CN109461039A - A kind of text handling method and intelligent customer service method - Google Patents
A kind of text handling method and intelligent customer service method Download PDFInfo
- Publication number
- CN109461039A CN109461039A CN201810986605.0A CN201810986605A CN109461039A CN 109461039 A CN109461039 A CN 109461039A CN 201810986605 A CN201810986605 A CN 201810986605A CN 109461039 A CN109461039 A CN 109461039A
- Authority
- CN
- China
- Prior art keywords
- text
- name entity
- intent information
- intent
- corpus
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0613—Third-party assisted
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
Abstract
The present invention relates to network communication technology fields, provide a kind of text handling method, and the method includes step: intent classifier and name Entity recognition are carried out to the text respectively, to obtain the intent information and name entity that the text is included；In the recognition result, when including the intent information and the name entity, the text is rewritten based on the intent information and the name entity；In the recognition result, when only including one of the intent information or the name entity, the text is rewritten based on the intent information or the name entity got.By the above method to text-processing after, more accurately meaning expressed by text accurately can be understood, and provide good basis for subsequent analysis.In addition, the present invention also provides a kind of intelligent customer service methods.
Description
Technical field
The present invention relates to network communication technology field more particularly to a kind of text handling method and intelligent customer service methods.
Background technique
With the rapid development of Internet, the raising of people's service awareness, network customer service has spread to all trades and professions, deep
Enter the links to everyday commerce service.
Current customer service system is usually made of machine customer service and artificial customer service, and wherein machine customer service is generally based on net
The immediate communication tool of page can specifically be realized based on intelligence chat robots or FAQ system, by internet towards use
Radix is big in the group of family, and limited artificial customer service can not carry out real-time response to advisory customers constantly, so, in customer service processes
In, when receiving the conversation message from customer, first serviced by machine customer service.When customer thinks that machine customer service can not solve
When certainly the problem of its proposition, or having determined that customer identification and infrastructure service information, and needing to further provide for professional service, it can connect
Enter and handled to artificial customer service, so that traveller is lost caused by avoiding network client from falling into a long wait.
In the prior art, the machine customer service either still realized based on FAQ system based on chat robots, requires base
Real-time Feedback is carried out in the information of client's input, and since the form of presentation of Chinese is ever-changing, it is desirable to accurately understand client
The information of input then needs to perform corresponding processing the input of client, to promote the precision of question and answer.
Summary of the invention
The embodiment of the present invention provides a kind of text handling method, and the method includes step: respectively to the text
Intent classifier and name Entity recognition are carried out, to obtain the intent information and name entity that the text is included；When the knowledge
In other result, when including the intent information and the name entity, based on the intent information and the name entity to institute
Text is stated to be rewritten；In the recognition result, when only including one of the intent information or the name entity, base
The text is rewritten in the intent information or the name entity got.
In implementing one, the method for carrying out intent classifier to the text respectively and naming Entity recognition is specifically wrapped
Contain: the intent information for being included to the text based on default intent classifier model identifies；Known based on default name entity
The name entity that other model is included to the text identifies.
In implementing one, the training method of the default intent classifier model includes step: collecting original language material, constructs language
Expect library；Training corpus is extracted from the corpus, and intent information mark is carried out to the training corpus；Calculate the training
The sentence vector of corpus；The default intent classifier model is trained based on the training corpus.
In implementing one, the training corpus is dialogue wheel number more than the question sentence in the session log of preset threshold.
In implementing one, the method for the sentence vector for calculating the training corpus specifically includes: in the corpus
All corpus segmented, and learn the term vector of each participle；The training corpus is segmented, and based on upper
The term vector of all participles, determines the term vector segmented in the training corpus in the corpus that one step obtains；It will be described
The term vector superposition of the participle of each in training corpus is averaged, and the sentence vector of the training corpus is obtained.
It is described to work as in the recognition result in implementing one, when including the intent information and the name entity, it is based on
The intent information and the name entity, which rewrite to the text, specifically includes: splicing the intent information and the life
Name entity.
It is described to work as in the recognition result in implementing one, only comprising in the intent information or the name entity
When a kind of, the text rewrite based on the intent information or the name entity got specifically include: by institute
It states intent information or the name entity is spliced with the text.
In implementing one, when that can not identify the intent information or the name entity that the text is included,
The intent classifier or the name Entity recognition are carried out to the context data of the text, and in the context data
The intent information or the name entity of the intent information or the name entity for being included as the text.
Text handling method provided by the embodiment of the present invention, based on real to intent information included in text and name
Body is obtained, thus the apparent content for having simplified text, and accessed intent information and name entity are settable
For practical application content of interest, further key content accurately can be obtained, and be subsequent analysis or feedback
Basis is provided, so that result is more acurrate, promotes user experience.
In addition, the method includes the present invention also provides a kind of intelligent customer service method: receiving the textual data of user's input
According to；The text data is handled based on above-mentioned text handling method；Based on treated the text data, inquiry is pre-
If Q & A database, the corresponding answer information of the text data is obtained；The answer information is exported to the user.
In implementing one, the construction method of the default Q & A database includes step: collecting history customer service dialogue data；
The session log that dialogue wheel number is more than preset threshold is extracted from the dialogue data；Based in claim 1 to 8
Described in any item text handling methods handle the question sentence in the session log；Save treated the question sentence and
The question sentence corresponding answer in the session log.
Detailed description of the invention
One or more embodiments are illustrated by the picture in corresponding attached drawing, these are exemplary
Illustrate not constitute the restriction to embodiment, the element in attached drawing with same reference numbers label is expressed as similar member
Part, unless there are special statement, composition does not limit the figure in attached drawing.
Fig. 1 is painted text handling method flow chart provided by first embodiment of the invention；
Fig. 2 is painted the training method flow chart that intent classifier model is preset in embodiment illustrated in fig. 1；
Fig. 3 is painted the method flow diagram that training corpus sentence vector is calculated in embodiment illustrated in fig. 2；
Fig. 4 is painted intelligent customer service method flow diagram provided by one embodiment of the invention.
Specific embodiment
To keep the purposes, technical schemes and advantages of embodiment of the present invention clearer, below in conjunction with attached drawing to this hair
Bright each embodiment is explained in detail.However, it will be understood by those skilled in the art that in each implementation of the invention
In mode, in order to make the reader understand this application better, many technical details are proposed.But it is even if thin without these technologies
Section and various changes and modifications based on the following respective embodiments, also may be implemented the application technical solution claimed.
First embodiment provided by the present invention is a kind of text handling method.
Fig. 1 is please referred to, Fig. 1 is painted text handling method flow chart provided by first embodiment of the invention.
As shown in Figure 1, text handling method comprises the steps of:
Step 101, intent classifier and name Entity recognition are carried out to the text respectively, is included to obtain the text
Intent information and name entity.
Text handling method provided by the present embodiment can be applied to the processing of the natural language inputted to user, can also be with
For the processing to the original language material data in corpus, that is to say, that text can be the natural language that user directly inputs,
It is also possible to the corpus data based on application demand, collected.
The method for carrying out intent classifier to the text respectively and naming Entity recognition specifically includes: based on default intention point
The intent information that class model is included to the text identifies；And based on default Named Entity Extraction Model to the text
This name entity for being included is identified.
Specifically, intent information refers to expressed user intent out in content of text, for example, with medical and beauty treatment industry
For, customer may propose problem to the contents of a project, price etc. when seeking advice from relevant item, and these sentences then can include
Relevant intent information, for example, " may I ask and cut how much double-edged eyelid probably need? ", " so, the price of this set meal is how many
? " Deng in consulting question sentence, " inquiry price " this intent information is all contained.It therefore, can be by text in the present embodiment
Intent information is identified, clearer can be understood user's input, be avoided due to expression way difference, and generates mistake
Solution.Further, it is set by the training method to intent classifier model, may make and identified based on intent classifier model
Result out is intent information of interest in practical application, to can guarantee that key subscriber information can be accurately identified.
In the present embodiment, the training method of default intent classifier model can be found in Fig. 2, and Fig. 2 is painted in embodiment illustrated in fig. 1
The training method flow chart of default intent classifier model.
As shown in Fig. 2, the training method of default intent classifier model includes step:
Step 201, original language material is collected, corpus is constructed.
Original language material can be collected according to actual application demand, for example, when text handling method need to finally be applied
When medical and beauty treatment seeks advice from customer service system, the collection of original language material then can be based on the history of medical and beauty treatment industry artificial customer service and client
The modes such as session log, industry Zone Information and terminological dictionary are obtained, the specific collection mode present invention and with no restriction.
Due to that may include the noise datas such as idle character in original language material, therefore those original language material data need to be carried out
Cleaning and pretreatment, comprising rejecting idle character etc..
Step 202, training corpus is extracted from the corpus, and intent information mark is carried out to the training corpus.
Wherein, the rule for extracting training corpus, which can be, to be extracted in corpus, and dialogue wheel number is more than the dialogue of preset threshold
Question sentence in record, as training corpus.
By taking practical application as an example, in customer service system, need to carry out intent classifier is usually the problem of customer proposes, from
And understand the intention of customer, therefore when being trained to intent classifier model, used training corpus is preferable that ground can be chosen
The corpus for having actual analysis to be worth in the session log that dialogue wheel number is more than preset threshold, is cared for through summarizing and analyzing to corpus
Visitor is intended to more specific to know.
And specific preset threshold can be determined based on the actual conditions of corpus.For example, if it is social category to language
Material, dialogue wheel number are that 3-6 wheel is likely to be also successfully dialogue；For the field of medical and beauty treatment, the thing being related to is relatively more,
User can pay close attention to price, safety, time, expert, validity etc. content, statistically see, successfully dialogue generally can be in 6 wheels
It is more than dialogue；For order customer service system, possible 2-3 wheel dialogue is it is known that client will look into order logistics information, know visitor
Then the order number at family directly returns to search result.Therefore the specific value of preset threshold, should be according to specifically using need
It asks to be set.
It selects the question sentence in dialogue as training corpus, and artificial intent information is carried out to training corpus and is marked, wherein
The intent information marked can be the intent information of certain concerns, be also possible to whole intent informations, and the present invention does not make
Limitation.
Step 203, the sentence vector of the training corpus is calculated.
Wherein, the method for calculating the sentence vector of the training corpus can refer to shown in Fig. 3, and Fig. 3 is painted embodiment illustrated in fig. 2
The middle method flow diagram for calculating training corpus sentence vector.
As shown in figure 3, calculate training corpus sentence vector method comprising the following steps:
Step 301, all corpus in the corpus are segmented, and learn the term vector of each participle,
Specifically, the term vector of participle can be calculated based on word2vec.
Step 302, all participles in the corpus for segmenting to the training corpus, and being obtained based on previous step
Term vector, determine the term vector segmented in the training corpus.
Step 303, the term vector superposition of the participle of each in the training corpus is averaged, obtains the instruction
Practice the sentence vector of corpus.
Step 204, the default intent classifier model is trained based on the training corpus.
Based on training corpus and corresponding sentence vector that above-mentioned steps obtain, default intent classifier model is trained,
To improve the accuracy rate of model.Specifically, intent classifier model can be based on traditional machine learning method such as support vector machines
(Support Vector Machine, SVM), random forest (Random Forest) or deep learning method such as convolution mind
It is built-up through network C NN, shot and long term memory network (Long Short-Term Memory, LSTM) scheduling algorithm.
Further, it completes repeatedly after training, desk checking can be carried out to the accuracy of intent classifier model, to
As a result when accuracy reaches preset standard, application can be put into.
In a step 101, the intent information that text is included can be obtained based on above-mentioned default intent classifier model, text
Name entity included in this can be obtained based on Named Entity Extraction Model.
And in some specific implementations, the meaning that text is included possibly can not be identified by default intent classifier model
Figure information, or the name entity that the text that can not be obtained by Named Entity Extraction Model is included.It in this case, can be to text
This context data carries out the intent classifier or the name Entity recognition, and included in the context data
The intent information or the name entity of the intent information or the name entity as the text.
That is, can pass through when that can not know the intent information or name entity that it is included from current text
Intent classifier or name Entity recognition are carried out to the question sentence in the context data of the text, and in this, as the meaning of current text
Figure information or name entity.
Step 102, in the recognition result, when including the intent information and the name entity, it is based on the meaning
Figure information and the name entity rewrite the text.
Specific rewrite method is to splice the intent information and the name entity, that is to say, that by using text
Intent information substitutes the text with name entity, the basis as subsequent analysis.
In this way, by the intent information to text and entity be named to obtain, and the text is substituted with this, as
The basis of subsequent analysis, and acquired intent information and name entity are exactly information of interest in practical application scene, from
And the key message in text can be more accurately got, for subsequent processing.
For example, current text is, " cut double-edged eyelid how much " can recognize to obtain to be intended that included in it and " ask valence
Lattice ", name entity is " cutting double-edged eyelid ", then being " cut double-edged eyelid and ask price " after splicing.Work as to more clearly embody
Meaning expressed by preceding text.
Step 103, in the recognition result, when only including one of the intent information or the name entity,
The text is rewritten based on the intent information or the name entity got.
As it was noted above, current text possibly can not identify included intent information or life in specific implement
Name entity is one such, although the intent information or name can further be supplemented by the analysis to context data in fact
Body, but in practical applications, corresponding intent information or name entity can not also may be obtained from context data, then
In this case, when one of the intent information that can only obtain text or name entity, mode that text is rewritten
It can be with are as follows: splice the intent information or the name entity with the text.That is, when can only be from ought be above
When obtaining one of intent information or name entity in this, it can be spliced with current text, as subsequent analysis
Basis.
And when by the above method can not obtain text intent information and name entity when, then not to current text into
Row processing.
In conclusion text handling method provided by the embodiment of the present invention, based on to intention letter included in text
Breath and name entity are obtained, thus the apparent content for having simplified text, and accessed intent information and name
Entity may be configured as practical application content of interest, further can accurately be obtained to key content, and be subsequent
Analysis or feedback provide basis, so that result is more acurrate, promote user experience.
Based on above-mentioned text handling method, the embodiment of the present invention also provides a kind of intelligent customer service method.
Referring to figure 4., Fig. 4 is painted intelligent customer service method flow diagram provided by one embodiment of the invention.
As shown in figure 4, intelligent customer service method provided by the present embodiment includes step:
Step 401, the text data of user's input is received.
Intelligent customer service method provided by the present embodiment can realize that customer service system provides user's input based on customer service system
Interface when getting user and inputting information, can directly acquire or inverted comprising the input modes such as keyboard, touch-control, voice
To text data.
Step 402, text data is handled.
Specific processing method can refer to the text handling method in embodiment illustrated in fig. 1, repeat no more.
Step 403, based on treated the text data, default Q & A database is inquired, the text data is obtained
Corresponding answer information.
Wherein, default Q & A database is built in advance, and question and answer data are preserved in database, is being actually used
In, database lookup can be carried out based on treated text data, to obtain corresponding reply.
Specifically, the construction method of default Q & A database may include following steps:
Collect history customer service dialogue data；So-called history customer service dialogue data can be to be got from artificial customer service system
Dialogue data is also possible to the data obtained from service system, network.
The session log that dialogue wheel number is more than preset threshold is extracted from the dialogue data；It is analyzed, dialogue wheel number is super
When crossing certain predetermined threshold value, it can be just confirmed to be Effective Dialogue, Effective Dialogue can provide more reference informations, for example, if
It is the dialogue corpus of social category, dialogue wheel number is that 3-6 wheel is likely to be also successfully dialogue；For the field of medical and beauty treatment, relate to
And the thing arrived is relatively more, user can pay close attention to price, safety, time, expert, validity etc. content, statistically see, success
Dialogue generally can 6 wheel dialogue more than；For order customer service system, the dialogue of possible 2-3 wheel is it is known that client will look into orders
Then single logistics information, the order number for knowing client directly return to search result.Therefore the specific value of preset threshold,
It should be set according to specific application demand.
Question sentence in the session log is handled, specific processing method can be found in embodiment illustrated in fig. 1
Text handling method repeats no more.
Save treated the question sentence and the question sentence corresponding answer in the session log.
To generate a question and answer data in database.
Step 404, the answer information is exported to the user.
Intelligent customer service method provided by the embodiment of the present invention first handles the text data of user's input, can obtain
Intent information and name entity of interest in practical application are got, and text data progress database is looked into based on treated
It looks for, to obtain corresponding answer, since question and answer data are asked based on extracting in history customer service data and processed in database
The processing mode of sentence is identical as the text data processing mode that user inputs, so as to more accurately navigate to corresponding answer,
And it exports to user.The accuracy replied for user's input text data is improved to a certain extent, and can more precisely
It is rapidly accurately replied for key message, promotes user experience.
It will be understood by those skilled in the art that the respective embodiments described above are to realize specific embodiment party of the invention
Formula, and in practical applications, can to it, various changes can be made in the form and details, without departing from spirit and model of the invention
It encloses.
Claims (10)
1. a kind of text handling method, which is characterized in that the method includes step:
Intent classifier and name Entity recognition are carried out to the text respectively, with obtain intent information that the text is included and
Name entity；
In the recognition result, when including the intent information and the name entity, it is based on the intent information and described
Name entity rewrites the text；
In the recognition result, when only including one of the intent information or the name entity, based on what is got
The intent information or the name entity rewrite the text.
2. text handling method as described in claim 1, which is characterized in that described to carry out intent classifier to the text respectively
And the method for name Entity recognition specifically includes:
The intent information for being included to the text based on default intent classifier model identifies；
The name entity for being included to the text based on default Named Entity Extraction Model identifies.
3. text handling method as claimed in claim 2, which is characterized in that the training method of the default intent classifier model
Include step:
Original language material is collected, corpus is constructed；
Training corpus is extracted from the corpus, and intent information mark is carried out to the training corpus；
Calculate the sentence vector of the training corpus；
The default intent classifier model is trained based on the training corpus.
4. text handling method as claimed in claim 3, which is characterized in that the training corpus is that dialogue wheel number is more than default
Question sentence in the session log of threshold value.
5. text handling method as claimed in claim 3, which is characterized in that the sentence vector for calculating the training corpus
Method specifically includes:
All corpus in the corpus are segmented, and learn the term vector of each participle；
The term vector of all participles in the corpus for segmenting to the training corpus, and being obtained based on previous step, really
The term vector segmented in the fixed training corpus；
The superposition of the term vector of the participle of each in the training corpus is averaged, obtain the sentence of the training corpus to
Amount.
6. text handling method as described in claim 1, which is characterized in that it is described to work as in the recognition result, comprising described
When intent information and the name entity, the text rewrite based on the intent information and the name entity specific
Include:
Splice the intent information and the name entity.
7. text handling method as described in claim 1, which is characterized in that it is described to work as in the recognition result, it only include institute
When stating one of intent information or the name entity, based on the intent information or the name entity got to institute
It states text and rewrite and specifically include:
The intent information or the name entity are spliced with the text.
8. text handling method as described in claim 1, which is characterized in that when the institute that can not identify that the text is included
When stating intent information or the name entity, the intent classifier is carried out to the context data of the text or the name is real
Body identification, and using the intent information included in the context data or the name entity as the institute of the text
State intent information or the name entity.
9. a kind of intelligent customer service method, which is characterized in that the method includes:
Receive the text data of user's input；
The text data is handled based on text handling method described in any item of the claim 1 to 8；
Based on treated the text data, default Q & A database is inquired, obtains the corresponding answer letter of the text data
Breath；
The answer information is exported to the user.
10. intelligent customer service method as claimed in claim 9, which is characterized in that the construction method of the default Q & A database
Include step:
Collect history customer service dialogue data；
The session log that dialogue wheel number is more than preset threshold is extracted from the dialogue data；
Question sentence in the session log is handled based on text handling method described in any item of the claim 1 to 8；
Save treated the question sentence and the question sentence corresponding answer in the session log.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810986605.0A CN109461039A (en) | 2018-08-28 | 2018-08-28 | A kind of text handling method and intelligent customer service method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810986605.0A CN109461039A (en) | 2018-08-28 | 2018-08-28 | A kind of text handling method and intelligent customer service method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109461039A true CN109461039A (en) | 2019-03-12 |
Family
ID=65606421
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810986605.0A Pending CN109461039A (en) | 2018-08-28 | 2018-08-28 | A kind of text handling method and intelligent customer service method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109461039A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110008325A (en) * | 2019-03-29 | 2019-07-12 | 海南中智信信息技术有限公司 | A kind of conversational language understanding and Improvement based on commercial conversational system |
| CN110046242A (en) * | 2019-04-22 | 2019-07-23 | 北京六行君通信息科技股份有限公司 | A kind of automatic answering device and method |
| CN110096570A (en) * | 2019-04-09 | 2019-08-06 | 苏宁易购集团股份有限公司 | A kind of intension recognizing method and device applied to intelligent customer service robot |
| CN110399609A (en) * | 2019-06-25 | 2019-11-01 | 众安信息技术服务有限公司 | Intension recognizing method, device, equipment and computer readable storage medium |
| CN110427611A (en) * | 2019-06-26 | 2019-11-08 | 深圳追一科技有限公司 | Text handling method, device, equipment and storage medium |
| CN110569343A (en) * | 2019-08-16 | 2019-12-13 | 华东理工大学 | question and answer based clinical text structuring method |
| CN110827831A (en) * | 2019-11-15 | 2020-02-21 | 广州洪荒智能科技有限公司 | Voice information processing method, device, equipment and medium based on man-machine interaction |
| CN111177358A (en) * | 2019-12-31 | 2020-05-19 | 华为技术有限公司 | Intention recognition method, server, and storage medium |
| CN111428483A (en) * | 2020-03-31 | 2020-07-17 | 华为技术有限公司 | Voice interaction method and device and terminal equipment |
| CN111597342A (en) * | 2020-05-22 | 2020-08-28 | 北京慧闻科技(集团)有限公司 | Multitask intention classification method, device, equipment and storage medium |
| CN112487179A (en) * | 2019-09-11 | 2021-03-12 | 珠海格力电器股份有限公司 | Spoken language semantic understanding method, device and system |
| CN112699233A (en) * | 2019-10-17 | 2021-04-23 | 中国移动通信集团浙江有限公司 | Service processing method and device and electronic equipment |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101377777A (en) * | 2007-09-03 | 2009-03-04 | 北京百问百答网络技术有限公司 | Automatic inquiring and answering method and system |
| US8285539B2 (en) * | 2007-06-18 | 2012-10-09 | International Business Machines Corporation | Extracting tokens in a natural language understanding application |
| CN105893465A (en) * | 2016-03-28 | 2016-08-24 | 北京京东尚科信息技术有限公司 | Automatic question answering method and device |
| US9588960B2 (en) * | 2014-01-15 | 2017-03-07 | Abbyy Infopoisk Llc | Automatic extraction of named entities from texts |
| CN106484801A (en) * | 2016-09-23 | 2017-03-08 | 厦门快商通科技股份有限公司 | A kind of dialogue method of intelligent customer service robot and its knowledge base management system |
| CN106815192A (en) * | 2015-11-27 | 2017-06-09 | 北京国双科技有限公司 | Model training method and device and sentence emotion identification method and device |
| CN107133259A (en) * | 2017-03-22 | 2017-09-05 | 北京晓数聚传媒科技有限公司 | A kind of searching method and device |
| CN108038234A (en) * | 2017-12-26 | 2018-05-15 | 众安信息技术服务有限公司 | A kind of question sentence template automatic generation method and device |
| CN108427722A (en) * | 2018-02-09 | 2018-08-21 | 卫盈联信息技术（深圳）有限公司 | intelligent interactive method, electronic device and storage medium |
-
2018
- 2018-08-28 CN CN201810986605.0A patent/CN109461039A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8285539B2 (en) * | 2007-06-18 | 2012-10-09 | International Business Machines Corporation | Extracting tokens in a natural language understanding application |
| CN101377777A (en) * | 2007-09-03 | 2009-03-04 | 北京百问百答网络技术有限公司 | Automatic inquiring and answering method and system |
| US9588960B2 (en) * | 2014-01-15 | 2017-03-07 | Abbyy Infopoisk Llc | Automatic extraction of named entities from texts |
| CN106815192A (en) * | 2015-11-27 | 2017-06-09 | 北京国双科技有限公司 | Model training method and device and sentence emotion identification method and device |
| CN105893465A (en) * | 2016-03-28 | 2016-08-24 | 北京京东尚科信息技术有限公司 | Automatic question answering method and device |
| CN106484801A (en) * | 2016-09-23 | 2017-03-08 | 厦门快商通科技股份有限公司 | A kind of dialogue method of intelligent customer service robot and its knowledge base management system |
| CN107133259A (en) * | 2017-03-22 | 2017-09-05 | 北京晓数聚传媒科技有限公司 | A kind of searching method and device |
| CN108038234A (en) * | 2017-12-26 | 2018-05-15 | 众安信息技术服务有限公司 | A kind of question sentence template automatic generation method and device |
| CN108427722A (en) * | 2018-02-09 | 2018-08-21 | 卫盈联信息技术（深圳）有限公司 | intelligent interactive method, electronic device and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 朱晓明: "《走向数字经济》", 31 January 2018, 上海交通大学出版社 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110008325B (en) * | 2019-03-29 | 2020-02-07 | 海南中智信信息技术有限公司 | Spoken language understanding and rewriting method based on commercial conversation system |
| CN110008325A (en) * | 2019-03-29 | 2019-07-12 | 海南中智信信息技术有限公司 | A kind of conversational language understanding and Improvement based on commercial conversational system |
| WO2020206957A1 (en) * | 2019-04-09 | 2020-10-15 | 苏宁云计算有限公司 | Intention recognition method and device for intelligent customer service robot |
| CN110096570A (en) * | 2019-04-09 | 2019-08-06 | 苏宁易购集团股份有限公司 | A kind of intension recognizing method and device applied to intelligent customer service robot |
| CN110096570B (en) * | 2019-04-09 | 2021-03-30 | 苏宁易购集团股份有限公司 | Intention identification method and device applied to intelligent customer service robot |
| CN110046242A (en) * | 2019-04-22 | 2019-07-23 | 北京六行君通信息科技股份有限公司 | A kind of automatic answering device and method |
| CN110399609A (en) * | 2019-06-25 | 2019-11-01 | 众安信息技术服务有限公司 | Intension recognizing method, device, equipment and computer readable storage medium |
| CN110399609B (en) * | 2019-06-25 | 2023-12-01 | 众安信息技术服务有限公司 | Intention recognition method, device, equipment and computer readable storage medium |
| CN110427611A (en) * | 2019-06-26 | 2019-11-08 | 深圳追一科技有限公司 | Text handling method, device, equipment and storage medium |
| CN110569343A (en) * | 2019-08-16 | 2019-12-13 | 华东理工大学 | question and answer based clinical text structuring method |
| CN112487179A (en) * | 2019-09-11 | 2021-03-12 | 珠海格力电器股份有限公司 | Spoken language semantic understanding method, device and system |
| CN112699233A (en) * | 2019-10-17 | 2021-04-23 | 中国移动通信集团浙江有限公司 | Service processing method and device and electronic equipment |
| CN110827831A (en) * | 2019-11-15 | 2020-02-21 | 广州洪荒智能科技有限公司 | Voice information processing method, device, equipment and medium based on man-machine interaction |
| CN111177358A (en) * | 2019-12-31 | 2020-05-19 | 华为技术有限公司 | Intention recognition method, server, and storage medium |
| CN111177358B (en) * | 2019-12-31 | 2023-05-12 | 华为技术有限公司 | Intention recognition method, server and storage medium |
| CN111428483A (en) * | 2020-03-31 | 2020-07-17 | 华为技术有限公司 | Voice interaction method and device and terminal equipment |
| CN111597342A (en) * | 2020-05-22 | 2020-08-28 | 北京慧闻科技(集团)有限公司 | Multitask intention classification method, device, equipment and storage medium |
| CN111597342B (en) * | 2020-05-22 | 2024-01-26 | 北京慧闻科技(集团)有限公司 | Multitasking intention classification method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109461039A (en) | A kind of text handling method and intelligent customer service method | |
| CN109522556B (en) | Intention recognition method and device | |
| CN111062220B (en) | End-to-end intention recognition system and method based on memory forgetting device | |
| CN103593412B (en) | A kind of answer method and system based on tree structure problem | |
| CN109885664A (en) | A kind of Intelligent dialogue method, robot conversational system, server and storage medium | |
| CN113360622B (en) | User dialogue information processing method and device and computer equipment | |
| CN108984655B (en) | Intelligent customer service guiding method for customer service robot | |
| CN113268610B (en) | Intent jump method, device, equipment and storage medium based on knowledge graph | |
| CN110704586A (en) | Information processing method and system | |
| CN109684444A (en) | A kind of intelligent customer service method and system | |
| CN109460451A (en) | Actively obtain method, the intelligent customer service method and system of user information | |
| CN113672720A (en) | Power audit question and answer method based on knowledge graph and semantic similarity | |
| CN114722839A (en) | Man-machine collaborative dialogue interaction system and method | |
| CN114218375A (en) | Dialogue guiding method, device, equipment and medium based on atlas | |
| CN113239698A (en) | Information extraction method, device, equipment and medium based on RPA and AI | |
| CN112328812B (en) | Domain knowledge extraction method and system based on self-adjusting parameters and electronic equipment | |
| CN115455151A (en) | AI emotion visual identification method and system and cloud platform | |
| CN111324722B (en) | Method and system for training word weight model | |
| CN111984764A (en) | Stream bill reason analysis method and device, computer equipment and readable storage medium | |
| Moberg et al. | Artificial Intelligence Adoption–Is it more than just hype? | |
| CN110956027A (en) | Method and device for generating digital short message content | |
| CN112395855A (en) | Comment-based evaluation method and device | |
| Galitsky | Finding a Lattice of Needles in a Haystack: Forming a Query from a Set of Items of Interest. | |
| CN113051388B (en) | Intelligent question-answering method and device, electronic equipment and storage medium | |
| CN117271778B (en) | Insurance outbound session information output method and device based on generation type large model |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190312 |
|
| RJ01 | Rejection of invention patent application after publication |