CN108920603A - A kind of customer service bootstrap technique based on customer service machine mould - Google Patents
A kind of customer service bootstrap technique based on customer service machine mould Download PDFInfo
- Publication number
- CN108920603A CN108920603A CN201810684550.8A CN201810684550A CN108920603A CN 108920603 A CN108920603 A CN 108920603A CN 201810684550 A CN201810684550 A CN 201810684550A CN 108920603 A CN108920603 A CN 108920603A
- Authority
- CN
- China
- Prior art keywords
- customer service
- demand
- client
- product classification
- machine mould
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H80/00—ICT specially adapted for facilitating communication between medical practitioners or patients, e.g. for collaborative diagnosis, therapy or health monitoring
Abstract
The present invention provides a kind of customer service bootstrap techniques based on customer service machine mould, and the related phrases in a field are inputted training pattern, obtain corresponding robot and reply；Field chat corpus is analyzed, data cleansing is carried out according to the rule that analysis obtains, invalid data is filtered, deletes redundancy；With the entitled label of the product classification in the field, to every a pair of of question and answer to marking；Question sentences all in corpus and corresponding answer are done and segmented respectively, the semantic feature of each word is learnt；The slot position for guiding client filling associated according to every demand of the required product classification of client and product classification, it is mapped according to client demand and product classification, it formulates client-side issue and replys rule, obtain every demand of the required product classification of client and product classification, the associated slot position of guidance client filling.Due to joined name of product classification and personal information state recording, so that entire dialogue is unlikely to sideslip, improve service quality.
Description
Technical field
The present invention relates to a kind of customer service bootstrap techniques based on customer service machine mould, are related to intelligent customer service field.
Background technique
Intelligent robot common at present includes chat robots and FAQ question answering system.Chat robots are one and are used to
The program for simulating human conversation or chat is carrying out interacting Question-Answer with people since the corpus field that chat robots are covered is too big
When, the phenomenon that can not often making more accurate answer or give an irrelevant answer；And FAQ question answering system then only from
Most similar question and answer are found in question and answer corpus and return to answer to client, can not be obtained the theme of entire session, be caused people
Machine conversation content deviates theme, i.e., when original data cover is sufficiently complete, is difficult to realize and carries out conventional guidance to user
And rules guide.
Summary of the invention
The present invention provides a kind of customer service bootstrap techniques based on customer service machine mould, to improve the accurate of robot answer
Degree, and can be when data volume is sufficiently large it can be concluded that good effect, reduces entire human-computer dialogue content offset theme
Probability, improve user experience.
A kind of customer service bootstrap technique based on customer service machine mould, specific method include,
The related phrases in one field are inputted into training pattern, corresponding robot is obtained and replys；
Field chat corpus is analyzed, data cleansing is carried out according to the rule that analysis obtains, invalid data is filtered, deletes superfluous
Remainder；
With the entitled label of the product classification in the field, to every a pair of of question and answer to marking；
Question sentences all in corpus and corresponding answer are done and segmented respectively, the semanteme for learning each word with skip-gram is special
Sign；
The average value of the sum of tf-idf weighting term vector by participle each in question sentence and/or answer splices relevant product
Sentence semantics feature of the term vector of specific name as each sentence；
Binding rule and Arithmetic of Semantic Similarity guidance client fill slot position, and specific method includes,
It is guided associated by client filling according to every demand of the required product classification of client and product classification
Slot position is mapped according to client demand and product classification, formulates client-side issue response rule and rhetorical question rule, thus
To every demand of the required product classification of client and product classification, the slot position for guiding client filling associated；
The associated slot position includes the basic contact details of user；
The items demand includes product demand, technical solution demand, time demand, demand for services, price demand, safety
Any one or a few in demand and risk demand.
The method also includes using the Average Accuracy of every section of dialogue as the evaluation index of training pattern.
The field is medical and beauty treatment fields, the entitled position of the product classification and classification of the items title.
The items demand includes project details/science popularization, Project Technical, the course for the treatment of/therapeutic scheme, price, side effect/multiple
Hair, diet, nursing, reservation/ask address/time, operative failure, in material/apparatus/product and safety any one or it is several
Kind.
The associated slot position includes name, gender, the age, symptom, position, project, examines history, the choice of technology and acquisition
Any one or a few in user's phone number.
The mapping ruler of the mapping is position->Project->Be intended to->As a result.
Compared with prior art, technical solution of the present invention by being analyzed corpus, extraction unit divider then, make robot
More accurate answer can be carried out according to the user's intention, avoids the phenomenon that giving an irrelevant answer, and improve intelligence, and in data
Available preferable effect, uses manpower and material resources sparingly when measuring sufficiently large；Simultaneously as joined name of product classification and
Personal information state recording, can reduce the probability of entire human-computer dialogue content offset theme, preferably carry out to user conventional
Guidance and rules guide, to improve service quality.
Detailed description of the invention
Fig. 1 is flow chart of the invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to embodiments, to the present invention
It is further elaborated.It should be appreciated that described herein, specific examples are only used to explain the present invention, is not used to limit
The fixed present invention.
Any feature disclosed in this specification (including abstract) unless specifically stated can or tool equivalent by other
There are the alternative features of similar purpose to be replaced.That is, unless specifically stated, each feature is a series of equivalent or similar characteristics
In an example.
As shown in Figure 1, a kind of customer service bootstrap technique based on customer service machine mould, specific method include,
The related phrases in one field are inputted into training pattern, corresponding robot is obtained and replys；
Field chat corpus is analyzed, data cleansing is carried out according to the rule that analysis obtains, invalid data is filtered, deletes superfluous
Remainder；
With the entitled label of the product classification in the field, to every a pair of of question and answer to marking；
Question sentences all in corpus and corresponding answer are done and segmented respectively, the semanteme for learning each word with skip-gram is special
Sign；
The average value of the sum of tf-idf weighting term vector by participle each in question sentence and/or answer splices relevant product
Sentence semantics feature (word2vec) of the term vector of specific name as each sentence；
Binding rule and Arithmetic of Semantic Similarity guidance client fill slot position, and specific method includes,
It is guided associated by client filling according to every demand of the required product classification of client and product classification
Slot position is mapped according to client demand and product classification, formulates client-side issue response rule and rhetorical question rule, thus
To every demand of the required product classification of client and product classification, the slot position for guiding client filling associated；
The associated slot position includes the basic contact details of user；
The items demand includes product demand, technical solution demand, time demand, demand for services, price demand, safety
Any one or a few in demand and risk demand.
In the present invention program, the kernel algorithm being related to includes FAQ, Word2vec, tf-idf, names Entity recognition and depth
Learning network is spent, in the form of inputting a short text, output and the input associated answer of sentence.
Skip-gram is a kind of model of Word2vec training, and there are two types of the word2vec methods of Google：CBOW and
Skip-gram, both are all the methods of trained term vector, and rule of thumb, CBOW will more faster, but skip-gram is imitated
Fruit wants better.Statistical language model statistical language model is exactly to give you several words, is gone out in these words
(subsequent) probability of some word appearance is calculated under the premise of existing.CBOW is also one kind of statistical language model, as the term suggests just
It is according to the C word or front and back C continuous words before some word, to calculate the probability of some word appearance.Skip-Gram
Model is on the contrary, be then to calculate separately its front and back according to some word and each probability of certain several word occur.
TF-IDF (term frequency-inverse document frequency) be it is a kind of for information retrieval with
The common weighting technique of data mining.TF means word frequency (Term Frequency), and IDF means inverse document frequency
(Inverse Document Frequency).TF-IDF is a kind of statistical method, to assess a words for a file
The significance level of collection or a copy of it file in a corpus.The number that the importance of words occurs hereof with it
Directly proportional increase, but the frequency that can occur in corpus with it simultaneously is inversely proportional decline.
The present invention program uses manpower and material resources sparingly, it can be concluded that good effect when data volume is sufficiently large.Due to adding
Name of product classification (being equivalent to conversation subject content) and personal information state recording are entered, so that entire dialogue is unlikely to run
Partially, it improves service quality.
One field related phrases (usually question sentence) input system is obtained as one embodiment of the present invention
Corresponding robot is replied, using the Average Accuracy of every section of dialogue as the evaluation index of training pattern.
As one embodiment of the present invention, the field is medical and beauty treatment fields, the entitled portion of product classification
Position and classification of the items title (such as face, taking off lip hair).
As one embodiment of the present invention, it is described items demand include project details/science popularization, Project Technical, the course for the treatment of/
Address/time, operative failure, material/apparatus/product are asked in therapeutic scheme, price, side effect/recurrence, diet, nursing, reservation/
With any one or a few in safety.
As one embodiment of the present invention, the associated slot position includes name, gender, age, symptom, portion
Position, project, examine history, the choice of technology and obtain user's phone number in any one or a few.With every demand of client
To be intended to user, guides user to fill in associated slot position according to disparity items.
As one embodiment of the present invention, classified according to every demand of user, data preparation obtains described
The mapping ruler of mapping is position->Project->Be intended to->As a result, for example：Nose->Augmentation rhinoplasty->Technology (material, product, production
Ground ...)->Price.
Rulemaking is carried out by taking following several customer problems as an example, and (slot point refers to herein：A problem is answered to be known
All associated informations in road, slot template=》【Question sentence classification, position, item types/title, symptom keyword】)：
1. client's question sentence contains all slot points (directly asking project) of project
Such as：I wants to do face scar reparation
【Question sentence classification->Ask project, position->Skin (learnt) by face, face scar reparation->Project】, according to obtaining
The available relevant technology of mapping data【It is ' cutting ' technology herein】, price, the information such as the course for the treatment of.
2. question sentence includes partial groove point (directly asking project, short slot point)
Such as:I wants to do scar reparation？
Position is not mentioned, corresponding technology, Price Range may be different
Robot needs that client is guided to say skin area【Obtain face or elsewhere skin, the scar at each position
Recovery technique and price may be different】.
For another example：I want to allow nose more very
Classification, obtains similar key：Augmentation rhinoplasty (it needs to carry out identical semantic analysis, replacement herein, it can be by arranging in advance
Synonym dictionary solve the problems, such as this, can also more be endured with the standard keyword replacement in similar semantic sentence, nose herein
It is not the professional term of standard).
Augmentation rhinoplasty relevant item has 9 kinds, (Artecoll augmentation rhinoplasty, subperiosteum augmentation rhinoplasty, Korean style augmentation rhinoplasty, augmentation rhinoplasty, prosthese augmentation rhinoplasty, augmentation rhinoplasty
Art operative failure reparation, augmentation rhinoplasty reparation, sodium hyaluronate augmentation rhinoplasty, self augmentation rhinoplasty)；
Impossible project is excluded in 9 options according to existing information；
The further suitable project of guidance user selection, (including technology introduction, equipment, product, price ...)；
Nose, very==》Augmentation rhinoplasty (entity name or Keywords matching, synonym replacement)；
【Question sentence classification, position, item types, symptom keyword】—>>【Ask project, nose, augmentation rhinoplasty, [nose, very]】.
3. including multiple positions (be intended to) more
Such as：client：How much loses hair or feathers？
Server：You want to improve the hair problem at which position？
Client：Oxter
Client：Lip
Client：Two positions will
Slot position can be filled up by project guidance (can be stored to all items bulleted list, guidance fills out slot or by visitor in order
Slot is filled out in family speech)
【Question sentence classification, position, item types, symptom keyword】—>>【Ask price, oxter, oxter depilation, [lip takes off
Hair, how much]】.
Problems Concerning Their Recurrence after treatment
Such as：Having put mole can also grow again later？It can also be regenerated after depilation？
Keyword：Mole has been put, then has been grown；Depilation, regeneration
Corresponding template is【Ask recurrence, mole is put at the position xx, [having put mole, regenerate]】.
The answer of customer problem provides strategy
Example：You lose hair or feathers technology how
【Ask that Project Technical, skin lose hair or feathers, [depilation, technology, how]】.
The problem of for being not required to exact details, can be in this way：Keyword ' depilation '【Project name】；Obtain continuous item
Purpose corpus (avoiding disposable all items categorical data from analyzing, to reduce unnecessary calculation amount)-Question sentence parsing
Matching, obtains the highest question and answer pair of similarity.
Empirically summarize, if feel this problem need answer it is more specific in detail, can use in template own
Corresponding slot point, by rule searching result：{ position { project：{ it is intended to：As a result } } }.
This method is for a small amount of data since its rule plus semantic similarity combine leading dialogue, question and answer effect meeting
It is relatively good；And a large amount of data are needed manpower and material resources is spent to go to summarize large-scale classification and rule.
Art is finally talked about according to business, guidance user reserves or leave telephone number.
Claims (6)
1. a kind of customer service bootstrap technique based on customer service machine mould, which is characterized in that specific method includes：
The related phrases in one field are inputted into training pattern, corresponding robot is obtained and replys；
Field chat corpus is analyzed, data cleansing is carried out according to the rule that analysis obtains, invalid data is filtered, deletes redundancy
?；
With the entitled label of the product classification in the field, to every a pair of of question and answer to marking；
Question sentences all in corpus and corresponding answer are done and segmented respectively, learn the semantic feature of each word with skip-gram；
The average value of the sum of tf-idf weighting term vector by participle each in question sentence and/or answer splices relevant product classification
Sentence semantics feature of the term vector of title as each sentence；
Binding rule and Arithmetic of Semantic Similarity guidance client fill slot position, and specific method includes,
Slot position associated by client filling is guided according to every demand of the required product classification of client and product classification,
It is mapped according to client demand and product classification, client-side issue response rule and rhetorical question rule is formulated, to obtain visitor
Every demand of the required product classification in family end and product classification, the associated slot position of guidance client filling；
The associated slot position includes the basic contact details of user；
The items demand includes product demand, technical solution demand, time demand, demand for services, price demand, demand for security
With any one or a few in risk demand.
2. the customer service bootstrap technique according to claim 1 based on customer service machine mould, which is characterized in that the method is also
Including：Using the Average Accuracy of every section of dialogue as the evaluation index of training pattern.
3. the customer service bootstrap technique according to claim 1 based on customer service machine mould, it is characterised in that：The field is
Medical and beauty treatment fields, the entitled position of the product classification and classification of the items title.
4. the customer service bootstrap technique according to claim 3 based on customer service machine mould, it is characterised in that：The items need
/ recurrence, diet, nursing, reservation/including project details/science popularization, Project Technical, the course for the treatment of/therapeutic scheme, price, side effect is asked to ask
Address/time, operative failure, any one or a few in material/apparatus/product and safety.
5. the customer service bootstrap technique according to claim 4 based on customer service machine mould, it is characterised in that：It is described associated
Slot position include name, gender, the age, symptom, position, project, examine history, the choice of technology and obtain user's phone number in appoint
Meaning is one or more of.
6. based on the customer service bootstrap technique of customer service machine mould according to one of claim 3 to 5, it is characterised in that：Institute
The mapping ruler for stating mapping is position->Project->Be intended to->As a result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810684550.8A CN108920603B (en) | 2018-06-28 | 2018-06-28 | Customer service guiding method based on customer service machine model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810684550.8A CN108920603B (en) | 2018-06-28 | 2018-06-28 | Customer service guiding method based on customer service machine model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108920603A true CN108920603A (en) | 2018-11-30 |
| CN108920603B CN108920603B (en) | 2021-12-21 |
Family
ID=64421745
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810684550.8A Active CN108920603B (en) | 2018-06-28 | 2018-06-28 | Customer service guiding method based on customer service machine model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108920603B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108984655A (en) * | 2018-06-28 | 2018-12-11 | 厦门快商通信息技术有限公司 | A kind of customer service robot intelligent customer service bootstrap technique |
| CN109514586A (en) * | 2019-01-30 | 2019-03-26 | 第四范式（北京）技术有限公司 | Realize the method and system of intelligent customer service robot |
| CN109767818A (en) * | 2018-12-27 | 2019-05-17 | 厦门快商通信息技术有限公司 | A kind of customization medical treatment is answerred questions interrogation guidance system |
| WO2020133470A1 (en) * | 2018-12-29 | 2020-07-02 | 深圳市优必选科技有限公司 | Chat corpus cleaning method and apparatus, computer device, and storage medium |
Citations (15)
| 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 |
| US20140236577A1 (en) * | 2013-02-15 | 2014-08-21 | Nec Laboratories America, Inc. | Semantic Representations of Rare Words in a Neural Probabilistic Language Model |
| CN105589848A (en) * | 2015-12-28 | 2016-05-18 | 百度在线网络技术（北京）有限公司 | Dialog management method and device |
| CN105824933A (en) * | 2016-03-18 | 2016-08-03 | 苏州大学 | Automatic question-answering system based on theme-rheme positions and realization method of automatic question answering system |
| CN105975478A (en) * | 2016-04-09 | 2016-09-28 | 北京交通大学 | Word vector analysis-based online article belonging event detection method and device |
| CN106156003A (en) * | 2016-06-30 | 2016-11-23 | 北京大学 | A kind of question sentence understanding method in question answering system |
| US20160342685A1 (en) * | 2015-05-22 | 2016-11-24 | Microsoft Technology Licensing, Llc | Ontology-Crowd-Relevance Deep Response Generation |
| CN106570708A (en) * | 2016-10-31 | 2017-04-19 | 厦门快商通科技股份有限公司 | Management method and management system of intelligent customer service knowledge base |
| US20170177715A1 (en) * | 2015-12-21 | 2017-06-22 | Adobe Systems Incorporated | Natural Language System Question Classifier, Semantic Representations, and Logical Form Templates |
| CN107273913A (en) * | 2017-05-11 | 2017-10-20 | 武汉理工大学 | A kind of short text similarity calculating method based on multi-feature fusion |
| CN107562863A (en) * | 2017-08-30 | 2018-01-09 | 深圳狗尾草智能科技有限公司 | Chat robots reply automatic generation method and system |
| CN107679234A (en) * | 2017-10-24 | 2018-02-09 | 上海携程国际旅行社有限公司 | Customer service information providing method, device, electronic equipment, storage medium |
| CN107704563A (en) * | 2017-09-29 | 2018-02-16 | 广州多益网络股份有限公司 | A kind of question sentence recommends method and system |
| CN107886948A (en) * | 2017-11-16 | 2018-04-06 | 百度在线网络技术（北京）有限公司 | Voice interactive method and device, terminal, server and readable storage medium storing program for executing |
| CN107958091A (en) * | 2017-12-28 | 2018-04-24 | 北京贝塔智投科技有限公司 | A kind of NLP artificial intelligence approaches and interactive system based on financial vertical knowledge mapping |
-
2018
- 2018-06-28 CN CN201810684550.8A patent/CN108920603B/en active Active
Patent Citations (15)
| 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 |
| US20140236577A1 (en) * | 2013-02-15 | 2014-08-21 | Nec Laboratories America, Inc. | Semantic Representations of Rare Words in a Neural Probabilistic Language Model |
| US20160342685A1 (en) * | 2015-05-22 | 2016-11-24 | Microsoft Technology Licensing, Llc | Ontology-Crowd-Relevance Deep Response Generation |
| US20170177715A1 (en) * | 2015-12-21 | 2017-06-22 | Adobe Systems Incorporated | Natural Language System Question Classifier, Semantic Representations, and Logical Form Templates |
| CN105589848A (en) * | 2015-12-28 | 2016-05-18 | 百度在线网络技术（北京）有限公司 | Dialog management method and device |
| CN105824933A (en) * | 2016-03-18 | 2016-08-03 | 苏州大学 | Automatic question-answering system based on theme-rheme positions and realization method of automatic question answering system |
| CN105975478A (en) * | 2016-04-09 | 2016-09-28 | 北京交通大学 | Word vector analysis-based online article belonging event detection method and device |
| CN106156003A (en) * | 2016-06-30 | 2016-11-23 | 北京大学 | A kind of question sentence understanding method in question answering system |
| CN106570708A (en) * | 2016-10-31 | 2017-04-19 | 厦门快商通科技股份有限公司 | Management method and management system of intelligent customer service knowledge base |
| CN107273913A (en) * | 2017-05-11 | 2017-10-20 | 武汉理工大学 | A kind of short text similarity calculating method based on multi-feature fusion |
| CN107562863A (en) * | 2017-08-30 | 2018-01-09 | 深圳狗尾草智能科技有限公司 | Chat robots reply automatic generation method and system |
| CN107704563A (en) * | 2017-09-29 | 2018-02-16 | 广州多益网络股份有限公司 | A kind of question sentence recommends method and system |
| CN107679234A (en) * | 2017-10-24 | 2018-02-09 | 上海携程国际旅行社有限公司 | Customer service information providing method, device, electronic equipment, storage medium |
| CN107886948A (en) * | 2017-11-16 | 2018-04-06 | 百度在线网络技术（北京）有限公司 | Voice interactive method and device, terminal, server and readable storage medium storing program for executing |
| CN107958091A (en) * | 2017-12-28 | 2018-04-24 | 北京贝塔智投科技有限公司 | A kind of NLP artificial intelligence approaches and interactive system based on financial vertical knowledge mapping |
Non-Patent Citations (5)
| Title |
|---|
| D BOGDANOVA等: "This is how we do it: Answer reranking for open-domain how questions with paragraph vectors and minimal feature engineering", 《PROCEEDINGS OF NAACL-HLT 2016》 * |
| JOO-KYUNG KIM等: "Intent detection using semantically enriched word embeddings", 《2016 IEEE SPOKEN LANGUAGE TECHNOLOGY WORKSHOP (SLT)》 * |
| 成昊: "基于Word2Vec的中文问句检索技术研究及系统实现", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 杨燕: "面向电商领域的智能问答系统若干关键技术研究", 《中国博士学位论文全文数据库信息科技辑》 * |
| 白晓雷: "基于信息抽取的语义框架填充技术研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108984655A (en) * | 2018-06-28 | 2018-12-11 | 厦门快商通信息技术有限公司 | A kind of customer service robot intelligent customer service bootstrap technique |
| CN108984655B (en) * | 2018-06-28 | 2021-01-01 | 厦门快商通信息技术有限公司 | Intelligent customer service guiding method for customer service robot |
| CN109767818A (en) * | 2018-12-27 | 2019-05-17 | 厦门快商通信息技术有限公司 | A kind of customization medical treatment is answerred questions interrogation guidance system |
| WO2020133470A1 (en) * | 2018-12-29 | 2020-07-02 | 深圳市优必选科技有限公司 | Chat corpus cleaning method and apparatus, computer device, and storage medium |
| CN109514586A (en) * | 2019-01-30 | 2019-03-26 | 第四范式（北京）技术有限公司 | Realize the method and system of intelligent customer service robot |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108920603B (en) | 2021-12-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108984655A (en) | A kind of customer service robot intelligent customer service bootstrap technique | |
| WO2020007028A1 (en) | Medical consultation data recommendation method, device, computer apparatus, and storage medium | |
| CN108920603A (en) | A kind of customer service bootstrap technique based on customer service machine mould | |
| CN110175227A (en) | A kind of dialogue auxiliary system based on form a team study and level reasoning | |
| CN110750616A (en) | Retrieval type chatting method and device and computer equipment | |
| CN110675944A (en) | Triage method and device, computer equipment and medium | |
| KR101971582B1 (en) | Method of providing health care guide using chat-bot having user intension analysis function and apparatus for the same | |
| Baur et al. | eXplainable cooperative machine learning with NOVA | |
| US20230394247A1 (en) | Human-machine collaborative conversation interaction system and method | |
| CN111309887B (en) | Method and system for training text key content extraction model | |
| CN111339284A (en) | Product intelligent matching method, device, equipment and readable storage medium | |
| CN109325780A (en) | A kind of exchange method of the intelligent customer service system in E-Governance Oriented field | |
| CN110309114A (en) | Processing method, device, storage medium and the electronic device of media information | |
| CN113569023A (en) | Chinese medicine question-answering system and method based on knowledge graph | |
| CN113724882A (en) | Method, apparatus, device and medium for constructing user portrait based on inquiry session | |
| CN113157885B (en) | Efficient intelligent question-answering system oriented to knowledge in artificial intelligence field | |
| CN115292457A (en) | Knowledge question answering method and device, computer readable medium and electronic equipment | |
| CN107562911A (en) | More wheel interaction probabilistic model training methods and auto-answer method | |
| CN113590783A (en) | Traditional Chinese medicine health-preserving intelligent question-answering system based on NLP natural language processing | |
| CN114238607A (en) | Deep interactive AI intelligent job-searching consultant method, system and storage medium | |
| CN114783421A (en) | Intelligent recommendation method and device, equipment and medium | |
| CN113010657A (en) | Answer processing method and answer recommending method based on answering text | |
| EP3901875A1 (en) | Topic modelling of short medical inquiries | |
| CN117235354A (en) | User personalized service strategy and system based on multi-mode large model | |
| CN111353290A (en) | Method and system for automatically responding to user inquiry |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |