CN114090789A - Intelligent multi-round interaction system for traditional Chinese medicine health maintenance based on knowledge graph - Google Patents
Intelligent multi-round interaction system for traditional Chinese medicine health maintenance based on knowledge graph Download PDFInfo
- Publication number
- CN114090789A CN114090789A CN202111369070.0A CN202111369070A CN114090789A CN 114090789 A CN114090789 A CN 114090789A CN 202111369070 A CN202111369070 A CN 202111369070A CN 114090789 A CN114090789 A CN 114090789A
- Authority
- CN
- China
- Prior art keywords
- chinese medicine
- traditional chinese
- entity
- graph
- round
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06F16/33—Querying
- G06F16/338—Presentation of query results
-
- 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 Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
本发明属于人工智能知识图谱领域，提供了一种基于知识图谱的中医养生智能多轮交互系统，因为具备记忆能力可以主动关联与用户交互的内容，用户向系统输入临床表现后，系统会根据用户输入的临床表现的特征，主动向用户提问相关主诉表现，用户根据系统提问的问题输入反馈内容，智能交互系统使用这样的方式，经过与用户多轮交互后，最终给予用户有较高准确率的中医养生建议。本发明的交互系统采用多轮对话的交互形式，在终端用户的体验上和中医养生咨询问答的结论准确度上均高于普通单轮交互的问答系统。本发明针对前后文中识别到的实体进行实体链接，结合前后文的文本信息基于记忆序列的实体链接模型来解决多轮会话中语句的实体省略和指代消解。
The invention belongs to the field of artificial intelligence knowledge graphs, and provides an intelligent multi-round interaction system for traditional Chinese medicine health care based on knowledge graphs. Because it has memory ability, it can actively associate with the content of interaction with users. After the user inputs clinical manifestations into the system, the system will The characteristics of the input clinical manifestations, actively ask the user about the relevant main complaints, and the user inputs the feedback content according to the questions asked by the system. The intelligent interactive system uses this method. After multiple rounds of interaction with the user, it finally gives the user a higher accuracy rate. Chinese medicine health advice. The interactive system of the present invention adopts the interactive form of multiple rounds of dialogue, and is higher than the common single-round interactive question-and-answer system in terms of the end user's experience and the accuracy of the conclusion of the traditional Chinese medicine health care consultation question and answer. The present invention performs entity linking for entities identified in the context, and combines the text information of the context and the entity link model based on memory sequences to solve entity omission and metaphorical resolution of sentences in multiple rounds of conversations.
Description
Technical Field
The invention belongs to the field of artificial intelligence knowledge maps, and particularly relates to a traditional Chinese medicine health-preserving intelligent multi-round interactive system based on knowledge maps.
Background
The state establishes a management system which accords with the characteristics of the traditional Chinese medicine, so that the traditional Chinese medicine can fully play an important role in the field of Chinese medicine and health, and the traditional Chinese medicine health preservation is an important component of the traditional Chinese medicine in the field of Chinese medicine and health. With the popularization of the mobile internet, people have become a normalized application scene for consulting the traditional Chinese medicine health-preserving related knowledge through mobile terminals. The intelligent interactive system in the traditional Chinese medicine health preservation field is mainly a question-answering system, the question-answering systems are mostly single-wheel intelligent interactive systems, namely a user asks a question, the system returns an answer according to the question, when the user asks the question again, the previous question-asking system cannot be associated, namely the system has no memory capacity. This results in a large number of repeated questions being asked by the user, which may or may not result in an accurate answer to the question.
At present, some question-answering systems adopt a multi-round interaction mode, but in the multi-round interaction process, because a conventional conversation generation model splices front and back interactive texts and models the spliced texts, the front and back texts are composed of a plurality of word sentences, and the spliced texts are long, the front and back text vector information obtained by the modeling mode is imperfect because the information of the front text is scattered, the conversation between a user and the system has the situations of reference resolution and entity omission, and the situations cause great troubles to entity linking, so that the user cannot obtain accurate answers to the problems.
Disclosure of Invention
The invention is made to solve the above problems, and aims to provide a traditional Chinese medicine health preserving intelligent multi-round interactive system based on a knowledge graph.
The invention provides a traditional Chinese medicine health preserving intelligent multi-round interactive system based on a knowledge graph, which is characterized by comprising the following components: the entity extraction module is used for extracting entities in the clinical information input by the user in each round of interaction; the entity linking module is used for carrying out reference resolution and entity disambiguation on the entity to obtain a processed entity; the entity query module is used for acquiring a plurality of triples related to the processed entities from a pre-stored traditional Chinese medicine health preserving knowledge graph; the extraction and sorting module is used for screening a plurality of related triples of clinical information which needs to be input again by a user, sorting the triples according to preset priorities, obtaining a triple priority sequence and determining a triple with the highest priority; the conversation generation module is used for generating a question according to the triple with the highest priority and a pre-stored traditional Chinese medicine health preserving knowledge map so as to carry out a next round of questioning on the user; and the result generating module is used for generating a health-preserving suggestion result according to the clinical information input by the user for many times and a pre-stored traditional Chinese medicine health-preserving knowledge map, wherein the entity linking module comprises a context encoder, an input text encoder, a candidate text encoder, an attention mechanism layer and an output result layer.
Action and effects of the invention
According to the intelligent multi-round interaction system for traditional Chinese medicine health preservation based on the knowledge graph, which is disclosed by the invention, because the interaction system has the memory capacity, the content interacted with a user can be actively associated, after the user inputs clinical performance into the system, the system can actively ask the user about the relevant chief complaint performance according to the characteristics of the clinical performance input by the user, the user inputs the relevant feedback content into the system according to the question asked by the system, and the intelligent interaction system adopts the mode, and finally provides the traditional Chinese medicine health preservation suggestion with higher accuracy to the user after multi-round interaction with the user. Moreover, the interactive system of the invention adopts an interactive form of multi-turn conversation, and the accuracy of the end user experience and the accuracy of the Chinese medical health consultation question and answer conclusion are higher than that of the common question and answer system with single turn of interaction. In addition, the invention carries out entity linkage on the entities identified in the preceding and following texts, and proposes an entity linkage model based on memory sequences in combination with the text information of the preceding and following texts to solve entity omission and reference resolution of sentences in multiple rounds of conversations.
Drawings
FIG. 1 is a block diagram of a knowledge graph-based intelligent multi-round interaction system for health preservation in traditional Chinese medicine according to an embodiment of the present invention;
FIG. 2 is a hardware architecture diagram of an intelligent multi-round interaction system for health preservation in traditional Chinese medicine based on knowledge-graph in the embodiment of the present invention;
FIG. 3 is a software architecture diagram of an intelligent multi-round interaction system for health preservation in traditional Chinese medicine based on knowledge-graph in an embodiment of the present invention;
FIG. 4 is a logic diagram of the operation of the entity link module of the intellectual graph-based intelligent multi-round interactive system for health preservation in TCM in the embodiment of the present invention;
FIG. 5 is a flow chart of constructing a health preserving intellectual map of Chinese medicine based on the intellectual map of the intelligent multi-round interactive system for health preserving of Chinese medicine in the embodiment of the present invention;
FIG. 6 is a logic diagram of a session generation module of an intellectual graph-based intelligent multi-round interaction system for health preservation in traditional Chinese medicine according to an embodiment of the present invention;
FIG. 7 is a flowchart illustrating the generation of corpora in the corpus storage module of the intellectual-graph-based intelligent multi-round interactive system for health preservation in TCM, according to an embodiment of the present invention;
fig. 8 is a logic diagram of the operation of the intelligent multi-round interaction system for health preservation in traditional Chinese medicine based on knowledge graph in the embodiment of the invention.
Detailed Description
In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the following embodiments specifically describe the intellectual spectrum-based intelligent multi-round interaction system for the traditional Chinese medicine health preservation in combination with the accompanying drawings.
In the traditional Chinese medicine health-preserving intelligent multi-round interactive system based on the knowledge graph, the system provided by the invention can also have the following characteristics: wherein the context encoder is configured to extract features of entities in two adjacent rounds of interaction.
In the traditional Chinese medicine health-preserving intelligent multi-round interactive system based on the knowledge graph, the system provided by the invention can also have the following characteristics: the context coder, the input text coder and the candidate text coder all comprise a bidirectional long-short term memory neural network.
The intelligent multi-round interaction system for traditional Chinese medicine health preservation based on the knowledge graph provided by the invention can also have the characteristics that: the knowledge map storage module is used for storing the traditional Chinese medicine health preserving knowledge map, and the traditional Chinese medicine health preserving knowledge map comprises traditional Chinese medicine health preserving body knowledge and a traditional Chinese medicine health preserving custom dictionary.
In the traditional Chinese medicine health-preserving intelligent multi-round interactive system based on the knowledge graph, the system provided by the invention can also have the following characteristics: wherein, the source data of the traditional Chinese medicine health preserving knowledge map comprises a traditional Chinese medicine hospital case report table, and the content of the traditional Chinese medicine hospital case report table comprises: gender, age, occupation, ethnicity, marital, heart rate, blood pressure, respiration, time to first onset, course of disease, chief complaints, clinical manifestations, other disease history, brief medical history, exam notes, pathology report notes, tongue veins, western medicine diagnosis, traditional Chinese medicine syndrome differentiation, traditional Chinese medicine diagnosis, prescriptions, treatment regimen, maintenance regimen, diet, exercise, lifestyle habits, massage.
The intelligent multi-round interactive system for traditional Chinese medicine health preservation based on the knowledge graph, provided by the invention, can also have the following characteristics: the construction of the traditional Chinese medicine health-preserving knowledge graph comprises the following steps: step S1, performing entity extraction, relationship extraction and attribute extraction on the contents of the medical case report table of the intermediate hospital to obtain an entity extraction result, a relationship extraction result and an attribute extraction result; and step S2, node creation and relation creation are carried out according to the entity extraction result, the relation extraction result and the attribute extraction result, and a traditional Chinese medicine health preserving knowledge graph is obtained.
In the traditional Chinese medicine health-preserving intelligent multi-round interactive system based on the knowledge graph, the system provided by the invention can also have the following characteristics: and the corpus storage module is used for providing corpus to the session generation module and the result generation module.
In the traditional Chinese medicine health-preserving intelligent multi-round interactive system based on the knowledge graph, the system provided by the invention can also have the following characteristics: the generating process of the linguistic data stored in the linguistic data storage module comprises the following steps: step A1, analyzing entities and entity relations in the traditional Chinese medicine health preserving knowledge graph to obtain an entity relation graph; step A2, converting the entity relationship graph into a directed graph; step A3, taking each entity of the entity relationship graph as a starting point, traversing the directed graph to obtain a path graph; a4, splitting and combining paths in the path graph to obtain sub-paths; and A5, corresponding the entities and the entity relations in the traditional Chinese medicine health-preserving knowledge graph to each sub-path to obtain the corpus.
The intelligent multi-round interaction system for traditional Chinese medicine health preservation based on the knowledge graph provided by the invention can also have the characteristics that: and the service side communication module is used for receiving the clinical information data sent by the user.
In the traditional Chinese medicine health-preserving intelligent multi-round interactive system based on the knowledge graph, the system provided by the invention can also have the following characteristics: the service side communication module comprises an Nginx proxy cluster server, a load balancer and a Web application cluster server.
In this embodiment, an Android mobile phone terminal is taken as an example of a user terminal to explain the intelligent multi-round interaction system for traditional Chinese medicine health preservation based on a knowledge graph, wherein a plurality of user terminals are held by users.
Those skilled in the art can understand that although the Android mobile phone terminal is taken as the user terminal in the embodiment, the user terminal may also be an IOS mobile phone terminal, a PC terminal, or other user terminals.
Fig. 1 is a block diagram of a knowledge graph-based intelligent multi-round interaction system for health preservation in traditional Chinese medicine in an embodiment of the invention.
As shown in fig. 1, the intellectual graph-based intelligent multi-round interaction system 100 for health preservation in chinese medicine may include: the system comprises an entity extraction module 10, an entity linking module 20, an entity query module 30, an extraction sorting module 40, a session generation module 50, a result generation module 60, a knowledge graph storage module 70, a corpus storage module 80 and a service side communication module 90.
The service-side communication module 90 may be configured to accept clinical information data sent by a user.
Fig. 2 is a hardware architecture diagram of the intellectual graph-based intelligent multi-round interaction system for health preservation in traditional Chinese medicine in the embodiment of the present invention. Fig. 3 is a software architecture diagram of the intellectual graph-based intelligent multi-round interactive system for health preservation in traditional Chinese medicine in the embodiment of the present invention.
As shown in fig. 2 and fig. 3, the service-side communication module includes an Nginx proxy cluster server 91, a load balancer 92, and a Web application cluster server 93.
The service side communication module receives a data packet sent by a user through an API (application program interface) of a network layer, a carrier of the user is an Android client, the Android client sends a JSON (Java Server pages) character string to perform data interaction with a system according to the RESTFUL (representational state transfer language) interface requirement, and meanwhile, an interactive UI front-end interface is provided for the user; the Spring Boot is responsible for integrating services such as Nginx and the like into a server; spring MVC is responsible for providing Model, View, Controller interfaces to the server and RESTFUL interface services.
The entity extraction module 10 may be used to extract entities from the clinical information entered by the user in each round of interaction.
The entity linking module 20 may be configured to perform reference resolution and entity disambiguation on the entity to obtain a processed entity.
Fig. 4 is a logic diagram of the operation of the entity link module of the intellectual graph-based intelligent multi-round interactive system for health preservation in traditional Chinese medicine in the embodiment of the present invention.
As shown in FIG. 4, entity linking module 20 includes a postamble encoder, an input text encoder, a candidate text encoder, an attention mechanism layer, and an output results layer. Wherein, the context coder is responsible for extracting the features of the previous and the following text. The context coder, the input text coder and the candidate text coder all adopt bidirectional LSTM to input text codingCode device as an example, input text { w1，w2，...，wn}(wnI.e., the nth word in w), i.e., at time f, the positive LSTM yields
Inverse LSTM derivation
Output w at time ffCan be formed by splicing
And
thus obtaining the compound. The formula is as follows:
and finally, transmitting the candidate vector c, the context vector s and the input text vector w which are calculated by using the attention mechanism into a replay layer to obtain an entity link result. The calculation formula is as follows:
s＝Attention(c，w)
c＝Attention(s，w)
the embodiment extracts valuable features in multiple rounds of interactive sessions by using the attention mechanism layer, and extracts content features weakened or disappeared in the bidirectional LSTM due to forward transmission over time, so as to filter out invalid features, and finally obtain the valid entity link.
The traditional Chinese medicine health-preserving intelligent multi-round interactive system based on the knowledge graph fuses related information together by using modes of knowledge graph node entity discovery, linkage and the like, so that multi-round interaction is realized. In the conversation process of the health maintenance consultation of the traditional Chinese medicine, the consultation of the user is relevant and continuous, and in order to ensure the accuracy and richness of the reply content of the system, the reply needs to be carried out by combining interactive historical information and a knowledge map related to the interaction.
The modeling of the traditional Chinese medicine health-preserving intelligent multi-round interactive system based on the knowledge graph is completed by adopting a point-to-point generation model, the knowledge graph, the interactive history and the current input of a user are modeled, and then decoding reply is carried out according to modeling information. The interaction history, the knowledge graph and the generated reply are not in a one-to-one mapping relation, so that the correct selection of the knowledge graph and the interaction history and the correct selection of the modeling mode can have a great influence on the effect of generating the reply by the interactive system.
The entity query module 30 may be configured to obtain a plurality of triples related to the processed entity from a pre-stored traditional Chinese medicine health preserving knowledge graph.
The knowledge graph storage module 70 may be configured to store a traditional Chinese medicine health preserving knowledge graph, where the traditional Chinese medicine health preserving knowledge graph includes a traditional Chinese medicine health preserving body knowledge and a traditional Chinese medicine health preserving custom dictionary.
The source data of the traditional Chinese medicine health-preserving knowledge map comprises a traditional Chinese medicine hospital case report table, and the content of the traditional Chinese medicine hospital case report table comprises: gender, age, occupation, ethnicity, marital, heart rate, blood pressure, respiration, time to first onset, course of disease, chief complaints, clinical manifestations, other disease history, brief medical history, exam notes, pathology report notes, tongue veins, western medicine diagnosis, traditional Chinese medicine syndrome differentiation, traditional Chinese medicine diagnosis, prescriptions, treatment regimen, maintenance regimen, diet, exercise, lifestyle habits, massage.
Fig. 5 is a flow chart for constructing a traditional Chinese medicine health preserving knowledge graph of the traditional Chinese medicine health preserving intelligent multi-round interactive system based on the knowledge graph in the embodiment of the invention.
As shown in fig. 5, the construction of the health preserving knowledge graph of traditional Chinese medicine comprises the following steps:
step S1, performing entity extraction, relationship extraction and attribute extraction on the contents of the medical case report table of the hospital to obtain an entity extraction result, a relationship extraction result and an attribute extraction result;
and step S2, performing node creation and relationship creation according to the entity extraction result, the relationship extraction result and the attribute extraction result, and obtaining the traditional Chinese medicine health preserving knowledge graph.
And the extraction and sorting module 40 is configured to screen a plurality of relevant triples of clinical information that needs to be input again by the user, sort the plurality of triples according to a preset priority, obtain a triple priority sequence, and determine a highest priority triple.
In the process of intelligent multi-round interaction, the situations of reference resolution and entity omission occur in the conversation, and the situations cause great troubles to entity linkage, so that the invention aims at entity linkage of the entities identified in the preceding and following texts and proposes to solve the problem of entity omission and reference resolution of statements in the multi-round conversation based on the entity linkage module 20 by combining the text information in the preceding and following texts.
First, the triples related to the entities identified in the context are spliced into candidate texts, then a full connection layer is used for splicing the candidate vectors c output by the candidate text encoder, the context vectors s output by the context encoder and the text vectors w output by the input text encoder, and then linear change is carried out, and finally, the output result replay is calculated through the softmax function, wherein the formula is as follows:
reply＝softmax(F·[c；s；w]
f in the formula is the full link layer parameter.
The session generation module 50 may be configured to generate a question according to the triple with the highest priority and a pre-stored traditional Chinese medicine health preserving knowledge graph, so as to perform a next round of questioning on the user.
Fig. 6 is a logic diagram of the operation of the session generation module of the intellectual-graph-based intelligent multi-round interactive system for health preservation in traditional Chinese medicine in the embodiment of the present invention.
As shown in fig. 6, the present embodiment proposes that the text information of a sentence is modeled using one encoder, and the text information before and after the sentence is composed is modeled using another encoder, which generates a model based on the memory ability of the entity linking module 20. The encoder used by the session generation module 50 based on the entity link module 20 is composed of a unidirectional GRU and a bidirectional GRU, the encoder used by the text information before and after the sentence composition is composed of a unidirectional GRU, and the encoder used by the text information of the sentence is a bidirectional GRU.
One preceding or following text is given by { s1,s2,...,snThe components of the components are replaced by the components,snis the nth sentence in the preceding and following text, sn＝{wn1，w21，...，wnn}，wnnIs snThe Encoder used by Word Encoder is a bidirectional GRU that encodes each sentence { s } in the context1,s2,...,snEncoding into sentence vector
The Encoder used by the Context Encoder is a one-way GRU that vectors sentences
Coding into a context vector t, and calculating as follows:
t＝sn＝GRU(sn，sn-1)
the result generating module 60 may be configured to generate a health promotion suggestion result according to the clinical information input by the user for multiple times and a pre-stored traditional Chinese medicine health promotion knowledge graph.
The corpus storage module 80 may be used to provide corpus to the session generation module and the result generation module.
Fig. 7 is a flowchart illustrating the generation of corpora in the corpus storage module of the intellectual-graph-based intelligent multi-round interactive system for health preservation in traditional Chinese medicine according to the embodiment of the present invention.
The corpus generation process stored in the corpus storage module 80 as shown in fig. 7 includes the following steps:
step A1, analyzing entities and entity relations in the traditional Chinese medicine health preserving knowledge graph to obtain an entity relation graph;
step A2, converting the entity relationship graph into a directed graph;
step A3, taking each entity of the entity relationship graph as a starting point, traversing the directed graph to obtain a path graph;
a4, splitting and combining paths in the path graph to obtain sub-paths;
and A5, corresponding the entities and the entity relations in the traditional Chinese medicine health-preserving knowledge graph to each sub-path to obtain the corpus.
Fig. 8 is a logic diagram of the operation of the intelligent multi-round interaction system for health preservation in traditional Chinese medicine based on knowledge graph in the embodiment of the invention.
As shown in fig. 8, after conducting named entity recognition on the current conversation sentence and the historical conversation sentence of the traditional Chinese medicine health problem consulted by the user, the traditional Chinese medicine health preserving intelligent interaction system based on the knowledge graph obtains an entity list related to the user problem, uses a memory sequence-based entity link model to conduct reference resolution and entity disambiguation, extracts triple information related to the entity from the traditional Chinese medicine health preserving knowledge graph, transmits the triple information to the conversation system, and uses a memory sequence modeling-based generation model to generate a reply transmitted to the user.
Effects and effects of the embodiments
According to the traditional Chinese medicine health-preserving intelligent multi-round interaction system based on the knowledge graph, the interaction system has the memory capacity, and the content interacted with the user can be actively associated, so that the essential requirement is met. After a user inputs clinical manifestations into the system, the system actively asks the user for related chief complaint manifestations according to the characteristics of the clinical manifestations input by the user, the user inputs related feedback contents into the system according to questions asked by the system, and the intelligent interactive system uses the mode, and finally gives traditional Chinese medicine health-preserving suggestions with higher accuracy to the user after multi-round interaction with the user. Moreover, the interactive system of the embodiment adopts an interactive form of multi-turn conversation, so that the conclusion accuracy of the end user experience and the Chinese medical health consultation question and answer is higher than that of a common question and answer system with single-turn interaction. In addition, the embodiment carries out entity linking on the entities identified in the preceding and following texts, and proposes an entity linking model based on memory sequences in combination with the text information of the preceding and following texts to solve entity omission and reference resolution of sentences in multiple sessions.
The above embodiments are preferred examples of the present invention, and are not intended to limit the scope of the present invention.
Claims (10)
1. A traditional Chinese medicine health preserving intelligent multi-round interactive system based on knowledge graph is characterized by comprising:
the entity extraction module is used for extracting entities in the clinical information input by the user in each round of interaction;
the entity linking module is used for carrying out reference resolution and entity disambiguation on the entity to obtain the processed entity;
the entity query module is used for acquiring a plurality of triples related to the processed entities from a pre-stored traditional Chinese medicine health preserving knowledge graph;
the extraction and sorting module is used for screening a plurality of relevant triples of the clinical information which needs to be input again by the user, sorting the triples according to preset priorities, obtaining a triple priority sequence and determining a triple with the highest priority;
the session generation module is used for generating a question according to the highest priority triple and the pre-stored traditional Chinese medicine health preserving knowledge graph so as to carry out a next round of questioning on the user; and
a result generating module for generating health promotion suggestion results according to the clinical information input by the user for multiple times and the pre-stored traditional Chinese medicine health promotion knowledge graph,
the entity linking module comprises a context encoder, an input text encoder, a candidate text encoder, an attention mechanism layer and an output result layer.
2. The intellectual graph based intelligent multi-round interaction system for health preservation of traditional Chinese medicine according to claim 1, which is characterized in that:
wherein the preceding and following coders are configured to extract features of the entities in two adjacent rounds of interaction.
3. The intellectual graph based intelligent multi-round interaction system for health preservation of traditional Chinese medicine according to claim 1, which is characterized in that:
wherein the preceding and following coders, the input text coder and the candidate text coder each comprise a bidirectional long-short term memory neural network.
4. The intellectual graph based intelligent multi-round interaction system for health preservation of traditional Chinese medicine according to claim 1, further comprising:
a knowledge-map storage module for storing the knowledge-map,
wherein the knowledge map storage module is used for storing the traditional Chinese medicine health preserving knowledge map,
the traditional Chinese medicine health preserving knowledge map comprises traditional Chinese medicine health preserving body knowledge and a traditional Chinese medicine health preserving user-defined dictionary.
5. The intellectual graph based intelligent multi-round interaction system for health preservation of traditional Chinese medicine according to claim 1, which is characterized in that:
wherein the source data of the traditional Chinese medicine health-preserving knowledge graph comprises a traditional Chinese medicine hospital case report form,
the content of the traditional Chinese medical hospital case report table comprises the following components: gender, age, occupation, ethnicity, marital, heart rate, blood pressure, respiration, time to first onset, medical history, chief complaints, clinical manifestations, other disease history, brief medical history, examination sheet, pathology report sheet, tongue pulse, western medicine diagnosis, traditional Chinese medicine syndrome differentiation, traditional Chinese medicine diagnosis, prescription, treatment, maintenance, diet, exercise, lifestyle habits, massage and massage.
6. The intellectual graph based intelligent multi-round interaction system for health preservation of traditional Chinese medicine according to claim 5, wherein:
the construction of the traditional Chinese medicine health-preserving knowledge graph comprises the following steps:
step S1, performing entity extraction, relationship extraction and attribute extraction on the contents of the traditional Chinese medical hospital case report table to obtain an entity extraction result, a relationship extraction result and an attribute extraction result;
and step S2, performing node creation and relationship creation according to the entity extraction result, the relationship extraction result and the attribute extraction result, and obtaining the traditional Chinese medicine health preserving knowledge graph.
7. The intellectual graph based intelligent multi-round interaction system for health preservation of traditional Chinese medicine according to claim 1, further comprising:
a corpus storage module for storing the corpus of the user,
the corpus storage module is used for providing corpus to the session generation module and the result generation module.
8. The intellectual graph based intelligent multi-round interaction system for health preservation of traditional Chinese medicine according to claim 7, which is characterized in that:
the generating process of the corpus stored in the corpus storage module comprises the following steps:
step A1, analyzing the entity and the entity relationship in the traditional Chinese medicine health preserving knowledge graph to obtain an entity relationship graph;
step A2, converting the entity relationship graph into a directed graph;
step A3, taking each entity of the entity relationship graph as a starting point, traversing the directed graph to obtain a path graph;
step A4, splitting and combining the paths in the path graph to obtain sub-paths;
and A5, corresponding the entities and the entity relations in the traditional Chinese medicine health-preserving knowledge graph to the sub-paths to obtain the corpus.
9. The intellectual graph based intelligent multi-round interaction system for health preservation of traditional Chinese medicine according to claim 1, further comprising:
a service-side communication module for communicating with a service-side communication module,
the service side communication module is used for receiving the clinical information data sent by the user.
10. The intellectual graph based intelligent multi-round interaction system for health preservation of traditional Chinese medicine according to claim 1, which is characterized in that:
the service side communication module comprises an Nginx proxy cluster server, a load balancer and a Web application cluster server.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111369070.0A CN114090789B (en) | 2021-11-12 | 2021-11-12 | Knowledge-graph-based intelligent multi-round interaction system for traditional Chinese medicine health maintenance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111369070.0A CN114090789B (en) | 2021-11-12 | 2021-11-12 | Knowledge-graph-based intelligent multi-round interaction system for traditional Chinese medicine health maintenance |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114090789A true CN114090789A (en) | 2022-02-25 |
| CN114090789B CN114090789B (en) | 2023-04-21 |
Family
ID=80301686
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111369070.0A Active CN114090789B (en) | 2021-11-12 | 2021-11-12 | Knowledge-graph-based intelligent multi-round interaction system for traditional Chinese medicine health maintenance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114090789B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116504414A (en) * | 2023-05-19 | 2023-07-28 | 北京百度网讯科技有限公司 | Medical data query method, device, electronic device and storage medium |
| CN117407509A (en) * | 2023-10-31 | 2024-01-16 | 短歌有限公司 | Human-computer interaction method and system based on opinion map |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090144609A1 (en) * | 2007-10-17 | 2009-06-04 | Jisheng Liang | NLP-based entity recognition and disambiguation |
| CN110377715A (en) * | 2019-07-23 | 2019-10-25 | 天津汇智星源信息技术有限公司 | Reasoning type accurate intelligent answering method based on legal knowledge map |
| CN113254659A (en) * | 2021-02-04 | 2021-08-13 | 天津德尔塔科技有限公司 | File studying and judging method and system based on knowledge graph technology |
| CN113590783A (en) * | 2021-07-28 | 2021-11-02 | 复旦大学 | Traditional Chinese medicine health-preserving intelligent question-answering system based on NLP natural language processing |
-
2021
- 2021-11-12 CN CN202111369070.0A patent/CN114090789B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090144609A1 (en) * | 2007-10-17 | 2009-06-04 | Jisheng Liang | NLP-based entity recognition and disambiguation |
| CN110377715A (en) * | 2019-07-23 | 2019-10-25 | 天津汇智星源信息技术有限公司 | Reasoning type accurate intelligent answering method based on legal knowledge map |
| CN113254659A (en) * | 2021-02-04 | 2021-08-13 | 天津德尔塔科技有限公司 | File studying and judging method and system based on knowledge graph technology |
| CN113590783A (en) * | 2021-07-28 | 2021-11-02 | 复旦大学 | Traditional Chinese medicine health-preserving intelligent question-answering system based on NLP natural language processing |
Non-Patent Citations (1)
| Title |
|---|
| 朱宗奎: "基于神经网络的中文知识图谱问答研究" * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116504414A (en) * | 2023-05-19 | 2023-07-28 | 北京百度网讯科技有限公司 | Medical data query method, device, electronic device and storage medium |
| CN116504414B (en) * | 2023-05-19 | 2024-03-22 | 北京百度网讯科技有限公司 | Medical data query method, device, electronic equipment and storage medium |
| CN117407509A (en) * | 2023-10-31 | 2024-01-16 | 短歌有限公司 | Human-computer interaction method and system based on opinion map |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114090789B (en) | 2023-04-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112271001B (en) | Medical consultation dialogue system and method applying heterogeneous graph neural network | |
| CN115309877B (en) | Dialogue generation method, dialogue model training method and device | |
| CN112084789B (en) | Text processing method, device, equipment and storage medium | |
| CN111666477B (en) | A data processing method, device, intelligent device and medium | |
| CN110993093B (en) | Method and device for pre-diagnosis in ophthalmology based on deep learning | |
| CN110476169A (en) | Provide emotional care during the session | |
| CN112100406B (en) | Data processing method, device, equipment and medium | |
| CN114330372B (en) | Model training method, related device and equipment | |
| US20250272556A1 (en) | Data processing method and apparatus, and electronic device | |
| CN110929532B (en) | Data processing method, device, equipment and storage medium | |
| WO2023029501A1 (en) | Smart interrogation method and apparatus, electronic device, and storage medium | |
| CN116894498B (en) | Training methods, policy recognition methods, devices and equipment for network models | |
| CN113761156A (en) | Data processing method, device, medium and electronic device for human-computer interaction dialogue | |
| CN112417117B (en) | Method, device and equipment for generating conversation message | |
| CN118350616A (en) | Business process construction method, device, electronic device and storage medium | |
| Inupakutika et al. | Integration of NLP and speech-to-text applications with chatbots | |
| KR102722913B1 (en) | System and method for generating medical text based on large language model | |
| CN114090789B (en) | Knowledge-graph-based intelligent multi-round interaction system for traditional Chinese medicine health maintenance | |
| CN118692450A (en) | Question and answer processing method, device, equipment, storage medium and program product | |
| CN116956934A (en) | Task processing method, device, equipment and storage medium | |
| CN119106739A (en) | Alignment model training method, information processing method and device | |
| CN113571184A (en) | A dialogue interaction design method and system for mental health assessment | |
| CN120072356A (en) | Automatic question answering method and medicine recommendation method | |
| CN109002498B (en) | Man-machine conversation method, device, equipment and storage medium | |
| CN116911260A (en) | Text evaluation method, device, equipment and medium based on artificial intelligence |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |