An intelligent question answering system for multi-source hospital database
Through an intelligent question-and-answer system for multi-source hospital databases, the keyword index set and pointing vector set are built, which solves the problem of low processing efficiency of multi-source data and improves the accuracy and user experience of answers.
Patent Information
- Application Number
- CN202410775695.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-17
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-06-17
AI Technical Summary
The existing multi-source data intelligent question and answer system has too much data processing volume and low efficiency when processing multi-source data, which has affected the user experience.
Design an intelligent question-and-answer system for multi-source hospital databases, obtain data from multiple data sources through the data source fusion module, generate target data sets and build keyword index sets, fuzzy word index sets and pointing vector sets, and the data set matching module is based on users The problem determines fuzzy keywords and intermediate answers, and the intelligent question and answer module generates user answers, reducing data processing volume and improving retrieval efficiency.
By building keyword index sets and pointing vector sets that are easy to retrieve, data processing efficiency and answer accuracy are improved, and user experience is improved.
Smart Images

Figure CN118643136B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent question answering technology, and in particular to an intelligent question answering system oriented to a multi-source hospital database. Background Art
[0002] With the development of artificial intelligence technology, the technology of human-intelligent system interaction is constantly improving, and related intelligent applications are developing rapidly. Intelligent question-answering system is a major research hotspot in the field of artificial intelligence. Its application in the medical field is receiving more and more attention, playing a positive role in saving human resources, improving the efficiency of information processing, and reducing website operating costs. Intelligent question-answering system is an important direction in the field of natural language processing, which aims to allow users to ask questions directly in natural language and get answers. Due to the complexity of natural language, it is very difficult for computers to correctly understand human language, especially in some special fields with a large number of professional terms. The application effect of intelligent question-answering system is far from meeting the needs of users.
[0003] For intelligent question-answering systems, matching questions with corresponding answers is an essential step. Text information matching in intelligent question-answering systems refers to entering pre-defined questions and corresponding answers into a database, then searching the database for questions based on the questions asked by the user, and finally extracting the answers corresponding to the questions with the highest similarity and feeding them back to the user. However, the traditional method is usually to perform fuzzy queries in the knowledge database, and select the query results that match the most keywords in the user's question description as the target. This matching method can easily lead to the user's question not being correctly understood and thus returning an incorrect answer. Existing knowledge databases rely on a single data source and expert knowledge, and do not make full use of multi-source big data resources, which can easily lead to the user's question being small in scope or not being correctly understood, thus returning an incorrect answer.
[0004] Chinese patent publication number CN112749265A discloses an intelligent question-answering system based on multiple information sources, which specifically includes a question-answering module KBQA based on a knowledge base, a question-answering module DBQA based on a document set, a question-answering module QuesSimQA based on a frequently asked question set, a third-party API module and a multiple answer verification module; for the user's questions, the KBQA module, the DBQA module, the QuesSimQA module and the third-party API module respectively give an answer, and then the answers provided by each module are verified, scored and ranked through the multi-source answer verification module, and the answer with the highest score is submitted to the user.
[0005] It can be seen that the intelligent question and answer system based on multiple information sources provided by the invention has the following problems: in the face of multi-source data, by establishing different question and answer modules, scoring the answers output by each answer module to determine the answers to user questions, it will increase the data processing volume of the intelligent question and answer system, increase the computing cost, and reduce the response efficiency, affecting the user experience. Summary of the invention
[0006] To this end, the present invention provides an intelligent question-answering system for multi-source hospital databases, so as to overcome the problems of excessive data processing volume and low efficiency faced by the multi-source data intelligent question-answering system in the prior art.
[0007] To achieve the above object, the present invention provides an intelligent question-answering system for a multi-source hospital database, comprising:
[0008] A data source fusion module is used to obtain data from several data sources, classify the data types of the data in each data source to generate several target data sets, and determine the keyword index set, fuzzy word index set and fuzzy word and keyword pointing vector set corresponding to each target data set; each target data set has a corresponding answer corpus;
[0009] A data set matching module, which is connected to the data source fusion module, is used to determine the question keywords according to the user question, determine the fuzzy keywords according to the relationship between the question keywords and each of the keyword index sets, and determine the retrieval method of the intermediate answer according to the attributes of the fuzzy keywords, including determining the intermediate answer based on the fuzzy word index set and determining the intermediate answer based on the pointing vector set;
[0010] And, determining a target data set corresponding to the user's answer based on the obtained intermediate answer and the question keyword; wherein the obtained intermediate answer is used as a keyword to replace the fuzzy keyword to form a new question keyword;
[0011] An intelligent question-answering module, which is connected to the data source fusion module and the data set matching module respectively, and is used to determine the user answer keywords according to the user question, determine the answer corpus based on the target data set corresponding to the user answer, and output the corresponding answer form according to the answer corpus and a preset output model;
[0012] The answer generation module is connected to the intelligent question and answer module and is used to output the user answer corresponding to the user question according to the user answer keywords and the answer form.
[0013] Furthermore, the data source fusion module includes:
[0014] A data classification submodule, which is used to classify data in a plurality of data sources according to their data types to obtain a plurality of initial data sets; the data types of the data in each of the initial data sets are the same;
[0015] A data preprocessing submodule, which is connected to the data classification submodule and is used to perform data preprocessing on the data in each initial data set to generate a plurality of target data sets;
[0016] A keyword extraction submodule, which is connected to the data preprocessing submodule, is used to extract keywords of each data in each target data set, and to form fuzzy words based on the analysis of each keyword, so as to obtain a keyword index set and a fuzzy word index set corresponding to each target data set, and to determine a pointing vector set of fuzzy words and keywords according to the keywords and fuzzy words of each data in a number of target data sets; the fuzzy words and keywords have a superior-fewer relationship, and each fuzzy word has at least one corresponding keyword;
[0017] Among them, if the keywords corresponding to a single fuzzy word are all located in the same target data set, the fuzzy word is collected into the fuzzy word index set corresponding to the target data set; if the keywords corresponding to a single fuzzy word are located in at least two target data sets, the fuzzy word is collected into the pointing vector set.
[0018] Furthermore, the data set matching module includes:
[0019] A question keyword determination submodule, which is connected to the data source fusion module and is used to determine question keywords according to the user's question; there is at least one question keyword;
[0020] A fuzzy keyword determination submodule, which is connected to the question keyword determination submodule and is used to determine the fuzzy keyword according to the relationship between the question keyword and each of the keyword index sets;
[0021] in,
[0022] Under the first relationship condition, the corresponding question keyword is processed according to a preset keyword fuzzy method to obtain a fuzzy keyword corresponding to the question keyword;
[0023] Under the second relationship condition, determining fuzzy keywords according to the user question based on a preset fuzzy word comparison table;
[0024] An intermediate answer determination submodule, which is connected to the fuzzy keyword determination submodule and is used to determine a retrieval method of an intermediate answer according to the attributes of the fuzzy keyword, including determining an intermediate answer based on the fuzzy word index set and determining an intermediate answer based on the pointing vector set;
[0025] The data set matching submodule is connected to the intermediate answer determination submodule, and determines the target data set corresponding to the user's answer based on the obtained intermediate answer and question keywords.
[0026] Furthermore, the first relationship condition is that there is at least one of the problem keywords and the number of occurrences in each of the keyword index sets is 0.
[0027] Furthermore, the second relationship condition is that the number of occurrences of each of the question keywords in each of the keyword index sets is not 0, and each of the question keywords does not correspond to the same target data set.
[0028] Furthermore, the method of determining the retrieval method of the intermediate answer according to the attribute of the fuzzy keyword includes:
[0029] If the fuzzy keyword is a preset first attribute, determining an intermediate answer based on the fuzzy word index set;
[0030] If the fuzzy keyword is a preset second attribute, an intermediate answer is determined based on the pointing vector set.
[0031] Further, the determining of the intermediate answer based on the fuzzy word index set includes: determining a corresponding target data set according to the fuzzy keywords based on the fuzzy word index set, and generating an intermediate answer according to corresponding data in the target data set.
[0032] Furthermore, the determining of the intermediate answer based on the pointing vector set includes: determining subordinate keywords of the fuzzy keywords based on the pointing vector set of the fuzzy words and keywords, and determining the target data set where the intermediate answer is located based on the subordinate keywords, and generating the intermediate answer based on the corresponding data in the target data set.
[0033] Furthermore, the intelligent question-answering module includes:
[0034] A user answer keyword determination submodule, which is used to determine the user answer keywords according to the user question based on a preset answer keyword comparison table;
[0035] An answer corpus determination submodule, which is connected to the data source fusion module and the data set matching module respectively, and is used to determine the answer corpus based on the target data set corresponding to the user answer;
[0036] The answer form output submodule is connected to the answer corpus determination submodule and is used to output the corresponding answer form according to the answer corpus, the user question and the preset output model.
[0037] Furthermore, the answer generation module includes:
[0038] An answer generation submodule, which is connected to the intelligent question and answer module and is used to match the user answer keywords and the answer form to generate a user answer;
[0039] The answer output submodule is connected to the answer generation submodule and is used to output the user answer corresponding to the user question.
[0040] Compared with the prior art, the beneficial effect of the present invention is that the intelligent question-answering system for multi-source hospital database provided by the present invention processes multi-source data to obtain several target data sets by setting a data source fusion module to process multi-source data, constructs a keyword index set, a fuzzy word index set, and a pointing vector set of fuzzy words and keywords that are easy to search, which can reduce the subsequent data processing volume and improve the search efficiency. By setting a data set matching module, an intelligent question-answering module, and an answer generation module, the target data set corresponding to the user answer is determined based on the question keywords and fuzzy keywords, thereby determining the answer corpus to generate the answer form, and outputting the user answer corresponding to the user question based on the user answer keywords and the answer form, the accuracy of retrieving the user answer can be improved, thereby improving the accuracy of the intelligent question-answering, so as to improve the user experience.
[0041] Furthermore, the data source fusion module of the present invention can classify the data in several data sources according to the data type by setting the data classification submodule, so as to form the corresponding answer corpus according to the data of the same data type, thereby improving the efficiency of subsequent data processing. Since the data standards and formats in each data source are not uniform, and there may be duplicate and noisy data, by setting the data preprocessing submodule to preprocess the data in each initial data set, the amount of subsequent data processing can be reduced and the accuracy of question and answer can be improved. By setting the keyword extraction submodule to construct the keyword index set, the fuzzy word index set, and the pointing vector set of the fuzzy word and the keyword, the efficiency of subsequent retrieval can be improved.
[0042] Furthermore, the data set matching module of the present invention, by setting a question keyword determination submodule, converts the retrieval of answers based on user questions into retrieval based on each question keyword corresponding to the user question, which can improve the answer retrieval efficiency. By setting a fuzzy keyword determination submodule, fuzzy keywords are determined according to the relationship between the question keywords and each keyword index set, which can ensure that the user answer can be retrieved even when the user question semantics are relatively vague and not precise enough, so as to improve the user experience. The intermediate answer determination submodule determines the retrieval method of the intermediate answer according to the attributes of the fuzzy keyword, which can avoid the problem of excessive data processing and inaccurate intermediate answers retrieved.
[0043] Furthermore, if any question keyword determined based on the user's question is not retrieved in each keyword index set, it will affect the accuracy of the output answer to the question. Therefore, when there is at least one question keyword that appears 0 times in each keyword index set, the non-appearing question keyword is processed to determine the corresponding fuzzy keyword, which can improve the accuracy of question and answer.
[0044] Furthermore, each question keyword determined according to the user's question can be retrieved in each keyword index set, but each question keyword does not correspond to the same target data set, indicating that the retrieved target data set is not accurate enough and may miss relevant information, resulting in inaccurate answers to the questions, thereby affecting the user experience. Therefore, determining fuzzy keywords according to user questions can improve the accuracy of question and answer.
[0045] Furthermore, the different attributes of fuzzy keywords will affect the accuracy of the intermediate answers and the amount of data processing. By setting different retrieval methods, the amount of data processing can be reduced and the accuracy of questions and answers can be improved.
[0046] Furthermore, based on the fuzzy word index set, the target data set corresponding to the fuzzy keyword can be directly determined, thereby determining the intermediate answer, which can improve the retrieval efficiency.
[0047] Furthermore, the subordinate keywords of the fuzzy keywords are determined based on the directional vector set of the fuzzy words and keywords, and the target data set where the intermediate answers are located is determined according to the subordinate keywords, which can reduce the amount of data processing.
[0048] Furthermore, the intelligent question and answer module of the present invention determines the user answer keywords by setting a user answer keyword determination submodule, determines the answer corpus according to the target data set corresponding to the user answer by setting an answer corpus determination submodule, and outputs the corresponding answer form according to the answer form output submodule, thereby ensuring the accuracy of the user's answer.
[0049] Furthermore, the answer generation module of the present invention outputs the user answer corresponding to the user question by setting an answer generation submodule and an answer output submodule, thereby ensuring that the form of the user answer is relatively standard and improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] Figure 1 It is a structural block diagram of the intelligent question-answering system for multi-source hospital databases of the present invention;
[0051] Figure 2 A structural block diagram of the data source fusion module provided by the present invention;
[0052] Figure 3 A structural block diagram of a data set matching module provided by the present invention;
[0053] Figure 4 A structural block diagram of the intelligent question-answering module provided by the present invention;
[0054] Figure 5 This is a structural block diagram of the answer generation module provided by the present invention. DETAILED DESCRIPTION
[0055] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0056] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.
[0057] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.
[0058] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0059] See also Figure 1 As shown, it is a structural block diagram of the intelligent question-answering system for multi-source hospital databases of the present invention. The present invention provides an intelligent question-answering system for multi-source hospital databases, including:
[0060] A data source fusion module is used to obtain data from several data sources, classify the data types of the data in each data source to generate several target data sets, and determine the keyword index set, fuzzy word index set and fuzzy word and keyword pointing vector set corresponding to each target data set; each target data set has a corresponding answer corpus;
[0061] In implementation, the data sources include patients' diagnostic information, electronic medical records, examination reports, treatment information, medication information, etc., the hospital's drug procurement information, usage, management information, etc., staff's job information, personnel files, diagnosis and treatment information, etc., medical literature, medical dictionary resources, disease classification manuals, etc.
[0062] Specifically, data types can be classified into structured data, semi-structured data, and unstructured data.
[0063] See also Figure 2As shown, it is a structural block diagram of the data source fusion module of the present invention.
[0064] Specifically, the data source fusion module includes:
[0065] A data classification submodule, which is used to classify data in a plurality of data sources according to their data types to obtain a plurality of initial data sets; the data types of the data in each of the initial data sets are the same;
[0066] A data preprocessing submodule, which is connected to the data classification submodule and is used to perform data preprocessing on the data in each initial data set to generate a plurality of target data sets;
[0067] In implementation, data preprocessing includes: format unification, noise data removal, duplicate data removal, data segmentation and part-of-speech tagging, etc.
[0068] It is worth noting that those skilled in the art know that any method capable of performing data preprocessing in the prior art falls within the protection scope of the present invention and will not be described in detail herein.
[0069] A keyword extraction submodule, which is connected to the data preprocessing submodule, is used to extract keywords of each data in each target data set, and to form fuzzy words based on the analysis of each keyword, so as to obtain a keyword index set and a fuzzy word index set corresponding to each target data set, and to determine a pointing vector set of fuzzy words and keywords according to the keywords and fuzzy words of each data in a number of target data sets; the fuzzy words and keywords have a superior-fewer relationship, and each fuzzy word has at least one corresponding keyword;
[0070] Among them, if the keywords corresponding to a single fuzzy word are all located in the same target data set, the fuzzy word is collected into the fuzzy word index set corresponding to the target data set; if the keywords corresponding to a single fuzzy word are located in at least two target data sets, the fuzzy word is collected into the pointing vector set.
[0071] In implementation, each data has at least one keyword, each target data set has a corresponding keyword index set, and each keyword can be set with a corresponding fuzzy word. For example, if the keyword is amoxicillin capsule, the fuzzy word can be antibiotics, medicines, or penicillin antibiotics; and if the keyword is cefazolin, the fuzzy word can also be antibiotics. The actual implementation personnel can set it according to the actual situation. Generally speaking, a fuzzy word is a broad definition (hypernym) covering a specific keyword.
[0072] The data source fusion module of the present invention classifies the data in several data sources according to the data type by setting the data classification submodule, which can facilitate the formation of corresponding answer corpus according to the data of the same data type, thereby improving the efficiency of subsequent data processing. Since the data standards and formats in each data source are not uniform, and there may be repeated and noisy data, by setting the data preprocessing submodule to preprocess the data in each initial data set, the subsequent data processing amount can be reduced and the accuracy of question and answer can be improved. By setting the keyword extraction submodule to construct the keyword index set, the fuzzy word index set, and the pointing vector set of the fuzzy word and the keyword, the subsequent retrieval efficiency can be improved.
[0073] A data set matching module, which is connected to the data source fusion module, is used to determine the question keywords according to the user question, determine the fuzzy keywords according to the relationship between the question keywords and each of the keyword index sets, and determine the retrieval method of the intermediate answer according to the attributes of the fuzzy keywords, including determining the intermediate answer based on the fuzzy word index set and determining the intermediate answer based on the pointing vector set;
[0074] And, determining a target data set corresponding to the user's answer based on the obtained intermediate answer and the question keyword; wherein the obtained intermediate answer is used as a keyword to replace the fuzzy keyword to form a new question keyword;
[0075] Specifically, the data set matching module includes:
[0076] A question keyword determination submodule, which is connected to the data source fusion module and is used to determine question keywords according to the user's question; there is at least one question keyword;
[0077] A fuzzy keyword determination submodule, which is connected to the question keyword determination submodule and is used to determine the fuzzy keyword according to the relationship between the question keyword and each of the keyword index sets;
[0078] in,
[0079] Under the first relationship condition, the corresponding question keyword is processed according to a preset keyword fuzzy method to obtain a fuzzy keyword corresponding to the question keyword;
[0080] Under the second relationship condition, determining fuzzy keywords according to the user question based on a preset fuzzy word comparison table;
[0081] In implementation, the preset keyword fuzzy method can be to determine the corresponding text library according to the part of speech of the question keyword, and define the fuzzy matching algorithm according to the text similarity; or determine the corresponding fuzzy word according to the search engine library in the prior art; or determine the corresponding fuzzy word based on the knowledge base or historical retrieval records.
[0082] In actual application scenarios, actual implementers can set a preset fuzzy word comparison table according to actual conditions, and the fuzzy word comparison table can be updated (including expansion, deletion and replacement) according to the update of the data source.
[0083] An intermediate answer determination submodule, which is connected to the fuzzy keyword determination submodule and is used to determine a retrieval method of an intermediate answer according to the attributes of the fuzzy keyword, including determining an intermediate answer based on the fuzzy word index set and determining an intermediate answer based on the pointing vector set;
[0084] The data set matching submodule is connected to the intermediate answer determination submodule, and determines the target data set corresponding to the user's answer based on the obtained intermediate answer and question keywords.
[0085] The data set matching module of the present invention converts the retrieval of answers according to user questions into retrieval according to each question keyword corresponding to the user question by setting a question keyword determination submodule, which can improve the answer retrieval efficiency. By setting a fuzzy keyword determination submodule, fuzzy keywords are determined according to the relationship between the question keywords and each keyword index set, which can ensure that the user answer can be retrieved even when the user question semantics are relatively vague and not precise enough, so as to improve the user experience. The intermediate answer determination submodule determines the retrieval method of the intermediate answer according to the attributes of the fuzzy keyword, which can avoid the problem of excessive data processing and inaccurate intermediate answers retrieved.
[0086] Specifically, the first relationship condition is that there is at least one of the problem keywords and the number of occurrences in each of the keyword index sets is 0.
[0087] If any question keyword determined by the present invention based on the user's question is not retrieved in each keyword index set, it will affect the accuracy of the output answer to the question. Therefore, when there is at least one question keyword with a number of appearances of 0 in each keyword index set, the non-appearing question keyword is processed to determine the corresponding fuzzy keyword, which can improve the accuracy of question and answer.
[0088] Specifically, the second relationship condition is that the number of times each of the question keywords appears in each of the keyword index sets is not 0, and each of the question keywords does not correspond to the same target data set.
[0089] The present invention determines each question keyword based on the user's question and can be retrieved in each keyword index set, but each question keyword does not correspond to the same target data set, indicating that the retrieved target data set is not accurate enough and may miss relevant information, resulting in inaccurate answers to the questions, thereby affecting the user experience. Therefore, determining fuzzy keywords based on user questions can improve the accuracy of question and answer.
[0090] Specifically, the method of determining the retrieval method of the intermediate answer according to the attribute of the fuzzy keyword includes:
[0091] If the fuzzy keyword is a preset first attribute, determining an intermediate answer based on the fuzzy word index set;
[0092] If the fuzzy keyword is a preset second attribute, an intermediate answer is determined based on the pointing vector set.
[0093] Specifically, the preset first attribute is divergent. If the intermediate answer is determined based on the pointing vector set, a large number of intermediate answers may be determined, which will increase the amount of data processing. Therefore, the intermediate answer is determined based on the fuzzy word index set. The preset second attribute is contractive. If the intermediate answer is determined based on the fuzzy word index set, the intermediate answer may not be determined, which reduces the accuracy of the question and answer. Therefore, the intermediate answer is determined based on the pointing vector set.
[0094] In practice, the different attributes of fuzzy keywords will affect the accuracy of the intermediate answers and the amount of data processing. By setting different retrieval methods, the amount of data processing can be reduced and the accuracy of questions and answers can be improved.
[0095] Specifically, determining the intermediate answer based on the fuzzy word index set includes: determining a corresponding target data set according to the fuzzy keywords based on the fuzzy word index set, and generating an intermediate answer according to corresponding data in the target data set.
[0096] In implementation, the target data set corresponding to the fuzzy keyword can be directly determined based on the fuzzy word index set, thereby determining the intermediate answer, which can improve the retrieval efficiency.
[0097] Specifically, the determining of the intermediate answer based on the pointing vector set includes: determining the subordinate keywords of the fuzzy keywords based on the pointing vector set of the fuzzy words and keywords, and determining the target data set where the intermediate answer is located according to the subordinate keywords, and generating the intermediate answer according to the corresponding data in the target data set.
[0098] In implementation, the subordinate keywords of the fuzzy keywords are determined based on the pointing vector set of the fuzzy words and keywords, and the target data set where the intermediate answers are located is determined according to the subordinate keywords, which can reduce the amount of data processing.
[0099] An intelligent question-answering module, which is connected to the data source fusion module and the data set matching module respectively, and is used to determine the user answer keywords according to the user question, determine the answer corpus based on the target data set corresponding to the user answer, and output the corresponding answer form according to the answer corpus and a preset output model;
[0100] In implementation, a training sample set is determined based on historical data from several data sources, and an initial large language model is trained based on the training sample set to obtain a preset output model, wherein each training sample is a question and answer corpus determined based on each data, and the corresponding answer form is used as a label.
[0101] Specifically, the intelligent question-answering module includes:
[0102] A user answer keyword determination submodule, which is used to determine the user answer keywords according to the user question based on a preset answer keyword comparison table;
[0103] An answer corpus determination submodule, which is connected to the data source fusion module and the data set matching module respectively, and is used to determine the answer corpus based on the target data set corresponding to the user answer;
[0104] The answer form output submodule is connected to the answer corpus determination submodule and is used to output the corresponding answer form according to the answer corpus, the user question and the preset output model.
[0105] The intelligent question-and-answer module of the present invention determines the user answer keywords by setting a user answer keyword determination submodule, determines the answer corpus according to the target data set corresponding to the user answer by setting an answer corpus determination submodule, and outputs the corresponding answer form according to the answer form output submodule, thereby ensuring the accuracy of the user's answer.
[0106] The answer generation module is connected to the intelligent question and answer module and is used to output the user answer corresponding to the user question according to the user answer keywords and the answer form.
[0107] Specifically, the answer generation module includes:
[0108] An answer generation submodule, which is connected to the intelligent question and answer module and is used to match the user answer keywords and the answer form to generate a user answer;
[0109] The answer output submodule is connected to the answer generation submodule and is used to output the user answer corresponding to the user question.
[0110] The answer generation module of the present invention outputs the user answer corresponding to the user question by setting an answer generation submodule and an answer output submodule, which can ensure that the form of the user answer is relatively standard and improve the user experience.
[0111] The intelligent question-and-answer system for multi-source hospital databases provided by the present invention processes multi-source data to obtain several target data sets by setting a data source fusion module, constructs a keyword index set, a fuzzy word index set, and a pointing vector set of fuzzy words and keywords that are easy to search, which can reduce the amount of subsequent data processing and improve search efficiency. By setting a data set matching module, an intelligent question-and-answer module, and an answer generation module, the target data set corresponding to the user answer is determined based on the question keywords and fuzzy keywords, thereby determining the answer corpus to generate the answer form, and outputting the user answer corresponding to the user question based on the user answer keywords and the answer form, the accuracy of retrieving the user answer can be improved, thereby improving the accuracy of the intelligent question-and-answer, so as to improve the user experience.
[0112] In a specific embodiment, the user question is: Which products performed well in March?
[0113] The key words of the questions are: March, product, performance, and excellence;
[0114] It is found that product, performance and excellence are fuzzy keywords in each target data set, among which the attribute of "product" is the preset first attribute, and the intermediate answer is determined based on the fuzzy word index set. According to each fuzzy word index set, the data in the corresponding target data set with "product" is found, and the corresponding intermediate answer is generated according to the data. The attribute of "performance" is the preset first attribute, and the intermediate answer is determined based on the fuzzy word index set. According to each fuzzy word index set, the data in the corresponding target data set with "performance" is found, and the corresponding intermediate answer is generated according to the data. The attribute of "excellent" is the preset second attribute, and the intermediate answer is determined based on the pointing vector set. The lower keywords corresponding to "excellent" may be high sales volume, low cost, etc., and the data in the corresponding target data set with high sales volume and low cost is found, and the corresponding intermediate answer is generated according to the data. "March" is a keyword, and the data in the corresponding target data set is directly determined according to the keyword index set, and the corresponding intermediate answer is generated according to the data. The common target data set corresponding to the four question keywords "March, product, performance, excellent" is used as the target data set corresponding to the user answer. If the number of times "March" appears in each of the keyword index sets is 0, then "March" is set as the fuzzy keyword "month", and the attribute of "month" is the preset second attribute. The intermediate answer is determined based on the pointing vector set, and the data in the corresponding target data set of the subordinate keyword "March" corresponding to "month" is found according to the pointing vector set, and the corresponding intermediate answer is generated according to the data. If the number of times "March" appears in each of the keyword index sets is not 0, and each of the question keywords does not correspond to the same target data set, then the fuzzy keyword is determined according to the user question based on the preset fuzzy word comparison table, which may be "first quarter", and the attribute of "first quarter" is the preset second attribute. The intermediate answer is determined based on the pointing vector set, and the data in the corresponding target data set of the subordinate keyword "March" corresponding to "month" is found according to the pointing vector set, and the corresponding intermediate answer is generated according to the data. The answer corpus is determined based on the target data set corresponding to the user answer, and then the user answer keywords are determined according to the user question based on the preset answer keyword comparison table. The user answer keyword corresponding to "Which products performed well in March?" is excellent performance. The answer corpus and the user question are input into the preset output model to output the corresponding answer form, and the user answer keywords are matched with the answer form to generate a user answer and output it to the user.
[0115] In another specific embodiment, the user question is: How is Zhang San's condition?
[0116] It is found in each target data set that the condition is a fuzzy keyword, and the attribute of "condition" is the preset first attribute. The intermediate answer is determined based on the fuzzy word index set, and the data in the corresponding target data set with "condition" is found according to each fuzzy word index set, and the corresponding intermediate answer is generated according to the data. "Zhang San" is a keyword, and the data in the corresponding target data set is directly determined according to the keyword index set, and the corresponding intermediate answer is generated according to the data. The common target data set corresponding to the two question keywords "Zhang San" and "condition" is used as the target data set corresponding to the user answer, and the answer corpus is determined based on the target data set corresponding to the user answer, and then the user answer keyword is determined according to the user question based on the preset answer keyword comparison table. The user answer keyword corresponding to "How is Zhang San's condition?" is the condition, and the answer corpus and the user question are input into the preset output model to output the corresponding answer form, and the user answer keyword is matched with the answer form to generate a user answer and output it to the user.
[0117] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.
Claims
1. An intelligent question-answering system for multi-source hospital databases, characterized in that: include: A data source fusion module is used to obtain data from several data sources, classify the data in each data source according to the data type to generate several target data sets, and determine the keyword index set, fuzzy word index set and fuzzy word and keyword pointing vector set corresponding to each target data set; Each target dataset has a corresponding answer corpus; A data set matching module, which is connected to the data source fusion module, is used to determine the question keywords according to the user question, determine the fuzzy keywords according to the relationship between the question keywords and each of the keyword index sets, and determine the retrieval method of the intermediate answer according to the attributes of the fuzzy keywords, including determining the intermediate answer based on the fuzzy word index set and determining the intermediate answer based on the pointing vector set; And, determining a target data set corresponding to the user's answer based on the obtained intermediate answer and the question keyword; wherein the obtained intermediate answer is used as a keyword to replace the fuzzy keyword to form a new question keyword; The method of determining the intermediate answer according to the attributes of the fuzzy keyword includes: If the fuzzy keyword is a preset first attribute, determining an intermediate answer based on the fuzzy word index set; If the fuzzy keyword is a preset second attribute, determining an intermediate answer based on the pointing vector set; An intelligent question-answering module, which is connected to the data source fusion module and the data set matching module respectively, and is used to determine the user answer keywords according to the user question, determine the answer corpus based on the target data set corresponding to the user answer, and output the corresponding answer form according to the answer corpus and a preset output model; The answer generation module is connected to the intelligent question and answer module and is used to output the user answer corresponding to the user question according to the user answer keywords and the answer form.
2. The intelligent question-answering system for multi-source hospital database according to claim 1 is characterized in that: The data source fusion module includes: A data classification submodule, which is used to classify data in a plurality of data sources according to their data types to obtain a plurality of initial data sets; the data types of the data in each of the initial data sets are the same; A data preprocessing submodule, which is connected to the data classification submodule and is used to perform data preprocessing on the data in each initial data set to generate a plurality of target data sets; A keyword extraction submodule, which is connected to the data preprocessing submodule, is used to extract keywords of each data in each target data set, and to form fuzzy words based on the analysis of each keyword, so as to obtain a keyword index set and a fuzzy word index set corresponding to each target data set, and to determine a pointing vector set of fuzzy words and keywords according to the keywords and fuzzy words of each data in a number of target data sets; the fuzzy words and keywords have a superior-fewer relationship, and each fuzzy word has at least one corresponding keyword; Among them, if the keywords corresponding to a single fuzzy word are all located in the same target data set, the fuzzy word is collected into the fuzzy word index set corresponding to the target data set; if the keywords corresponding to a single fuzzy word are located in at least two target data sets, the fuzzy word is collected into the pointing vector set.
3. The intelligent question-answering system for multi-source hospital database according to claim 2 is characterized in that: The data set matching module comprises: A question keyword determination submodule, which is connected to the data source fusion module and is used to determine question keywords according to the user's question; there is at least one question keyword; A fuzzy keyword determination submodule, which is connected to the question keyword determination submodule and is used to determine the fuzzy keyword according to the relationship between the question keyword and each of the keyword index sets; in, Under the first relationship condition, the corresponding question keyword is processed according to a preset keyword fuzzy method to obtain a fuzzy keyword corresponding to the question keyword; Under the second relationship condition, determining fuzzy keywords according to the user question based on a preset fuzzy word comparison table; An intermediate answer determination submodule, which is connected to the fuzzy keyword determination submodule and is used to determine a retrieval method of an intermediate answer according to the attributes of the fuzzy keyword, including determining an intermediate answer based on the fuzzy word index set and determining an intermediate answer based on the pointing vector set; The data set matching submodule is connected to the intermediate answer determination submodule, and determines the target data set corresponding to the user's answer based on the obtained intermediate answer and question keywords.
4. The intelligent question-answering system for multi-source hospital database according to claim 3 is characterized in that: The first relationship condition is that there is at least one of the problem keywords and the number of occurrences in each of the keyword index sets is 0.
5. The intelligent question-answering system for multi-source hospital database according to claim 4 is characterized in that: The second relationship condition is that the number of times each of the question keywords appears in each of the keyword index sets is not 0, and each of the question keywords does not correspond to the same target data set.
6. The intelligent question-answering system for multi-source hospital database according to claim 5, characterized in that: The determining of the intermediate answer based on the fuzzy word index set includes: determining a corresponding target data set according to the fuzzy keywords based on the fuzzy word index set, and generating an intermediate answer according to corresponding data in the target data set.
7. The intelligent question-answering system for multi-source hospital database according to claim 6, characterized in that: The determining of the intermediate answer based on the pointing vector set includes: determining subordinate keywords of the fuzzy keywords based on the pointing vector set of the fuzzy words and keywords, and determining the target data set where the intermediate answer is located according to the subordinate keywords, and generating the intermediate answer according to the corresponding data in the target data set.
8. The intelligent question-answering system for multi-source hospital database according to claim 7, characterized in that: The intelligent question-answering module comprises: A user answer keyword determination submodule, which is used to determine the user answer keywords according to the user question based on a preset answer keyword comparison table; An answer corpus determination submodule, which is connected to the data source fusion module and the data set matching module respectively, and is used to determine the answer corpus based on the target data set corresponding to the user answer; The answer form output submodule is connected to the answer corpus determination submodule and is used to output the corresponding answer form according to the answer corpus, the user question and the preset output model.
9. The intelligent question-answering system for multi-source hospital database according to claim 8, characterized in that: The answer generation module comprises: An answer generation submodule, which is connected to the intelligent question and answer module and is used to match the user answer keywords and the answer form to generate a user answer; The answer output submodule is connected to the answer generation submodule and is used to output the user answer corresponding to the user question.
