CN110019698A - A kind of intelligent Service method and system of medicine question and answer - Google Patents
A kind of intelligent Service method and system of medicine question and answer Download PDFInfo
- Publication number
- CN110019698A CN110019698A CN201710785947.1A CN201710785947A CN110019698A CN 110019698 A CN110019698 A CN 110019698A CN 201710785947 A CN201710785947 A CN 201710785947A CN 110019698 A CN110019698 A CN 110019698A
- Authority
- CN
- China
- Prior art keywords
- answer
- customer problem
- candidate answers
- inquiry
- user
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Human Computer Interaction (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of intelligent Service method and system of medicine question and answer.Method includes: to establish medical domain database；Obtain customer problem；Natural language processing is carried out to the customer problem, obtaining indicates about the inquiry of the customer problem；The relevant documentation in searching database is indicated according to the inquiry, obtains search file collection；Based on the search file collection, candidate answers are extracted using cosine similarity calculation method；Based on the candidate answers, model answer is obtained using correlation similarity calculating method；By the answer feedback to user.Method and system of the invention, it is concentrated using cosine similarity calculation method in the search file retrieved according to customer problem and extracts candidate answers group, using correlation similarity calculating method, the selection standard answer in candidate answers group, and model answer is fed back into user, retrieval content is selected, judged and summarized to obtain the model answer of customer problem without user.
Description
Technical field
The present invention relates to medical field, in particular to a kind of intelligent Service method and system of medicine question and answer.
Background technique
Existing medical assistant's system is limited only to that more medicine or disease can not be obtained with the communication of doctor perhaps patient
Relevant information can only also feed back to user even if can be realized the retrieval of the medical problem proposed to user to a certain extent
Search result also needs user oneself to select search result, judge and summarize, can not to obtain the answer of proposition problem
Realize automatically generating and feeding back for the model answer of customer problem.
Summary of the invention
The object of the present invention is to which automatically generating and feed back for the model answer for realizing customer problem, provides a kind of doctor
The intelligent Service method and system that knowledge is answered.
To achieve the above object, the present invention provides following schemes:
A kind of intelligent Service method of medicine question and answer, the intelligent Service method include the following steps:
Medical domain database is established, the document about multi-class problem is stored in the medical domain database；
Obtain customer problem；
Natural language processing is carried out to the customer problem, obtaining indicates about the inquiry of the customer problem；
The relevant documentation in retrieval medical domain database is indicated according to the inquiry, obtains search file collection；
Based on the search file collection, candidate answers are extracted using cosine similarity calculation method；
Based on the candidate answers, model answer is obtained using correlation similarity calculating method；
The model answer is fed back into user.
Optionally, the acquisition customer problem, specifically includes:
It obtains user and inputs problem；
Judge that the user inputs whether problem is text, obtains the first judging result；
If the first judging result be it is yes, current text is as customer problem；
If the first judging result be it is no, judge user input problem whether be voice, obtain the second judging result；
If second judging result be it is yes, current speech is identified using automatic speech recognition mode, obtains conversion text
This information, the conversion text information is as customer problem；
If second judging result be it is no, user input problem be inefficiency, remind user use text or language
The mode of sound is re-entered.
Optionally, the inquiry that natural language processing is carried out to the customer problem, obtains about the customer problem
It indicates, specifically includes:
The customer problem is segmented, the entry set of the customer problem is obtained；
Entity recognition, semantic disambiguation, synonym extension process are named to each word in the entry set, obtained
Entry set that treated；
According to treated, entry set carries out intention analysis, obtains being intended to analysis result；
Analyze result judge whether customer problem is complete according to the intention, if so, in general's treated entry set
Entry be merged into inquiry indicate, if it is not, then require user supplement customer problem, the customer problem after being supplemented.
Optionally, described to be based on the search file collection, candidate answers are extracted using cosine similarity calculation method, specifically
Include:
According to the interrogative in the inquiry expression, answer type is determined；
Each document that search file is concentrated is divided into N number of part, wherein N is more than or equal to 1, calculates using cosine similarity
Method calculates the similarity that each part is indicated with inquiry, chooses the part that similarity is greater than the first setting value, forms the document
Text snippet；The text snippet for each document that search file is concentrated forms abstract group；
Occur in the sentence in the abstract group according to the noun phrase in the answer type and the abstract group
Order extracts noun phrase, forms candidate answers.
Optionally, described to be based on the candidate answers, model answer is obtained using correlation similarity calculating method, specifically
Include:
The number that each noun phrase in the candidate answers occurs in the medical domain database is calculated, is obtained
The correlation of each noun phrase in the candidate answers；
The number that the noun phrase in the inquiry expression occurs in the medical domain database is calculated, is obtained in institute
State the correlation that inquiry indicates；
Calculate the similarity of the correlation and the correlation of the inquiry expression of the candidate answers；
The candidate answers that similarity is greater than the second setting value are chosen, the model answer of the customer problem is combined into.
Optionally, the answer feedback is specifically included to user:
Whether the answer mode for judging user setting is speech pattern, if so, the model answer of text formatting is converted
At phonetic matrix, if it is not, then keeping text formatting；
The model answer of phonetic matrix or text formatting is exported to user.
A kind of intelligent service system of medicine question and answer, comprising:
Database module, for establishing medical domain database, be stored in the medical domain database about
The document of multi-class problem；
Customer problem obtains module, for obtaining customer problem；
Customer problem processing module is obtained for carrying out natural language processing to the customer problem about the user
The inquiry of problem indicates；
File retrieval submodule is obtained for indicating the relevant documentation in retrieval medical domain database according to the inquiry
Obtain search file collection；
Candidate answers extraction module is extracted using cosine similarity calculation method and is waited for being based on the search file collection
Select answer；
Model answer obtains module, for being based on the candidate answers, is obtained and is marked using correlation similarity calculating method
Quasi- answer；
Answer feedback module, for the model answer to be fed back to user.
Optionally, the customer problem processing module, specifically includes:
It segments submodule and obtains the entry set of the customer problem for segmenting to the customer problem；
Entry handles submodule, is named Entity recognition to each entry in the entry set, semanteme disambiguates, same
Adopted word extension process, the entry set that obtains that treated；
It is intended to analysis submodule, entry set carries out intention analysis according to treated, obtains being intended to analysis result；
Integrality judging submodule analyzes result judges whether customer problem is complete according to the intention, if so, will place
The entry in entry set after reason is merged into inquiry and indicates, if it is not, then requiring user to supplement customer problem, after being supplemented
Customer problem.
Optionally, the candidate answers extraction module, specifically includes:
Answer type determines submodule, for the interrogative in indicating according to the inquiry, determines answer type；
Abstract extraction submodule, each document for concentrating search file are divided into N number of part, and wherein N is more than or equal to
1, the similarity that each part is indicated with inquiry is calculated using cosine similarity calculation method, similarity is chosen and is greater than the first setting
The part of value forms the text snippet of the document；The text snippet for each document that search file is concentrated forms abstract group；
Candidate answers generate submodule, for according to the noun phrase in the answer type and the abstract group described
The order occurred in sentence in abstract group extracts noun phrase, forms candidate answers.
Optionally, the model answer obtains module, specifically includes:
Candidate answers correlation calculations submodule, for calculating each noun phrase in the candidate answers in the doctor
The number occurred in field database is learned, the correlation of each noun phrase in the candidate answers group is obtained；
Inquiry indicates correlation calculations submodule, leads for calculating the noun phrase in the inquiry expression in the medicine
The number occurred in regional data base obtains the correlation indicated in the inquiry；
Similarity calculation submodule, the correlation that the correlation for calculating the candidate answers is indicated with the inquiry
Similarity；
Model answer generates submodule, and the candidate answers for being greater than the second setting value for choosing similarity are combined into described
The model answer of customer problem.
The specific embodiment provided according to the present invention, the invention discloses following technical effects:
The invention discloses a kind of intelligent Service method and system of medicine question and answer, are based on medical domain database, use
Cosine similarity calculation method is concentrated in the search file retrieved according to customer problem and extracts candidate answers group, using correlation
Property similarity calculating method, the selection standard answer in the candidate answers group, and model answer is fed back into user, using this
The corresponding model answer of customer problem directly can be fed back to user by the method for invention, be selected without user retrieval content
It selects, judge and summarizes to obtain the model answer of customer problem.
Detailed description of the invention
It in order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, below will be to institute in embodiment
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the invention
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is a kind of flow chart of the intelligent Service method of medicine question and answer provided by the invention.
Fig. 2 is a kind of structural block diagram of the intelligent service system of medicine question and answer provided by the invention.
Specific embodiment
The object of the present invention is to provide a kind of intelligent Service method and system of medicine question and answer, to realize the mark of customer problem
Quasi- answer is automatically generated and is fed back.
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing and specific real
Mode is applied to be described in further detail invention.
As shown in Figure 1, the present invention provides a kind of intelligent Service method of medicine question and answer, the intelligent Service method includes
Following steps:
Step 101, medical domain database is established, the text about multi-class problem is stored in the medical domain database
Shelves；Specifically, the artificial cost for marking corpus is excessively high in terms of semantic processes, now many is all using general conventional language
Material, content embrace a wide spectrum of ideas.But it is very low to the recognition capability of some specialized vocabularies of professional domain, so, it is done specially if brought
The question answering system in industry field, the discrimination of sentence critical entities be it is very low, be based on this, we profession doctor team
Under support, a large amount of domain knowledge is sorted out, establishes medical domain database.Such as: the general introduction of disease, symptom, the cause of disease are controlled
It treats, diet conservancy etc. belongs to medical domain knowledge of the invention.
Step 102, customer problem is obtained；
Step 103, natural language processing is carried out to the customer problem, obtains the inquiry table about the customer problem
Show；
Step 104, the relevant documentation in retrieval medical domain database is indicated according to the inquiry, obtains search file
Collection；
Step 105, it is based on the search file collection, candidate answers are extracted using cosine similarity calculation method；
Step 106, the candidate answers are based on, model answer is obtained using correlation similarity calculating method；
Step 107, the model answer is fed back into user.
Optionally, customer problem is obtained described in step 102, specifically included:
It obtains user and inputs problem；
Judge that the user inputs whether problem is text, obtains the first judging result；
If the first judging result be it is yes, current text is as customer problem；
If the first judging result be it is no, judge user input problem whether be voice, obtain the second judging result；Specifically
, port (Application Programming Interface, API) is identified using Baidu's opening voice, identification input is asked
Whether topic is voice；
If second judging result be it is yes, current speech is identified using automatic speech recognition mode, obtains conversion text
This information, the conversion text information is as customer problem；Specifically, calling automatic speech recognition (Automatic Speech
Recognition, ASR) system identification current speech；
If second judging result be it is no, user input problem be inefficiency, remind user use text or language
The mode of sound is re-entered.
Optionally, natural language processing is carried out to the customer problem described in step 103, obtained about the customer problem
Inquiry indicate, specifically include:
The customer problem is segmented, the entry set of the customer problem is obtained；
Entity recognition, semantic disambiguation, synonym extension process are named to each word in the entry set, obtained
Entry set that treated；
According to treated, entry set carries out intention analysis, obtains being intended to analysis result；
Specifically, the detailed process for being intended to analysis are as follows:
Firstly, intention analysis is carried out according to primitive rule, and such as: catch a cold and what medicine eaten? corresponding rule is " disease *
Eat * medicine "；
Secondly, being handled if rule cannot be hit by following method:
Manual sorting feature defines intention to a large amount of assemblage characteristic, carries out probability calculation according to bayes formula, generally
Rate is high to be as intended to；
Such as: does is the quotation (price) of xx operation how many? artificial mark characteristic formp is as follows
Quotation operation how operation quotation
Price operation how operation quotation
Wherein, how represents problem types, and type is divided into how, when, where, what, why, six major class of yesno.
Analyze result judge whether customer problem is complete according to the intention, if so, in general's treated entry set
Entry be merged into inquiry indicate, if it is not, then require user supplement customer problem, the customer problem after being supplemented.
Specifically, described judge the whether complete specific steps of customer problem are as follows:
Problem can correctly be classified and identify that intention is effective, then it is assumed that problem is effectively and complete, wherein Question Classification
Method is that, by classical file classification method, to Question Classification, problem category has: gynaecology, paediatrics etc., for all classes
Other suitable corpus of problem hand picking carries out model training, passes through trained model prediction problem category.
Optionally, it is based on the search file collection described in step 105, candidate is extracted using cosine similarity calculation method and is answered
Case specifically includes:
According to the interrogative in the inquiry expression, answer type is determined；
Each document that search file is concentrated is divided into N number of part, wherein N is more than or equal to 1, calculates using cosine similarity
Method calculates the similarity that each part is indicated with inquiry, chooses the part that similarity is greater than the first setting value, forms the document
Text snippet；The text snippet for each document that search file is concentrated forms abstract group；
Occur in the sentence in the abstract group according to the noun phrase in the answer type and the abstract group
Order extracts noun phrase, forms candidate answers.
Optionally, it is based on the candidate answers described in step 106, standard is obtained using correlation similarity calculating method and is answered
Case specifically includes:
The number that each noun phrase in the candidate answers occurs in the medical domain database is calculated, is obtained
The correlation of each noun phrase in the candidate answers；
The number that the noun phrase in the inquiry expression occurs in the medical domain database is calculated, is obtained in institute
State the correlation that inquiry indicates；
Calculate the similarity of the correlation and the correlation of the inquiry expression of the candidate answers；
The candidate answers that similarity is greater than the second setting value are chosen, the model answer of the customer problem is combined into.
Optionally, described in step 107 by the answer feedback to user, specifically include:
Whether the answer mode for judging user setting is speech pattern, if so, Chinese idiom is converted in the answer of text formatting
Sound format, if it is not, then keeping text formatting；
The answer of phonetic matrix or text formatting is exported to user.
As shown in Fig. 2, the present invention also provides a kind of intelligent service systems of medicine question and answer, comprising:
Database module 201 stores related for establishing medical domain database in the medical domain database
In the document of multi-class problem；
Customer problem obtains module 202, for obtaining customer problem；
Customer problem processing module 203 is obtained for carrying out natural language processing to the customer problem about the use
The inquiry of family problem indicates；
File retrieval submodule 204, for indicating the relevant documentation in retrieval medical domain database according to the inquiry,
Obtain search file collection；
Candidate answers extraction module 205 is extracted for being based on the search file collection using cosine similarity calculation method
Candidate answers；
Model answer obtains module 206, for being based on the candidate answers, is obtained using correlation similarity calculating method
Model answer；
Answer feedback module 207, for the model answer to be fed back to user.
Optionally, the customer problem obtains module 202, specifically includes:
Customer problem acquisition submodule inputs problem for obtaining user；
First judging result acquisition submodule obtains first and sentences for judging that the user inputs whether problem is text
Disconnected result；
First judging result handles submodule, if for the first judging result be it is yes, current text is as customer problem；
Second judging result acquisition submodule, if for the first judging result be it is no, judge user input problem whether
For voice, the second judging result is obtained；Specifically, identifying port (Application using Baidu's opening voice
Programming Interface, API), whether identification input problem is voice；
Second judging result handles submodule, if for second judging result be it is yes, using automatic speech recognition
Mode identifies current speech, obtains conversion text information, the conversion text information is as customer problem；Specifically, calling certainly
Dynamic speech recognition (Automatic Speech Recognition, ASR) system identification current speech；If the second judgement knot
Fruit be it is no, then user input problem be inefficiency, remind user re-entered by the way of text or voice.
Optionally, the customer problem processing module 203, specifically includes:
It segments submodule and obtains the entry set of the customer problem for segmenting to the customer problem；
Entry handles submodule, for each word in the entry set be named Entity recognition, it is semantic disambiguate,
Synonym extension process, the entry set that obtains that treated；
It is intended to analysis submodule, for entry set to carry out intention analysis according to treated, obtains being intended to analysis result；
Specifically, the detailed process for being intended to analysis are as follows:
Firstly, intention analysis is carried out according to primitive rule, and such as: catch a cold and what medicine eaten? corresponding rule is " disease *
Eat * medicine "；
Secondly, being handled if rule cannot be hit by following method:
Manual sorting feature defines intention to a large amount of assemblage characteristic, carries out probability calculation according to bayes formula, generally
Rate is high to be as intended to；
Such as: does is the quotation (price) of xx operation how many? artificial mark characteristic formp is as follows
Quotation operation how operation quotation
Price operation how operation quotation
Wherein, how represents problem types, and type is divided into how, when, where, what, why, six major class of yesno.
Problem integrality judging submodule, for analyzing result judges whether customer problem is complete according to the intention, if
It is that the entry in treated entry set, which is then merged into inquiry, indicates, if it is not, user is then required to supplement customer problem, obtains
Customer problem after must supplementing.
Specifically, described judge the whether complete specific steps of customer problem are as follows:
Problem can correctly be classified and identify that intention is effective, then it is assumed that problem is effectively and complete, wherein Question Classification
Method is that, by classical file classification method, to Question Classification, problem category has: gynaecology, paediatrics etc., for all classes
Other suitable corpus of problem hand picking carries out model training, passes through trained model prediction problem category.
Optionally, the candidate answers extraction module 205, specifically includes:
Answer type determines submodule, for the interrogative in indicating according to the inquiry, determines answer type；
Abstract extraction submodule, each document for concentrating search file are divided into N number of part, and wherein N is more than or equal to
1, the similarity that each part is indicated with inquiry is calculated using cosine similarity calculation method, similarity is chosen and is greater than the first setting
The part of value forms the text snippet of the document；The text snippet for each document that search file is concentrated forms abstract group；
Candidate answers generate submodule, for according to the noun phrase in the answer type and the abstract group described
The order occurred in sentence in abstract group extracts noun phrase, forms candidate answers.
Optionally, the model answer obtains module 206, specifically includes:
Candidate answers correlation calculations submodule, for calculating each noun phrase in the candidate answers in the doctor
The number occurred in field database is learned, the correlation of each noun phrase in the candidate answers group is obtained；
Inquiry indicates correlation calculations submodule, leads for calculating the noun phrase in the inquiry expression in the medicine
The number occurred in regional data base obtains the correlation indicated in the inquiry；
Similarity calculation submodule, the correlation that the correlation for calculating the candidate answers is indicated with the inquiry
Similarity；
Model answer generates submodule, and the candidate answers for being greater than the second setting value for choosing similarity are combined into described
The model answer of customer problem.
Optionally, the answer feedback module 207, specifically includes:
Format transform subblock, for judging whether the answer mode of user setting is speech pattern, if so, by text
The model answer of format is converted into phonetic matrix, if it is not, then keeping text formatting；
Answer feedback submodule, for exporting the model answer of phonetic matrix or text formatting to user.
The specific embodiment provided according to the present invention, the invention discloses following technical effects:
The invention discloses a kind of intelligent Service method and system of medicine question and answer, are based on medical domain Database Systems,
It is concentrated using cosine similarity calculation method in the search file retrieved according to customer problem and extracts candidate answers group, used
Correlation similarity calculating method, the selection standard answer in the candidate answers group, and model answer is fed back into user, it adopts
The corresponding model answer of customer problem directly can be fed back into user with method of the invention, without user to retrieval content into
Row selection judges and summarizes to obtain the model answer of customer problem.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other
The difference of embodiment, the same or similar parts in each embodiment may refer to each other.For system disclosed in embodiment
For, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is said referring to method part
It is bright.
Specific examples are used herein to describe the principles and implementation manners of the present invention, the explanation of above embodiments
Method and its core concept of the invention are merely used to help understand, described embodiment is only that a part of the invention is real
Example is applied, instead of all the embodiments, based on the embodiments of the present invention, those of ordinary skill in the art are not making creation
Property labour under the premise of every other embodiment obtained, shall fall within the protection scope of the present invention.
Claims (10)
1. a kind of intelligent Service method of medicine question and answer, which is characterized in that the intelligent Service method includes the following steps:
Medical domain database is established, the document about multi-class problem is stored in the medical domain database；
Obtain customer problem；
Natural language processing is carried out to the customer problem, obtaining indicates about the inquiry of the customer problem；
The relevant documentation in retrieval medical domain database is indicated according to the inquiry, obtains search file collection；
Based on the search file collection, candidate answers are extracted using cosine similarity calculation method；
Based on the candidate answers, model answer is obtained using correlation similarity calculating method；
The model answer is fed back into user.
2. the intelligent Service method of medicine question and answer according to claim 1, which is characterized in that the acquisition customer problem,
It specifically includes:
It obtains user and inputs problem；
Judge that the user inputs whether problem is text, obtains the first judging result；
If the first judging result be it is yes, current text is as customer problem；
If the first judging result be it is no, judge user input problem whether be voice, obtain the second judging result；
If second judging result be it is yes, current speech is identified using automatic speech recognition mode, obtains conversion text envelope
Breath, the conversion text information is as customer problem；
If second judging result be it is no, it is inefficiency that user, which inputs problem, reminds user using text or voice
Mode is re-entered.
3. the intelligent Service method of medicine question and answer according to claim 1, which is characterized in that described to the customer problem
Natural language processing is carried out, obtaining indicates about the inquiry of the customer problem, it specifically includes:
The customer problem is segmented, the entry set of the customer problem is obtained；
Entity recognition, semantic disambiguation, synonym extension process are named to each word in the entry set, handled
Entry set afterwards；
According to treated, entry set carries out intention analysis, obtains being intended to analysis result；
Analyze result judges whether customer problem is complete according to the intention, if so, by the word in treated entry set
Item is merged into inquiry and indicates, if it is not, user is then required to supplement customer problem, the customer problem after being supplemented.
4. the intelligent Service method of medicine question and answer according to claim 1, which is characterized in that described based on the retrieval text
Shelves collection extracts candidate answers using cosine similarity calculation method, specifically includes:
According to the interrogative in the inquiry expression, answer type is determined；
Each document that search file is concentrated is divided into N number of part, wherein N is more than or equal to 1, using cosine similarity calculation method
The similarity that each part is indicated with inquiry is calculated, the part that similarity is greater than the first setting value is chosen, forms the text of the document
This abstract；The text snippet for each document that search file is concentrated forms abstract group；
The order occurred in the sentence in the abstract group according to the noun phrase in the answer type and the abstract group
Noun phrase is extracted, candidate answers are formed.
5. the intelligent Service method of medicine question and answer according to claim 1, which is characterized in that described to be answered based on the candidate
Case obtains model answer using correlation similarity calculating method, specifically includes:
The number that each noun phrase in the candidate answers occurs in the medical domain database is calculated, is obtained in institute
State the correlation of each noun phrase in candidate answers；
The number that the noun phrase in the inquiry expression occurs in the medical domain database is calculated, obtains looking into described
Ask the correlation indicated；
Calculate the similarity of the correlation and the correlation of the inquiry expression of the candidate answers；
The candidate answers that similarity is greater than the second setting value are chosen, the model answer of the customer problem is combined into.
6. the intelligent Service method of medicine question and answer according to claim 1, which is characterized in that by the answer feedback to use
Family specifically includes:
Whether the answer mode for judging user setting is speech pattern, if so, Chinese idiom is converted in the model answer of text formatting
Sound format, if it is not, then keeping text formatting；
The model answer of phonetic matrix or text formatting is exported to user.
7. a kind of intelligent service system of medicine question and answer, which is characterized in that the intelligent service system includes:
Database module is stored in the medical domain database about multiclass for establishing medical domain database
The document of problem；
Customer problem obtains module, for obtaining customer problem；
Customer problem processing module is obtained for carrying out natural language processing to the customer problem about the customer problem
Inquiry indicate；
Document retrieval module is retrieved for indicating the relevant documentation in retrieval medical domain database according to the inquiry
Document sets；
Candidate answers extraction module extracts candidate using cosine similarity calculation method and answers for being based on the search file collection
Case；
Model answer obtains module, for being based on the candidate answers, obtains standard using correlation similarity calculating method and answers
Case；
Answer feedback module, for the model answer to be fed back to user.
8. the intelligent service system of medicine question and answer according to claim 7, which is characterized in that the customer problem handles mould
Block specifically includes:
It segments submodule and obtains the entry set of the customer problem for segmenting to the customer problem；
Entry handles submodule, is named Entity recognition, semantic disambiguation, synonym to each entry in the entry set
Extension process, the entry set that obtains that treated；
It is intended to analysis submodule, entry set carries out intention analysis according to treated, obtains being intended to analysis result；
Integrality judging submodule analyzes result judges whether customer problem is complete according to the intention, if so, after handling
Entry set in entry be merged into inquiry indicate, if it is not, then require user supplement customer problem, the user after being supplemented
Problem.
9. the intelligent service system of medicine question and answer according to claim 7, which is characterized in that the candidate answers extract mould
Block specifically includes:
Answer type determines submodule, for the interrogative in indicating according to the inquiry, determines answer type；
Abstract extraction submodule, each document for concentrating search file are divided into N number of part, and wherein N is more than or equal to 1, adopt
The similarity that each part is indicated with inquiry is calculated with cosine similarity calculation method, similarity is chosen and is greater than the first setting value
Part forms the text snippet of the document；The text snippet for each document that search file is concentrated forms abstract group；
Candidate answers generate submodule, for according to the noun phrase in the answer type and the abstract group in the abstract
The order occurred in sentence in group extracts noun phrase, forms candidate answers.
10. the intelligent service system of medicine question and answer according to claim 7, which is characterized in that the model answer obtains
Module specifically includes:
Candidate answers correlation calculations submodule is led for calculating each noun phrase in the candidate answers in the medicine
The number occurred in regional data base obtains the correlation of each noun phrase in the candidate answers group；
Inquiry indicates correlation calculations submodule, for calculating the noun phrase in the inquiry expression in the medical domain number
According to the number occurred in library, the correlation indicated in the inquiry is obtained；
Similarity calculation submodule, the correlation for calculating the candidate answers are similar with the correlation of the inquiry expression
Degree；
Model answer generates submodule, and the candidate answers for being greater than the second setting value for choosing similarity are combined into the user
The model answer of problem.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710785947.1A CN110019698A (en) | 2017-09-04 | 2017-09-04 | A kind of intelligent Service method and system of medicine question and answer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710785947.1A CN110019698A (en) | 2017-09-04 | 2017-09-04 | A kind of intelligent Service method and system of medicine question and answer |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110019698A true CN110019698A (en) | 2019-07-16 |
Family
ID=67186196
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710785947.1A Pending CN110019698A (en) | 2017-09-04 | 2017-09-04 | A kind of intelligent Service method and system of medicine question and answer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110019698A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110674272A (en) * | 2019-09-05 | 2020-01-10 | 科大讯飞股份有限公司 | Question answer determining method and related device |
| CN111985246A (en) * | 2020-08-27 | 2020-11-24 | 武汉东湖大数据交易中心股份有限公司 | Disease cognitive system based on main symptoms and accompanying symptom words |
| CN112380329A (en) * | 2020-11-13 | 2021-02-19 | 四川大学 | Training robot system and method under fine positive symptom background |
| CN112380330A (en) * | 2020-11-13 | 2021-02-19 | 四川大学 | Training robot system and method under background of fine yin syndrome |
| CN112380231A (en) * | 2020-11-13 | 2021-02-19 | 四川大学 | Training robot system and method with depressive disorder characteristics |
| CN112395399A (en) * | 2020-11-13 | 2021-02-23 | 四川大学 | Specific personality dialogue robot training method based on artificial intelligence |
| CN113868406A (en) * | 2021-12-01 | 2021-12-31 | 无码科技（杭州）有限公司 | Search method, search system, and computer-readable storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663129A (en) * | 2012-04-25 | 2012-09-12 | 中国科学院计算技术研究所 | Medical field deep question and answer method and medical retrieval system |
| US20140244565A1 (en) * | 2013-02-28 | 2014-08-28 | The Procter & Gamble Company | Systems And Methods For Customized Advice Messaging |
| CN106326640A (en) * | 2016-08-12 | 2017-01-11 | 上海交通大学医学院附属瑞金医院卢湾分院 | Medical speech control system and control method thereof |
-
2017
- 2017-09-04 CN CN201710785947.1A patent/CN110019698A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663129A (en) * | 2012-04-25 | 2012-09-12 | 中国科学院计算技术研究所 | Medical field deep question and answer method and medical retrieval system |
| US20140244565A1 (en) * | 2013-02-28 | 2014-08-28 | The Procter & Gamble Company | Systems And Methods For Customized Advice Messaging |
| CN106326640A (en) * | 2016-08-12 | 2017-01-11 | 上海交通大学医学院附属瑞金医院卢湾分院 | Medical speech control system and control method thereof |
Non-Patent Citations (1)
| Title |
|---|
| 刘红霞: ""面向慢性病海量数据问答系统智能摘要算法的研究与实现"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110674272A (en) * | 2019-09-05 | 2020-01-10 | 科大讯飞股份有限公司 | Question answer determining method and related device |
| CN110674272B (en) * | 2019-09-05 | 2022-12-06 | 科大讯飞股份有限公司 | Question answer determining method and related device |
| CN111985246A (en) * | 2020-08-27 | 2020-11-24 | 武汉东湖大数据交易中心股份有限公司 | Disease cognitive system based on main symptoms and accompanying symptom words |
| CN111985246B (en) * | 2020-08-27 | 2023-08-15 | 武汉东湖大数据交易中心股份有限公司 | Disease cognitive system based on main symptoms and accompanying symptom words |
| CN112380329A (en) * | 2020-11-13 | 2021-02-19 | 四川大学 | Training robot system and method under fine positive symptom background |
| CN112380330A (en) * | 2020-11-13 | 2021-02-19 | 四川大学 | Training robot system and method under background of fine yin syndrome |
| CN112380231A (en) * | 2020-11-13 | 2021-02-19 | 四川大学 | Training robot system and method with depressive disorder characteristics |
| CN112395399A (en) * | 2020-11-13 | 2021-02-23 | 四川大学 | Specific personality dialogue robot training method based on artificial intelligence |
| CN113868406A (en) * | 2021-12-01 | 2021-12-31 | 无码科技（杭州）有限公司 | Search method, search system, and computer-readable storage medium |
| CN113868406B (en) * | 2021-12-01 | 2022-03-11 | 无码科技（杭州）有限公司 | Search method, search system, and computer-readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110019698A (en) | A kind of intelligent Service method and system of medicine question and answer | |
| CN107480122B (en) | Artificial intelligence interaction method and artificial intelligence interaction device | |
| CN112231447B (en) | Method and system for extracting Chinese document events | |
| CN109508459B (en) | Method for extracting theme and key information from news | |
| CN110597952A (en) | Information processing method, server, and computer storage medium | |
| CN112201228A (en) | Multimode semantic recognition service access method based on artificial intelligence | |
| US20220254507A1 (en) | Knowledge graph-based question answering method, computer device, and medium | |
| CN110895559B (en) | Model training method, text processing method, device and equipment | |
| CN108388553B (en) | Method for eliminating ambiguity in conversation, electronic equipment and kitchen-oriented conversation system | |
| CN110164447B (en) | Spoken language scoring method and device | |
| CN113505586A (en) | Seat-assisted question-answering method and system integrating semantic classification and knowledge graph | |
| CN113821605B (en) | Event extraction method | |
| CN112699686B (en) | Semantic understanding method, device, equipment and medium based on task type dialogue system | |
| CN105868179A (en) | Intelligent asking-answering method and device | |
| CN110210036A (en) | A kind of intension recognizing method and device | |
| CN115292461B (en) | Man-machine interaction learning method and system based on voice recognition | |
| CN116822517B (en) | Multi-language translation term identification method | |
| CN111353026A (en) | Intelligent law attorney assistant customer service system | |
| CN109992651B (en) | Automatic identification and extraction method for problem target features | |
| CN109065015B (en) | Data acquisition method, device and equipment and readable storage medium | |
| CN107562907B (en) | Intelligent lawyer expert case response device | |
| CN117272977A (en) | Character description sentence recognition method and device, electronic equipment and storage medium | |
| CN114637852B (en) | Entity relation extraction method, device, equipment and storage medium of medical text | |
| CN114880994B (en) | Text style conversion method and device from direct white text to irony text | |
| CN111680493B (en) | English text analysis method and device, readable storage medium and computer equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |