CN102663129A - Medical field deep question and answer method and medical retrieval system - Google Patents
Medical field deep question and answer method and medical retrieval system Download PDFInfo
- Publication number
- CN102663129A CN102663129A CN2012101251578A CN201210125157A CN102663129A CN 102663129 A CN102663129 A CN 102663129A CN 2012101251578 A CN2012101251578 A CN 2012101251578A CN 201210125157 A CN201210125157 A CN 201210125157A CN 102663129 A CN102663129 A CN 102663129A
- Authority
- CN
- China
- Prior art keywords
- retrieval
- answer
- question
- database
- sentence
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 68
- 238000010801 machine learning Methods 0.000 claims abstract description 9
- 238000013480 data collection Methods 0.000 claims description 19
- 230000011218 segmentation Effects 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 17
- 239000000284 extract Substances 0.000 claims description 15
- 238000004458 analytical method Methods 0.000 claims description 6
- 239000012141 concentrate Substances 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 206010019233 Headaches Diseases 0.000 description 8
- 231100000869 headache Toxicity 0.000 description 8
- 201000010099 disease Diseases 0.000 description 7
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 7
- 239000003814 drug Substances 0.000 description 7
- 238000003745 diagnosis Methods 0.000 description 6
- 239000000463 material Substances 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000012797 qualification Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 230000035764 nutrition Effects 0.000 description 3
- 235000016709 nutrition Nutrition 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000001356 surgical procedure Methods 0.000 description 3
- 208000024891 symptom Diseases 0.000 description 3
- 206010037660 Pyrexia Diseases 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 230000013011 mating Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 235000015097 nutrients Nutrition 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000011282 treatment Methods 0.000 description 2
- BSYNRYMUTXBXSQ-UHFFFAOYSA-N Aspirin Chemical compound CC(=O)OC1=CC=CC=C1C(O)=O BSYNRYMUTXBXSQ-UHFFFAOYSA-N 0.000 description 1
- 244000188472 Ilex paraguariensis Species 0.000 description 1
- 241000700605 Viruses Species 0.000 description 1
- 229960001138 acetylsalicylic acid Drugs 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000013549 information retrieval technique Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
本发明提供一种深度问答方法，包括：步骤1、接收提问数据；步骤2、在知识库数据库中进行第一检索；所述知识库数据库包括百科中的事实性信息；和步骤3、在自定义数据集中进行第三检索，包括：步骤3.1、基于句型模式集，利用机器学习的方法给所述问题分类，确定问题的句型模式；和步骤3.2、用句型模式匹配问题，得到第二类关键字，用第二类关键字检索自定义数据集。与上述方法相对应的，本发明还提供一种医学检索系统，包括：输入模块；第一检索模块，用于检索知识库数据库；所述知识库数据库包括百科中的事实性信息；第三检索模块，用于检索自定义数据集。
The present invention provides an in-depth question answering method, comprising: step 1, receiving question data; step 2, performing a first search in a knowledge base database; the knowledge base database includes factual information in the encyclopedia; The third search is carried out in the defined data set, including: step 3.1, based on the sentence pattern set, using the method of machine learning to classify the problem, and determine the sentence pattern of the problem; and step 3.2, matching the problem with the sentence pattern to obtain the first Second-class keywords, use the second-class keywords to retrieve custom datasets. Corresponding to the above method, the present invention also provides a medical retrieval system, including: an input module; a first retrieval module for retrieving a knowledge base database; the knowledge base database includes factual information in the encyclopedia; a third retrieval module Module for retrieving custom datasets.
Description
Technical field
The present invention relates to medical information and handle and searching field, relate in particular to a kind of medical field degree of depth answering method and medical retrieval system.
Background technology
Computer aided technique has been penetrated into every field such as medical treatment, manufacturing, design, finance, commerce consultation at present, has quickened the development of all trades and professions.At medical field, medical expert's advisory system generally comprises the question and answer module, is used for system and patient or doctor and carries out alternately.
Because the answer meeting that medical expert's advisory system is returned influences patient's judgement, its result can cause the dual risk of life and property, so,, the accuracy of medical expert's advisory system can't extensively adopt before acquiring a certain degree.
The interrogation reply system of using in existing medical expert's advisory system mainly contains two kinds: community's question and answer and automatic question answering.
Community's question and answer belong to interpersonal question and answer, and the user submits a question in the website, and in the regular hour, other users answer this problem, and perhaps system returns associated answer according to the similar problem that had in the past.This will uncertain, the not accurate enough situation of answer form often occurs, and the time of answering a question be longer based on website user's the structure of knowledge.
The automatic question answering function that automatically request-answering system provides can be answered simple question, basically all is to open the field, and promptly the problem in any field can be imported, and causes the answer accuracy rate low like this, answers form fix, is not very hommization.
There are bottleneck in the knowledge that provides based on the expert consulting system of the medical domain of above-mentioned answering method and precision, the credibility of information, have restricted the development of medical expert's advisory system.
Summary of the invention
The technical matters that the present invention will solve provides medical field degree of depth answering method and medical retrieval system, improves question and answer precision as a result.
According to an aspect of the present invention, a kind of degree of depth answering method is provided, comprises:
Data are putd question in step 1, reception;
Step 2, in repository database, carry out first the retrieval; Said repository database comprises the factual information in the encyclopaedia; With
Step 3, concentrate in self-defining data and to carry out the 3rd retrieval, comprising:
Step 3.1, based on the sentence pattern set of patterns, utilize the method for machine learning to give said problem classification, the sentence pattern pattern of problem identificatioin; With
Step 3.2, use the sentence pattern pattern matching problem, obtain second type of key word, with second type of key search self-defining data collection.
Optional, described degree of depth answering method also comprises:
Step 4, carry out second retrieval in to database in question and answer; Said question and answer comprise that to database verified question and answer accurately are to information.
Optional, the generation method of the sentence pattern set of patterns of step 3.1 comprises:
Step 3.1.1, set up the self-defining data collection;
The sentence pattern set of patterns is set up in step 3.1.2a, manual work; And/or
Step 3.1.2b, through artificial labeled data, training data obtains the sentence pattern set of patterns.
Optional, said step 3 also comprises:
Step 3.3, utilize second type of keyword retrieval repository database.
Optional, step 1 comprises: from the text message of said enquirement extracting data problem.
Optional, step 2 comprises:
Step 2.1, be unit, each sentence is carried out word segmentation processing with the sentence in the problem;
Step 2.2, from the word segmentation processing result, extract first kind keyword; With
Step 2.3, utilize first kind keyword retrieval repository database.
Optional, step 4 comprises:
Step 4.1, retrieval question and answer are calculated said problem and the question and answer similarity to the record in the database to database; With
If step 4.2 exists similarity to reach the record of certain threshold value, according to the size of similarity said record is carried out rank, obtain the top n matching result, N is a natural number.
According to a further aspect of the present invention, a kind of medical retrieval system is provided, comprises:
Load module is used to receive the input from the user;
First retrieval module is used for the retrieval knowledge library database; Said repository database comprises the factual information in the encyclopaedia; With
The 3rd retrieval module is used to retrieve the self-defining data collection;
Wherein, said the 3rd retrieval module also comprises:
Sentence pattern pattern analysis module is used for based on the sentence pattern set of patterns, utilizes the method for machine learning to give said problem classification, the sentence pattern pattern of problem identificatioin; With
The second generic key word extracts and retrieval module, is used to utilize the sentence pattern pattern to come matching problem to obtain second type of key word, uses second type of key search self-defining data collection then.
Optional, said medical retrieval system also comprises:
Second retrieval module is used to retrieve question and answer to database; Said question and answer comprise that to database verified question and answer accurately are to information.
Optional, said second retrieval module also comprises:
Similarity calculation module is used to retrieve question and answer to database, calculates said problem and the question and answer similarity to the record in the database; With
Feedback module if exist similarity to reach the record of certain threshold value, carries out rank according to the size of similarity to said record as a result, obtains the top n matching result, and N is a natural number.
Optional, said load module also is used for from the text message of said enquirement extracting data problem.
Optional, said first retrieval module also comprises:
Word-dividing mode, the sentence that is used for problem is a unit, and each sentence is carried out word segmentation processing; With
First kind keyword extraction and retrieval module are used for extracting first kind keyword from the word segmentation processing result, utilize first kind keyword retrieval repository database then.
Compared with prior art, the invention has the advantages that: set up three kinds of databases, distinct type data-base is carried out the retrieval of distinct methods, improved the precision of result for retrieval; Wherein, extract the precision of all right further raising of attribute result for retrieval such as problem domain.
Description of drawings
Fig. 1 is the degree of depth answering method process flow diagram that provides in the one embodiment of the invention;
Fig. 2 is in the another embodiment of the present invention, the process flow diagram of step S20 among Fig. 1;
Fig. 3 is in the another embodiment of the present invention, the process flow diagram of step S30 among Fig. 1;
Fig. 4 is in the another embodiment of the present invention, the process flow diagram of step S40 among Fig. 1;
Fig. 5 is the generation method flow diagram of the sentence pattern set of patterns that provides in the another embodiment of the present invention;
Fig. 6 is the degree of depth question answering system structural representation that provides in the one embodiment of the invention.
Embodiment
In order to make the object of the invention, technical scheme and advantage clearer, below in conjunction with accompanying drawing, to further explain of the present invention.Should be appreciated that specific embodiment described herein only in order to explanation the present invention, and be not used in qualification the present invention.
(Question Answering System QA) is a kind of advanced form of information retrieval system to question answering system.It can use accurate, succinct natural language to answer the problem that the user proposes with natural language.The main cause that its research is risen is that people are to obtain the demand of information quickly and accurately.Question answering system is research direction that receives much attention and have the broad development prospect in present artificial intelligence and the natural language processing field.
The inventor finds after deliberation, based on the personal information and the health of patient's input, through in extensive text/data, excavating similar state of an illness information and returning the highest result of similarity (for example returning first three), can improve the question and answer precision.Wherein large-scale data comprises: (1) obtains factual information from encyclopaedia, the artificial knowledge hierarchy of setting up, set up " knowledge base "; (2) from community question and answer website, obtain question and answer, set up " question and answer are to database " information; (3) set up " self-defining data collection " by other data.
(1) repository database is called for short knowledge base, from the artificial knowledge hierarchy of setting up of encyclopaedia and other, obtains factual information, and with forming knowledge base in these information importing non-relational databases.Because the content of need clear in structure, mating easily when information such as importing encyclopaedia, is main with structured messages such as business card, form, tabulations mainly.The notion of some repetitions can be put in order through manual method; Be fused in the tangible system; For example: disease is divided into internal medicine, surgery, gynemetrics, paediatrics; Tell concrete disease below each subject again in detail; Every kind of disease all can have concrete symptom, every kind of symptom or disease be lack that which nutrient causes or infect that which virus causes or because other what reasons cause, every kind of disease generally be when, weather, which kind of environment morbidity be in the majority, need with what medicine or the like.All information of this method storage all are accurately, if mate successfully, the answer accuracy rate can be very high.Because not absolute contact between a lot of information in knowledge base the inside, generally without relational database, and use non-relational database in the time of storage, for example MongoDB.
(2) question and answer are from there are the information such as community of medical science relevant issues and answer community question and answer website etc., to obtain to the data in the database.Answer this complicacy of disease, highly professional problem; Best bet is exactly borrower's a strength, and the problem of in community, dishing out is answered by the user who knows the answer; Query person selects oneself satisfied answer, and the user of similar demand is recommended in these answers.This method has improved the accuracy of answer, if can the data centralization in these websites be got up, returns suitable answer to the user quickly and easily, then can very practical reference be provided for the user.
Question and answer are that question and answer are right to the data in the database, and the form that question and answer are right is: problem+optimum answer+other answers.When question and answer are set up database, at first will mark problem, for example, the mark problem belongs to which kind of problem in " internal medicine, surgery, gynemetrics, paediatrics ", and the classification label symbol is as shown in table 1.Then, machine can produce series of parameters according to the word and the classification of manual work mark, and these parameters have been represented the relation of word and contextual relation, word and classification.After manual work mark, machine generated parameter, for the word of unknown classification, system just can automatically identify the classification of word.
Because the relevant disease under the such section office of internal medicine, surgery, gynemetrics, paediatrics, medicine etc. have remarkable difference; If so can before retrieval, just can confirm the affiliated field of customer problem through automatic identification; Can dwindle range of search, get rid of extraneous data, make the result more accurate.
Table 1
(3) data of self-defining data collection can from: (a) retrieve, wherein comprise based on existing encyclopaedia language material: universal experience knowledge and objective fact etc., scope is wider, can be used for common knowledge question and medical diagnosis; (b) extract based on the answer of medical science voluminous dictionary, can be used for professional knowledge inquiry and medical diagnosis; (c) extract based on the answer of nutrition knowledge collection, comprise relevant knowledge such as books handbook etc., can be used to expand knowledge query and medical diagnosis, the suggestion of medical science voluminous dictionary; (d) open retrieval promptly at existing network retrieval system (for example Google) input keyword, is returned relevant documentation, in document, extracts the bigger part of probability as related data, can be used for the question and answer in open field, and is relevant with daily life.
The file layout of self-defining data intensive data is complicated, comprising: (a), the encyclopaedia language material of html format.Knowledge base before is that information is stored in the non-relational database with textual form, is with the original HTML stores of information at this moment.(b), medical science voluminous dictionary and nutrition knowledge collection all be with the textual form storage, comprises that form is like TXT, PDF, WORD.
Based on above-mentioned database,, a kind of degree of depth answering method is provided according to one embodiment of the invention.As shown in Figure 1, this method comprises:
Step S10, reception user's input (promptly puing question to data); These enquirement data can be text input, phonetic entry, image input etc.; Being input as example with text below describes;
Step S20, in repository database, retrieve (be also referred to as first retrieval); If first retrieval obtains required answer, finish; If first retrieval does not obtain required answer, carry out step S30;
Step S30, retrieve (being also referred to as second retrieval) in to database in question and answer; If second retrieval obtains required answer, finish; If second retrieval does not obtain required answer, carry out step S40;
Step S40, concentrate in self-defining data and to retrieve (being also referred to as the 3rd retrieval); If the 3rd retrieval obtains required answer, finish; If the 3rd retrieval does not obtain required answer, then guide the user to use other problems to describe, for example, the guiding user further or from different perspectives (for example other symptoms, patient's medical history, once make a definite diagnosis result etc.) describe.
According to one embodiment of present invention; In step S10; The user also can put question to or put question to through image through voice, in these cases, and after the enquirement data that receive the user; Carry out speech recognition (or speech recognition reciprocal process) or image recognition (or image recognition reciprocal process), confirm text message.
According to one embodiment of present invention,, also can utilize the keyword in the question sentence to carry out open retrieval, promptly utilize existing commercial search engine to retrieve if step S40 does not obtain required answer.
According to a further embodiment of the invention, as shown in Figure 2, step S20 first retrieval also comprises:
S201 is a unit to put question to the sentence in the data, and each sentence is carried out word segmentation processing (being also referred to as entryization);
S202 extracts keyword from the word segmentation processing result; Field under all right problem identificatioin;
S203 utilizes the keyword retrieval repository database; With
S204 if find required answer, is converted into natural language with answer and returns.
According to one embodiment of present invention, in step S201, word segmentation processing can utilize the Chinese word segmentation assembly of increasing income to carry out, for example: Pan Gu's participle, MMSEG4J, Paoding, CC-CEDICT, IK, ICTCLAS etc.Word segmentation processing also can be carried out based on the segmenting method of statistics, and combines named entity recognition, word sense disambiguation; Named entity recognition is to the identification like speech such as name, place name, mechanism's name, place name, ProductName, trade (brand) name, abbreviation, ellipsis; The effect of word sense disambiguation is to reduce wrong understanding.At last, sentence is cut into independent one by one speech.
According to one embodiment of present invention, in step S202, from word segmentation result, extract keyword and can carry out in two ways:
First kind of mode, remove stop words (speech such as ",, what ") after, stay the speech in the dictionaries such as named entity.For example, stop words, common subject (as you, I etc.) do not belong to the speech in the dictionary; The corresponding synon colloquial style saying of named entity, common specialized word (as, headache, aspirin etc.) belong to the speech in the dictionary.
The second way; After confirming the sentence pattern pattern, can also be corresponding with the position in the sentence pattern pattern speech suitable in the sentence, the extracting section that selection needs is come out; For example; " subject+predicate+illness ", the selected ci poem corresponding " illness " comes out as keyword, and the second way will be further explained below.
According to one embodiment of present invention, step S202 can further include:
S202a, keyword expansion; Keyword expansion has method in common, and for example " painful around the temple " " fever " can expand " headache " speech such as " fevers ", needs to use the synonym in the dictionary to expand;
S202b, part-of-speech tagging; Part-of-speech tagging has method in common, identifies noun in the sentence, verb, adjective etc.; With
S202c, the synonym expansion; Synonym expansion is close a bit with keyword expansion, represents a meaning but synonym is meant two speech, can mutual alternative, and for example " headache " can replace with " headache ".
Above-mentioned substep S202a～S202c can produce more multi-key word.
In step S203, the above-mentioned keyword that utilizes step S202 to generate comes the retrieval knowledge library database.If above-mentioned steps S203 does not find required answer, carry out step S30.
In step S204, the answer of returning in the knowledge base is simple, stiff, can become than the complex natural linguistic form.If the sentence pattern pattern of problem is made up of " statement+problem ", can replace problem with answer, become " statement+answer " and return to the user.
In other embodiment of the present invention; In order further to improve the accuracy of understanding to customer problem; Not all sentence pattern all only extracts key word with regard to (promptly can not ignore the effect of other speech in the sentence fully) much of that; To the speech of the normal association area of using of user in the reality test, also can such speech be listed in the lists of keywords.
Another embodiment is as shown in Figure 3 according to the present invention, and step S30 further comprises:
S301 retrieval question and answer are to database; When retrieving question and answer to database; Calculate the similarity of sentence; In the present embodiment, adopt support vector machine method, whether inspection customer problem and question and answer are enough big to the problem similarity in the database; Promptly whether reach preset threshold value, if enough greatly just return the corresponding answer of problem in this database.
It will be understood by those skilled in the art that then similarity can improve if comprise keyword in the sentence.
In addition, because the probability that matees in practical application fully is very low, so can utilize machine learning algorithm, searching and question sentence similarity maximum and similarity are returned the corresponding answer of this sentence greater than the sentence of certain threshold value.Can improve matching rate like this, more tally with the actual situation simultaneously.
If the problem that S302 retrieves in to database in question and answer and the similarity of customer problem reach certain threshold value, according to the size of similarity result for retrieval is carried out rank, obtain the top n matching result.
S303 is converted into natural language with the top n matching result and returns; If do not find required answer, execution in step S40.
Another embodiment is as shown in Figure 4 according to the present invention, and step S40 further comprises:
S401, problem identificatioin field and answer type; Through qualification answer field, reject the irrelevant candidate answers of a part (those records that promptly its domain attribute of eliminating and determined problem domain the database had nothing to do) from question and answer, can further improve recall precision.
Utilize part-of-speech tagging to confirm which word is keyword in the sentence, can confirm the field of keyword, thereby obtain the field of problem in conjunction with dictionary.The answer type also can be passed through key word recognition, for example, if occur in the keyword " how much ", the problem identificatioin type is a quantity.
S402, based on the sentence pattern set of patterns, utilize the method for machine learning to give question sentence (being problem) classification, the sentence pattern pattern of problem identificatioin.
Concrete, according to the classification in the sentence pattern set of patterns (being the sentence pattern pattern), give the problem mark classification of language material, learn to obtain series of parameters with the method for machine learning.These parameters have been represented the relation of speech and speech, can confirm through these parameters which sentence pattern pattern is new problem belong to.The foundation of sentence pattern set of patterns will go through below.
S403, use the sentence pattern pattern matching problem, obtain new key word, with new key search self-defining data collection.
Concrete, replace the keyword in the target sentences with specific forms, make the expression that similar sentence can corresponding some template (can utilize regular expression to carry out rule match.For example; I+any word+have+any word+illness; Regular expression can be expressed as " I ([]+?) have ([]+?) iDisease ", the speech (iDisease) of wherein representing illness is all in dictionary, as long as the speech that this dictionary the inside has occurs; Just be matched to merit at last, this speech of this position is exactly the speech that needs).For example, " I have a headache recently ", utilize " I ([]+?) iDisease arranged " pattern; Can match " recently " and " headache " two speech; Contrast with the word in the dictionary, can mate successfully with " headache ", that headache is exactly a keyword.
If S404 concentrates in self-defining data and retrieves data matching, according to the degree of correlation (being similarity) result for retrieval is carried out rank, obtain the top n matching result.
The sentence and the speech of answer position in S405, the selector matched moulds plate.
Go on foot N the matching result that obtains, the keyword in the matching template (being the sentence pattern pattern) and corresponding answer type, speech and sentence after obtaining mating to last one.
S406, select the high answer of the frequency of occurrences and return.
To speech and the sentence that a last step obtains, add up its frequency of occurrences, the reasonable distribution weight sorts, and the answer that the selection frequency of occurrences is the highest is also returned.
S204 is similar with step, also can return answer with the natural language form.
The keyword that the result of sentence pattern pattern match obtains generally is the subclass of the keyword of above-mentioned first method (S202) extraction; (retrieval knowledge library database for example in retrieving; The self-defining data collection); The weight of the keyword in this subclass is higher than the weight of other keywords, can improve recall precision and accuracy; The key word that the key word additional step S20 that promptly utilizes step S40 to obtain obtains comes databases such as retrieval knowledge library database, self-defining data collection, can improve recall precision and accuracy.
According to a further embodiment of the invention, as shown in Figure 5, a kind of generation method of sentence pattern set of patterns is provided, it comprises:
S501, set up the self-defining data collection;
The sentence pattern set of patterns is set up in S502a, manual work; And/or
S502b, through artificial labeled data, training data obtains the sentence pattern set of patterns.
In step S501, as stated, the data of self-defining data collection can comprise: (1) is retrieved based on existing encyclopaedia language material; (2) extract based on the answer of medical science voluminous dictionary; (3) extract based on the answer of nutrition knowledge collection; The webpage (document that comprises various forms) that obtains when (4) opening retrieval with various keywords.
In step S502a; Rule template is set up in manual work needs check, and several kinds of templates of general design are earlier tested on existing sentence; If being arranged, sentence can't confirm that perhaps sentence pattern is omitted; Omitted these or need improved sentence pattern schema modification to come, tested once more, till including most of sentence pattern.
In step S502b; At first get the problem of a plurality of (for example 10,000) medical aspect, mark into a kind of sentence pattern (for example artificial mark) to each problem, the sentence that these had marked is learnt with maximum entropy model; Obtain a plurality of parameters, these parameters are used for the sentence pattern pattern of problem identificatioin.
Further; Choose a plurality of (for example 500) again and marked but the sentence of not learning (being that it doesn't matter for Determination of Parameters and these sentences), through the classification of these sentences of parameter prediction before, corresponding two classification of each sentence at this moment; Whether the classification of contrast machine is consistent with the classification that marks before; If inconsistent, increase the quantity of learning sample, reach predetermined threshold value up to the quantity of unanimity.Can further guarantee to utilize these parameters to generate the accuracy of the sentence pattern pattern of question sentence like this, thereby improve the precision of result for retrieval.
According to another embodiment of the invention; After above-mentioned searching step S20, S30 and S40 return optimum answer, can also recommend the user whether to go large hospital to make a definite diagnosis, use which kind of medicine, points for attention etc. through the illness, nutrient, the medicine that relate in the optimum answer.
In the above-described embodiments, according to the order of accuarcy of the data content of storing in the database that each step was directed against, designed the execution sequence of step S20 to S40.It will be understood by those skilled in the art that according to other retrievals of carrying out in proper order three databases and can realize that also the present invention improves the purpose of result for retrieval precision.In addition; Even carry out S20, S30 or S40 separately,, also can improve the precision of result for retrieval because the present invention has improved search method (will be described in more detail below); Be that present embodiment is a preferred embodiment, whether each step is all carried out and execution sequence in the not qualification method.
And, even a searching step obtains required result, also can continue to carry out follow-up searching step, thereby, perhaps select more accurately for the user provides more selection.
According to one embodiment of the invention, a kind of degree of depth answering method is provided, comprising:
Step T10, reception user's input (promptly puing question to data); These enquirement data can be text input, phonetic entry, image input etc.; Being input as example with text below describes;
Step T20, in repository database, retrieve (above-mentioned first retrieval);
Step T30, concentrate in self-defining data and to retrieve (above-mentioned the 3rd retrieval).
Wherein, step T30 obtains utilizing this keyword to carry out the retrieval of self-defining data collection than the more accurate keyword of the keyword among the step T10 through the sentence pattern pattern of utilizing problem.
Wherein, the more accurate keyword that step T30 obtains can be used for the retrieval knowledge library database once more, for the user provides answer more accurately.
This method can also comprise: retrieve (above-mentioned second retrieval) in to database in question and answer.This step can before step T20 and the T30, between or carry out afterwards.
Based on above-mentioned degree of depth answering method,, a kind of degree of depth question answering system/medical retrieval system is provided according to one embodiment of the invention.As shown in Figure 6, this system 100 comprises: load module 101, first retrieval module 102, second retrieval module 103 and the 3rd retrieval module 104; The repository database 201 that three retrieval modules connect respectively, question and answer are to database 202 and self-defining data collection 203.Wherein, first, second, third retrieval module 102,103 and 104 working method are: if the previous stage retrieval module obtains required result then finishes, do not call the next stage retrieval module and retrieve if previous retrieval module obtains required result.
Concrete, load module 101 is used to receive the input from the user; User's input comprises phonetic entry, literal input etc., and under the situation of phonetic entry, load module 101 utilizes existing speech recognition system that phonetic entry is converted into the literal input.Load module 101 sends to first retrieval module 102 with ready literal input.
After first retrieval module 102 receives the literal input; At first question sentence is carried out word segmentation processing; From the result of word segmentation processing, extract keyword then and keyword is expanded; Utilize the result of expanded keyword in repository database 201, to retrieve at last,, start second retrieval module 103 if do not retrieve required result.The search method of first retrieval module 102 is aforesaid step S20, repeats no more here.
Preferably, first retrieval module 102 also comprises: word-dividing mode, and the sentence that is used for problem is a unit, and each sentence is carried out word segmentation processing; With first kind keyword extraction and retrieval module, be used for extracting first kind keyword from the word segmentation processing result, utilize first kind keyword retrieval repository database then.
Preferably, second retrieval module 103 also comprises similarity calculation module, is used to retrieve question and answer to database, calculates said problem and the question and answer similarity to the record in the database; With feedback module as a result, if exist similarity to reach the record of certain threshold value, said record is carried out rank according to the size of similarity, obtain the top n matching result, N is a natural number.
The 3rd retrieval module 104 is at first based on the sentence pattern set of patterns; Obtain the sentence pattern pattern of problem sentence; Utilize the sentence pattern pattern to obtain keyword (being called second type of keyword) more accurately then, utilize second type of keyword and sentence pattern pattern retrieval self-defining data collection 203, for example utilize regular expression to carry out rule match; According to the degree of correlation result for retrieval is carried out rank, return the top n matching result.The search method of the 3rd retrieval module 104 is aforesaid step S40, repeats no more here.
Preferably, said the 3rd retrieval module also comprises: sentence pattern pattern analysis module, be used for based on the sentence pattern set of patterns, and utilize the method for machine learning to give said problem classification, the sentence pattern pattern of problem identificatioin; Extract and retrieval module with the second generic key word, be used to utilize the sentence pattern pattern to come matching problem, use second type of key search self-defining data collection then to obtain second type of key word.
In the present embodiment,, designed the execution sequence of retrieval module 102 to 104 according to the order of accuarcy of the data content of storing in the database that each module was directed against.It will be understood by those skilled in the art that carrying out three retrieval modules in proper order according to other can realize that also the present invention improves the purpose of result for retrieval precision.In addition; Even only comprise one or two (second retrieval module 103 and the 3rd retrieval module 104 comprise one at least) in the retrieval module 102 to 104 in the system; Owing to improved search method (as indicated above); Also can improve the precision of result for retrieval, promptly present embodiment is a preferred embodiment, does not limit whether each module all comprises and execution sequence in three modules.
1), method and system provided by the invention reallocates resources, and sets up large scale database compared with prior art:, information is comprehensive, and is easy-to-look-up, helps adjusting parameter, improves accuracy, reduces the effect of interfere information; 2), native system is on the data basis of question and answer community; Adjustment parameter and degree of correlation requirement, and for example return three optimum answers, make answer more clear; When the answer degree of correlation is not enough; The answer of selecting for use additive method to return so also can avoid the user in question and answer community, oneself to sum up error message, for the crowd of amateur background important meaning is arranged more; 3), native system proposes and has used a kind of mark standard; In medical treatment qualification field, better effect is arranged; Be fit to native system more, and at language material of large-scale data mark like this, for from now on further investigation lays the first stone; For the researchist of back provides powerful support, this mark standard and the language material that has marked all are that the high precision of system provides support and prepares.
Though carry out explanation as an example with medical domain in the foregoing description, be appreciated that said method and system can also be applied in the data processing of other field, retrieval and the analysis, for example the long-distance education field to method and system.
The advantage of method and system provided by the invention comprises 1) set up three kinds of databases, distinct type data-base is carried out the retrieval of distinct methods, improved the precision of result for retrieval; Except utilizing keyword retrieves, to database, the similarity that rounds a sentence is mated to question and answer; Wherein, to the self-defining data storehouse, attributes such as the sentence pattern pattern of extraction problem, problem domain are to improve the precision of result for retrieval; 2) information retrieval technique is applied to the automatic question answering field; The inference method of abandoning tradition is from the accuracy of new angle increase automatic question answering, information such as tap/dip deep into medical field relevant professional knowledge, general knowledge; Therefrom excavate correct option, improved the precision of result for retrieval; 3) to the characteristic analysis of medical field, be suitable for being applied in the medical consultation system, realize intelligent medical analysis and diagnosis system based on degree of depth dialogue retrieve.
Though method and system provided by the invention all combines medical domain to describe in the above-described embodiments; But one of ordinary skill in the art will appreciate that; Method and system of the present invention also can be applied to any field; Through domain knowledge being decomposed into different levels, different levels is utilized different search method retrievals, can obtain the higher result of accuracy.
Should be noted that and understand, under the situation that does not break away from the desired the spirit and scope of the present invention of accompanying Claim, can make various modifications and improvement the present invention of above-mentioned detailed description.Therefore, the scope of the technical scheme of requirement protection does not receive the restriction of given any specific exemplary teachings.
Claims (12)
1. degree of depth answering method comprises:
Data are putd question in step 1, reception;
Step 2, in repository database, carry out first the retrieval; Said repository database comprises the factual information in the encyclopaedia; With
Step 3, concentrate in self-defining data and to carry out the 3rd retrieval, comprising:
Step 3.1, based on the sentence pattern set of patterns, utilize the method for machine learning to give said problem classification, the sentence pattern pattern of problem identificatioin; With
Step 3.2, use the sentence pattern pattern matching problem, obtain second type of key word, with second type of key search self-defining data collection.
2. degree of depth answering method according to claim 1 also comprises:
Step 4, carry out second retrieval in to database in question and answer; Said question and answer comprise that to database verified question and answer accurately are to information; This step was carried out before or after step 2.
3. degree of depth answering method according to claim 1, the generation method of the sentence pattern set of patterns of step 3.1 comprises:
Step 3.1.1, set up the self-defining data collection;
The sentence pattern set of patterns is set up in step 3.1.2a, manual work; And/or
Step 3.1.2b, through artificial labeled data, training data obtains the sentence pattern set of patterns.
4. degree of depth answering method according to claim 1, said step 3 also comprises:
Step 3.3, utilize second type of keyword retrieval repository database.
5. degree of depth answering method according to claim 1, wherein, step 1 comprises: from the text message of said enquirement extracting data problem.
6. degree of depth answering method according to claim 1, wherein, step 2 comprises:
Step 2.1, be unit, each sentence is carried out word segmentation processing with the sentence in the problem;
Step 2.2, from the word segmentation processing result, extract first kind keyword; With
Step 2.3, utilize first kind keyword retrieval repository database.
7. degree of depth answering method according to claim 2, wherein, step 4 comprises:
Step 4.1, retrieval question and answer are calculated said problem and the question and answer similarity to the record in the database to database; With
If step 4.2 exists similarity to reach the record of certain threshold value, according to the size of similarity said record is carried out rank, obtain the top n matching result, N is a natural number.
8. medical retrieval system comprises:
Load module is used to receive the input from the user;
First retrieval module is used for the retrieval knowledge library database; Said repository database comprises the factual information in the encyclopaedia; With
The 3rd retrieval module is used to retrieve the self-defining data collection;
Wherein, said the 3rd retrieval module also comprises:
Sentence pattern pattern analysis module is used for based on the sentence pattern set of patterns, utilizes the method for machine learning to give said problem classification, the sentence pattern pattern of problem identificatioin; With
The second generic key word extracts and retrieval module, is used to utilize the sentence pattern pattern to come matching problem to obtain second type of key word, uses second type of key search self-defining data collection then.
9. medical retrieval according to claim 8 system also comprises:
Second retrieval module is used to retrieve question and answer to database; Said question and answer comprise that to database verified question and answer accurately are to information.
10. medical retrieval according to claim 9 system, wherein, said second retrieval module also comprises:
Similarity calculation module is used to retrieve question and answer to database, calculates said problem and the question and answer similarity to the record in the database; With
Feedback module if exist similarity to reach the record of certain threshold value, carries out rank according to the size of similarity to said record as a result, obtains the top n matching result, and N is a natural number.
11. medical retrieval according to claim 8 system, wherein, said load module also is used for from the text message of said enquirement extracting data problem.
12. medical retrieval according to claim 8 system, wherein, said first retrieval module also comprises:
Word-dividing mode, the sentence that is used for problem is a unit, and each sentence is carried out word segmentation processing; With
First kind keyword extraction and retrieval module are used for extracting first kind keyword from the word segmentation processing result, utilize first kind keyword retrieval repository database then.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012101251578A CN102663129A (en) | 2012-04-25 | 2012-04-25 | Medical field deep question and answer method and medical retrieval system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012101251578A CN102663129A (en) | 2012-04-25 | 2012-04-25 | Medical field deep question and answer method and medical retrieval system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102663129A true CN102663129A (en) | 2012-09-12 |
Family
ID=46772620
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012101251578A Pending CN102663129A (en) | 2012-04-25 | 2012-04-25 | Medical field deep question and answer method and medical retrieval system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102663129A (en) |
Cited By (58)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103235787A (en) * | 2013-03-28 | 2013-08-07 | 北京百度网讯科技有限公司 | Method and equipment used for providing target information on target object |
| CN103955449A (en) * | 2014-04-21 | 2014-07-30 | 安一恒通（北京）科技有限公司 | Target sample positioning method and device |
| CN104199987A (en) * | 2014-09-28 | 2014-12-10 | 北京奇虎科技有限公司 | Method for displaying search result pages according to user-defined ordering, and search server |
| CN104239573A (en) * | 2014-10-08 | 2014-12-24 | 无锡慧方科技有限公司 | Medical system with novel open and integrated type search engine |
| CN104504163A (en) * | 2015-01-21 | 2015-04-08 | 北京智富者机器人科技有限公司 | Robot vision knowledge retrieval system |
| WO2015188425A1 (en) * | 2014-06-10 | 2015-12-17 | 网易公司 | Method and device for questioning and answering |
| CN105471712A (en) * | 2015-11-25 | 2016-04-06 | 深圳狗尾草智能科技有限公司 | Robot reply system and reply method thereof |
| CN105653840A (en) * | 2015-12-21 | 2016-06-08 | 青岛中科慧康科技有限公司 | Similar case recommendation system based on word and phrase distributed representation, and corresponding method |
| CN105701254A (en) * | 2016-03-09 | 2016-06-22 | 北京搜狗科技发展有限公司 | Information processing method and device and device for processing information |
| CN105786977A (en) * | 2016-02-05 | 2016-07-20 | 北京百度网讯科技有限公司 | Mobile search method and device based on artificial intelligence |
| CN106294341A (en) * | 2015-05-12 | 2017-01-04 | 阿里巴巴集团控股有限公司 | A kind of Intelligent Answer System and theme method of discrimination thereof and device |
| CN106295168A (en) * | 2016-08-08 | 2017-01-04 | 安徽易联众信息技术有限公司 | Medical information inquiry system based on speech recognition |
| CN106530159A (en) * | 2016-11-09 | 2017-03-22 | 平顶山学院 | Simulated teaching method for clinical medicine |
| CN106663125A (en) * | 2014-08-21 | 2017-05-10 | 国立研究开发法人情报通信研究机构 | Question sentence generation device and computer program |
| CN106682387A (en) * | 2016-10-26 | 2017-05-17 | 百度国际科技（深圳）有限公司 | Method and device used for outputting information |
| CN106844686A (en) * | 2017-01-26 | 2017-06-13 | 武汉奇米网络科技有限公司 | Intelligent customer service question and answer robot and its implementation based on SOLR |
| CN106919577A (en) * | 2015-12-24 | 2017-07-04 | 北京奇虎科技有限公司 | Based on method, device and search engine that search word scans for recommending |
| CN107180037A (en) * | 2016-03-09 | 2017-09-19 | 北京京东尚科信息技术有限公司 | Man-machine interaction method and device |
| CN107247868A (en) * | 2017-05-18 | 2017-10-13 | 深思考人工智能机器人科技（北京）有限公司 | A kind of artificial intelligence aids in interrogation system |
| CN107451388A (en) * | 2016-05-31 | 2017-12-08 | 百度（美国）有限责任公司 | For the methods, devices and systems for automating medical diagnosis |
| CN107491655A (en) * | 2017-08-31 | 2017-12-19 | 康安健康管理咨询（常熟）有限公司 | Liver diseases information intelligent consultation method and system based on machine learning |
| CN107610740A (en) * | 2017-07-27 | 2018-01-19 | 康美健康云服务有限公司 | Semantic analysis, electronic equipment, storage medium and system for medical treatment |
| CN107609017A (en) * | 2017-08-04 | 2018-01-19 | 陈剑辉 | The method and system of medical industry intelligent search consulting are realized by self-defined hot word |
| CN107766536A (en) * | 2017-10-30 | 2018-03-06 | 江西博瑞彤芸科技有限公司 | The searching method of related information |
| CN107993724A (en) * | 2017-11-09 | 2018-05-04 | 易保互联医疗信息科技（北京）有限公司 | A kind of method and device of medicine intelligent answer data processing |
| CN108108449A (en) * | 2017-12-27 | 2018-06-01 | 哈尔滨福满科技有限责任公司 | A kind of implementation method based on multi-source heterogeneous data question answering system and the system towards medical field |
| CN108595696A (en) * | 2018-05-09 | 2018-09-28 | 长沙学院 | A kind of human-computer interaction intelligent answering method and system based on cloud platform |
| CN108920603A (en) * | 2018-06-28 | 2018-11-30 | 厦门快商通信息技术有限公司 | A kind of customer service bootstrap technique based on customer service machine mould |
| CN108920488A (en) * | 2018-05-14 | 2018-11-30 | 平安科技（深圳）有限公司 | The natural language processing method and device that multisystem combines |
| CN108932349A (en) * | 2018-08-17 | 2018-12-04 | 齐鲁工业大学 | Medical automatic question-answering method and device, storage medium, electronic equipment |
| CN109036554A (en) * | 2018-08-07 | 2018-12-18 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating information |
| CN109086272A (en) * | 2018-08-01 | 2018-12-25 | 浙江蓝鸽科技有限公司 | Sentence pattern recognition methods and its system |
| CN109448857A (en) * | 2018-10-17 | 2019-03-08 | 南京理工大学 | Medical advice information fusion analysis method |
| CN109726269A (en) * | 2017-10-30 | 2019-05-07 | 松下知识产权经营株式会社 | Information processing method, information processing device, and program |
| CN109740928A (en) * | 2018-12-29 | 2019-05-10 | 北京中电普华信息技术有限公司 | Power consulting business processing method and device |
| CN109857847A (en) * | 2019-01-15 | 2019-06-07 | 北京搜狗科技发展有限公司 | A kind of data processing method, device and the device for data processing |
| CN109885823A (en) * | 2017-12-01 | 2019-06-14 | 武汉楚鼎信息技术有限公司 | A kind of distributed semantic recognition methods of financial industry and system and device |
| CN110019698A (en) * | 2017-09-04 | 2019-07-16 | 珠海健康云科技有限公司 | A kind of intelligent Service method and system of medicine question and answer |
| CN110111882A (en) * | 2011-10-24 | 2019-08-09 | 哈佛大学校长及研究员协会 | Enhancing diagnosis is carried out to illness by artificial intelligence and mobile health approach, in the case where not damaging accuracy |
| CN110168544A (en) * | 2016-12-27 | 2019-08-23 | 夏普株式会社 | Response device, response device control method, and control program |
| CN110727781A (en) * | 2019-10-21 | 2020-01-24 | 国网江苏省电力有限公司电力科学研究院 | Power multi-source knowledge retrieval result fusion method and device |
| CN110874403A (en) * | 2018-08-29 | 2020-03-10 | 株式会社日立制作所 | Question answering system, question answering processing method and question answering integrated system |
| CN111125150A (en) * | 2019-12-26 | 2020-05-08 | 成都航天科工大数据研究院有限公司 | Industrial field question-answering system retrieval method |
| CN112035627A (en) * | 2020-07-27 | 2020-12-04 | 深圳技术大学 | Automatic question answering method, device, equipment and storage medium |
| CN112052317A (en) * | 2020-08-19 | 2020-12-08 | 杭州古珀医疗科技有限公司 | Medical knowledge base intelligent retrieval system and method based on deep learning |
| WO2020244534A1 (en) * | 2019-06-05 | 2020-12-10 | 京东方科技集团股份有限公司 | Medical questioning and answering method, medical questioning and answering system, electronic device and computer-readable storage medium |
| CN112507179A (en) * | 2020-12-11 | 2021-03-16 | 杭州依图医疗技术有限公司 | Medical data processing method and retrieval method, device and storage medium |
| CN113609830A (en) * | 2021-04-07 | 2021-11-05 | 新大陆数字技术股份有限公司 | Literature question-answering method, system and storage medium based on NLP technology |
| CN114303141A (en) * | 2019-10-01 | 2022-04-08 | 杰富意钢铁株式会社 | Information retrieval system |
| CN114548100A (en) * | 2022-03-01 | 2022-05-27 | 深圳市医未医疗科技有限公司 | Clinical scientific research auxiliary method and system based on big data technology |
| CN114820040A (en) * | 2022-03-31 | 2022-07-29 | 江苏未来智慧信息科技有限公司 | Electric power intelligent consultation control system and control mode thereof |
| US11568997B2 (en) | 2019-07-23 | 2023-01-31 | International Business Machines Corporation | Dynamic context-based collaborative medical concept interpreter |
| CN117133448A (en) * | 2023-10-23 | 2023-11-28 | 长春中医药大学 | A gynecological disease consultation system based on big data |
| US11972336B2 (en) | 2015-12-18 | 2024-04-30 | Cognoa, Inc. | Machine learning platform and system for data analysis |
| CN118643136A (en) * | 2024-06-17 | 2024-09-13 | 北京法伯宏业科技发展有限公司 | An intelligent question answering system for multi-source hospital database |
| CN118689967A (en) * | 2024-08-23 | 2024-09-24 | 中科厚立信息技术(成都)有限公司 | Question query method, device, electronic device and computer readable storage medium |
| US12205725B2 (en) | 2016-11-14 | 2025-01-21 | Cognoa, Inc. | Methods and apparatus for evaluating developmental conditions and providing control over coverage and reliability |
| US12402840B2 (en) | 2015-08-11 | 2025-09-02 | Cognoa, Inc. | Methods and apparatus to determine developmental progress with artificial intelligence and user input |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2002075588A1 (en) * | 2001-03-21 | 2002-09-26 | Inter China Network Software Company Limited | System and method of personal and business web cards |
| CN1492367A (en) * | 2002-09-27 | 2004-04-28 | 株式会社东芝 | Inquiry-response system and inquiry-response method |
| CN1928864A (en) * | 2006-09-22 | 2007-03-14 | 浙江大学 | FAQ based Chinese natural language ask and answer method |
| CN101441636A (en) * | 2007-11-21 | 2009-05-27 | 中国科学院自动化研究所 | Hospital information search engine and system based on knowledge base |
| WO2010150910A1 (en) * | 2009-06-26 | 2010-12-29 | 楽天株式会社 | Information search device, information search method, information search program, and storage medium on which information search program has been stored |
-
2012
- 2012-04-25 CN CN2012101251578A patent/CN102663129A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2002075588A1 (en) * | 2001-03-21 | 2002-09-26 | Inter China Network Software Company Limited | System and method of personal and business web cards |
| CN1492367A (en) * | 2002-09-27 | 2004-04-28 | 株式会社东芝 | Inquiry-response system and inquiry-response method |
| CN1928864A (en) * | 2006-09-22 | 2007-03-14 | 浙江大学 | FAQ based Chinese natural language ask and answer method |
| CN101441636A (en) * | 2007-11-21 | 2009-05-27 | 中国科学院自动化研究所 | Hospital information search engine and system based on knowledge base |
| WO2010150910A1 (en) * | 2009-06-26 | 2010-12-29 | 楽天株式会社 | Information search device, information search method, information search program, and storage medium on which information search program has been stored |
Cited By (80)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110111882B (en) * | 2011-10-24 | 2024-03-15 | 哈佛大学校长及研究员协会 | Enhanced diagnosis of conditions without compromising accuracy through artificial intelligence and mobile health techniques |
| CN110111882A (en) * | 2011-10-24 | 2019-08-09 | 哈佛大学校长及研究员协会 | Enhancing diagnosis is carried out to illness by artificial intelligence and mobile health approach, in the case where not damaging accuracy |
| CN103235787A (en) * | 2013-03-28 | 2013-08-07 | 北京百度网讯科技有限公司 | Method and equipment used for providing target information on target object |
| CN103955449A (en) * | 2014-04-21 | 2014-07-30 | 安一恒通（北京）科技有限公司 | Target sample positioning method and device |
| US10944699B2 (en) | 2014-06-10 | 2021-03-09 | Netease, Inc. | Method and device for questioning and answering |
| WO2015188425A1 (en) * | 2014-06-10 | 2015-12-17 | 网易公司 | Method and device for questioning and answering |
| CN106663125B (en) * | 2014-08-21 | 2020-06-30 | 国立研究开发法人情报通信研究机构 | Question sentence generating device and recording medium |
| CN106663125A (en) * | 2014-08-21 | 2017-05-10 | 国立研究开发法人情报通信研究机构 | Question sentence generation device and computer program |
| CN104199987A (en) * | 2014-09-28 | 2014-12-10 | 北京奇虎科技有限公司 | Method for displaying search result pages according to user-defined ordering, and search server |
| CN104239573A (en) * | 2014-10-08 | 2014-12-24 | 无锡慧方科技有限公司 | Medical system with novel open and integrated type search engine |
| CN104504163A (en) * | 2015-01-21 | 2015-04-08 | 北京智富者机器人科技有限公司 | Robot vision knowledge retrieval system |
| CN106294341A (en) * | 2015-05-12 | 2017-01-04 | 阿里巴巴集团控股有限公司 | A kind of Intelligent Answer System and theme method of discrimination thereof and device |
| US12402840B2 (en) | 2015-08-11 | 2025-09-02 | Cognoa, Inc. | Methods and apparatus to determine developmental progress with artificial intelligence and user input |
| CN105471712A (en) * | 2015-11-25 | 2016-04-06 | 深圳狗尾草智能科技有限公司 | Robot reply system and reply method thereof |
| US11972336B2 (en) | 2015-12-18 | 2024-04-30 | Cognoa, Inc. | Machine learning platform and system for data analysis |
| CN105653840A (en) * | 2015-12-21 | 2016-06-08 | 青岛中科慧康科技有限公司 | Similar case recommendation system based on word and phrase distributed representation, and corresponding method |
| CN105653840B (en) * | 2015-12-21 | 2019-01-04 | 青岛中科慧康科技有限公司 | The similar case recommender system and corresponding method shown based on words and phrases distribution table |
| CN106919577A (en) * | 2015-12-24 | 2017-07-04 | 北京奇虎科技有限公司 | Based on method, device and search engine that search word scans for recommending |
| CN105786977B (en) * | 2016-02-05 | 2020-03-03 | 北京百度网讯科技有限公司 | Mobile search method and device based on artificial intelligence |
| CN105786977A (en) * | 2016-02-05 | 2016-07-20 | 北京百度网讯科技有限公司 | Mobile search method and device based on artificial intelligence |
| CN105701254B (en) * | 2016-03-09 | 2020-11-13 | 北京搜狗科技发展有限公司 | Information processing method and device for information processing |
| CN107180037A (en) * | 2016-03-09 | 2017-09-19 | 北京京东尚科信息技术有限公司 | Man-machine interaction method and device |
| CN105701254A (en) * | 2016-03-09 | 2016-06-22 | 北京搜狗科技发展有限公司 | Information processing method and device and device for processing information |
| CN107451388A (en) * | 2016-05-31 | 2017-12-08 | 百度（美国）有限责任公司 | For the methods, devices and systems for automating medical diagnosis |
| CN106295168A (en) * | 2016-08-08 | 2017-01-04 | 安徽易联众信息技术有限公司 | Medical information inquiry system based on speech recognition |
| CN106682387A (en) * | 2016-10-26 | 2017-05-17 | 百度国际科技（深圳）有限公司 | Method and device used for outputting information |
| CN106530159A (en) * | 2016-11-09 | 2017-03-22 | 平顶山学院 | Simulated teaching method for clinical medicine |
| US12205725B2 (en) | 2016-11-14 | 2025-01-21 | Cognoa, Inc. | Methods and apparatus for evaluating developmental conditions and providing control over coverage and reliability |
| CN110168544A (en) * | 2016-12-27 | 2019-08-23 | 夏普株式会社 | Response device, response device control method, and control program |
| CN106844686A (en) * | 2017-01-26 | 2017-06-13 | 武汉奇米网络科技有限公司 | Intelligent customer service question and answer robot and its implementation based on SOLR |
| CN107247868A (en) * | 2017-05-18 | 2017-10-13 | 深思考人工智能机器人科技（北京）有限公司 | A kind of artificial intelligence aids in interrogation system |
| CN107247868B (en) * | 2017-05-18 | 2020-05-12 | 深思考人工智能机器人科技（北京）有限公司 | Artificial intelligence auxiliary inquiry system |
| CN107610740A (en) * | 2017-07-27 | 2018-01-19 | 康美健康云服务有限公司 | Semantic analysis, electronic equipment, storage medium and system for medical treatment |
| CN107609017A (en) * | 2017-08-04 | 2018-01-19 | 陈剑辉 | The method and system of medical industry intelligent search consulting are realized by self-defined hot word |
| CN107491655B (en) * | 2017-08-31 | 2020-08-25 | 上海柯棣健康管理咨询有限公司 | Liver disease information intelligent consultation system based on machine learning |
| CN107491655A (en) * | 2017-08-31 | 2017-12-19 | 康安健康管理咨询（常熟）有限公司 | Liver diseases information intelligent consultation method and system based on machine learning |
| CN110019698A (en) * | 2017-09-04 | 2019-07-16 | 珠海健康云科技有限公司 | A kind of intelligent Service method and system of medicine question and answer |
| CN107766536A (en) * | 2017-10-30 | 2018-03-06 | 江西博瑞彤芸科技有限公司 | The searching method of related information |
| CN109726269A (en) * | 2017-10-30 | 2019-05-07 | 松下知识产权经营株式会社 | Information processing method, information processing device, and program |
| CN109726269B (en) * | 2017-10-30 | 2024-04-16 | 松下知识产权经营株式会社 | Information processing method, information processing device, and program |
| CN107993724B (en) * | 2017-11-09 | 2020-11-13 | 易保互联医疗信息科技（北京）有限公司 | Medical intelligent question and answer data processing method and device |
| CN107993724A (en) * | 2017-11-09 | 2018-05-04 | 易保互联医疗信息科技（北京）有限公司 | A kind of method and device of medicine intelligent answer data processing |
| CN109885823A (en) * | 2017-12-01 | 2019-06-14 | 武汉楚鼎信息技术有限公司 | A kind of distributed semantic recognition methods of financial industry and system and device |
| CN108108449A (en) * | 2017-12-27 | 2018-06-01 | 哈尔滨福满科技有限责任公司 | A kind of implementation method based on multi-source heterogeneous data question answering system and the system towards medical field |
| CN108595696A (en) * | 2018-05-09 | 2018-09-28 | 长沙学院 | A kind of human-computer interaction intelligent answering method and system based on cloud platform |
| WO2019218527A1 (en) * | 2018-05-14 | 2019-11-21 | 平安科技（深圳）有限公司 | Multi-system combined natural language processing method and apparatus |
| CN108920488A (en) * | 2018-05-14 | 2018-11-30 | 平安科技（深圳）有限公司 | The natural language processing method and device that multisystem combines |
| CN108920603A (en) * | 2018-06-28 | 2018-11-30 | 厦门快商通信息技术有限公司 | A kind of customer service bootstrap technique based on customer service machine mould |
| CN108920603B (en) * | 2018-06-28 | 2021-12-21 | 厦门快商通信息技术有限公司 | Customer service guiding method based on customer service machine model |
| CN109086272B (en) * | 2018-08-01 | 2023-02-17 | 浙江蓝鸽科技有限公司 | Sentence pattern recognition method and system |
| CN109086272A (en) * | 2018-08-01 | 2018-12-25 | 浙江蓝鸽科技有限公司 | Sentence pattern recognition methods and its system |
| CN109036554B (en) * | 2018-08-07 | 2020-02-14 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating information |
| CN109036554A (en) * | 2018-08-07 | 2018-12-18 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating information |
| CN108932349A (en) * | 2018-08-17 | 2018-12-04 | 齐鲁工业大学 | Medical automatic question-answering method and device, storage medium, electronic equipment |
| CN110874403A (en) * | 2018-08-29 | 2020-03-10 | 株式会社日立制作所 | Question answering system, question answering processing method and question answering integrated system |
| CN110874403B (en) * | 2018-08-29 | 2024-03-08 | 株式会社日立制作所 | Question and answer system, question and answer processing method, and question and answer integrated system |
| CN109448857B (en) * | 2018-10-17 | 2021-10-26 | 南京理工大学 | Medical consultation information aggregation analysis method |
| CN109448857A (en) * | 2018-10-17 | 2019-03-08 | 南京理工大学 | Medical advice information fusion analysis method |
| CN109740928A (en) * | 2018-12-29 | 2019-05-10 | 北京中电普华信息技术有限公司 | Power consulting business processing method and device |
| CN109857847A (en) * | 2019-01-15 | 2019-06-07 | 北京搜狗科技发展有限公司 | A kind of data processing method, device and the device for data processing |
| WO2020244534A1 (en) * | 2019-06-05 | 2020-12-10 | 京东方科技集团股份有限公司 | Medical questioning and answering method, medical questioning and answering system, electronic device and computer-readable storage medium |
| US11568997B2 (en) | 2019-07-23 | 2023-01-31 | International Business Machines Corporation | Dynamic context-based collaborative medical concept interpreter |
| CN114303141A (en) * | 2019-10-01 | 2022-04-08 | 杰富意钢铁株式会社 | Information retrieval system |
| CN110727781B (en) * | 2019-10-21 | 2022-11-01 | 国网江苏省电力有限公司电力科学研究院 | Power multi-source knowledge retrieval result fusion method and device |
| CN110727781A (en) * | 2019-10-21 | 2020-01-24 | 国网江苏省电力有限公司电力科学研究院 | Power multi-source knowledge retrieval result fusion method and device |
| CN111125150A (en) * | 2019-12-26 | 2020-05-08 | 成都航天科工大数据研究院有限公司 | Industrial field question-answering system retrieval method |
| CN111125150B (en) * | 2019-12-26 | 2023-12-26 | 成都航天科工大数据研究院有限公司 | Search method for industrial field question-answering system |
| CN112035627B (en) * | 2020-07-27 | 2023-11-17 | 深圳技术大学 | Automatic question and answer method, device, equipment and storage medium |
| CN112035627A (en) * | 2020-07-27 | 2020-12-04 | 深圳技术大学 | Automatic question answering method, device, equipment and storage medium |
| CN112052317A (en) * | 2020-08-19 | 2020-12-08 | 杭州古珀医疗科技有限公司 | Medical knowledge base intelligent retrieval system and method based on deep learning |
| CN112507179A (en) * | 2020-12-11 | 2021-03-16 | 杭州依图医疗技术有限公司 | Medical data processing method and retrieval method, device and storage medium |
| CN113609830A (en) * | 2021-04-07 | 2021-11-05 | 新大陆数字技术股份有限公司 | Literature question-answering method, system and storage medium based on NLP technology |
| CN114548100A (en) * | 2022-03-01 | 2022-05-27 | 深圳市医未医疗科技有限公司 | Clinical scientific research auxiliary method and system based on big data technology |
| CN114820040A (en) * | 2022-03-31 | 2022-07-29 | 江苏未来智慧信息科技有限公司 | Electric power intelligent consultation control system and control mode thereof |
| CN117133448A (en) * | 2023-10-23 | 2023-11-28 | 长春中医药大学 | A gynecological disease consultation system based on big data |
| CN117133448B (en) * | 2023-10-23 | 2024-01-23 | 长春中医药大学 | A gynecological disease consultation system based on big data |
| CN118643136B (en) * | 2024-06-17 | 2024-11-22 | 北京法伯宏业科技发展有限公司 | An intelligent question answering system for multi-source hospital database |
| CN118643136A (en) * | 2024-06-17 | 2024-09-13 | 北京法伯宏业科技发展有限公司 | An intelligent question answering system for multi-source hospital database |
| CN118689967A (en) * | 2024-08-23 | 2024-09-24 | 中科厚立信息技术(成都)有限公司 | Question query method, device, electronic device and computer readable storage medium |
| CN118689967B (en) * | 2024-08-23 | 2024-11-08 | 中科厚立信息技术(成都)有限公司 | Medical data collection method based on medical question-answering large model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102663129A (en) | Medical field deep question and answer method and medical retrieval system | |
| CN111475623B (en) | Case Information Semantic Retrieval Method and Device Based on Knowledge Graph | |
| CN110502621B (en) | Question answering method, question answering device, computer equipment and storage medium | |
| CN110147436B (en) | A hybrid automatic question answering method based on educational knowledge graph and text | |
| CN110968699B (en) | Logic map construction and early warning method and device based on fact recommendation | |
| Balsmeier et al. | Machine learning and natural language processing on the patent corpus: Data, tools, and new measures | |
| CN117076653B (en) | Improved contextual learning knowledge base Q&A method based on thinking chain and visualization | |
| US11521713B2 (en) | System and method for generating clinical trial protocol design document with selection of patient and investigator | |
| CN106776711B (en) | Chinese medical knowledge map construction method based on deep learning | |
| US10713571B2 (en) | Displaying quality of question being asked a question answering system | |
| US9280535B2 (en) | Natural language querying with cascaded conditional random fields | |
| CN104216913B (en) | Question answering method, system and computer-readable medium | |
| CN112667794A (en) | Intelligent question-answer matching method and system based on twin network BERT model | |
| CN112650840A (en) | Intelligent medical question-answering processing method and system based on knowledge graph reasoning | |
| CN109271505A (en) | A kind of question answering system implementation method based on problem answers pair | |
| RU2488877C2 (en) | Identification of semantic relations in indirect speech | |
| CN110674252A (en) | High-precision semantic search system for judicial domain | |
| CN118296120A (en) | Large-scale language model retrieval enhancement generation method for multi-mode multi-scale multi-channel recall | |
| CN104699730A (en) | Identifying and displaying relationships between candidate answers | |
| CN103838789A (en) | A Calculation Method of Text Similarity | |
| CN113868387B (en) | A word2vec medical similarity question retrieval method based on improved tf-idf weighting | |
| CN112328800A (en) | System and method for automatically generating programming specification question answers | |
| US20170169355A1 (en) | Ground Truth Improvement Via Machine Learned Similar Passage Detection | |
| CN112632250A (en) | Question and answer method and system under multi-document scene | |
| CN106407316B (en) | Software question and answer recommendation method and device based on topic model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C05 | Deemed withdrawal (patent law before 1993) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20120912 |