherein said identifying the medical entity to be queried included in the query sentence input by the user comprises:

A bidirectional long-term short-term memory network and a conditional random field are used to identify the medical entity to be queried contained in the query sentence entered by the user.
The data processing method according to claim 1, wherein the parsing the query sentence input by the user to obtain the medical entity to be queried and the medical relationship to be queried comprises:

By extracting the medical characteristic vocabulary contained in the query sentence, the medical entity to be queried contained in the query sentence is identified.
The data processing method according to claim 1, wherein the parsing the query sentence input by the user to obtain the medical entity to be queried and the medical relationship to be queried comprises:

The relationship recognition ruler is used to understand the semantics of the query sentence, match the relationship type in the medical knowledge graph, and obtain the medical relationship corresponding to the medical knowledge graph.
The data processing method according to claim 1, wherein the query of the medical knowledge graph created in advance by using the medical entity to be queried and the medical relationship to be queried comprises:

Use the medical entity to be queried and the medical relationship to be queried to generate query statements for the graph database;

Use graph database query sentences to query the pre-created medical knowledge graph.
The data processing method according to claim 7, wherein said generating a graph database query sentence using the medical entity to be queried and the medical relationship to be queried comprises:

Combine the medical entity to be queried and the medical relationship to be queried to form a logical expression of the query sentence;

The logical expression is converted into a graph database query statement corresponding to the query statement based on the medical knowledge graph. 9. The data processing method according to claim 7, wherein the graph database query statement comprises a one-degree query statement or a multiple-degree query statement.
8. The data processing method according to claim 7, wherein the graph database query sentence comprises a Cypher graph database query sentence.
The data processing method according to claim 1, wherein said outputting a response sentence according to the target medical entity comprises:

According to the medical relationship to be queried, determine the sentence template of the response sentence;

Fill in the sentence template of the response sentence with the medical entity to be queried and the target medical entity to generate the response sentence;

The response sentence is output.
The data processing method according to claim 11, wherein said filling the medical entity to be queried and the target medical entity into the sentence template of the response sentence, and generating the response sentence comprises:

According to the query results from the medical knowledge graph, the response sentence is generated by combining the medical knowledge, the sentence template of the response sentence, and the natural language that conforms to the user's expression habits.
The data processing method according to claim 1, further comprising:

Use each medical entity and the medical relationship between each medical entity to create a medical knowledge graph.
A data processing device includes:

The sentence parsing module is configured to parse the query sentence input by the user to obtain the medical entity to be queried and the medical relationship to be queried;

The knowledge graph query module is configured to use the medical entity to be queried and the medical relationship to be queried to query the pre-created medical knowledge graph to obtain the target medical entity. Among them, the nodes of the medical knowledge graph are each medical entity, and the edges of the medical knowledge graph For the medical relationship between various medical entities, each medical entity includes the medical entity to be queried and the target medical entity, and the medical relationship between each medical entity includes the medical relationship to be queried;

The sentence output module is configured to output response sentences according to the target medical entity.
A data processing device includes:

Memory; and

A processor coupled to the memory, and the processor is configured to execute the data processing method according to any one of claims 1 to 13 based on instructions stored in the memory.
A computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions that, when executed by a processor, implement the data processing method according to any one of claims 1 to 13.