CN107862005A - User view recognition methods and device - Google Patents
User view recognition methods and device Download PDFInfo
- Publication number
- CN107862005A CN107862005A CN201711005661.3A CN201711005661A CN107862005A CN 107862005 A CN107862005 A CN 107862005A CN 201711005661 A CN201711005661 A CN 201711005661A CN 107862005 A CN107862005 A CN 107862005A
- Authority
- CN
- China
- Prior art keywords
- key element
- question
- multiple predefined
- predefined key
- rhetorical
- 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
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Machine Translation (AREA)
- User Interface Of Digital Computer (AREA)
Abstract
This specification embodiment provides a kind of user view recognition methods and device, in user view recognition methods, obtains the dialog text of user first, and determine the dialog text and multiple predefined key elements associates angle value.Afterwards according to the association angle value of above-mentioned determination, the first element to match with dialog text is chosen.Finally according to the first element of selection, the knowledge point corresponding to matching from knowledge base.After knowledge point is matched, user view recognition result is determined according to the knowledge point.
Description
Technical field
This specification one or more embodiment is related to field of computer technology, more particularly to a kind of user view identification side
Method and device.
Background technology
In online or hotline service, hot line system can receive the various problems of user's transmission, and reception is asked
Topic is analyzed.The problem of by receiving, is analyzed, and identifies the intention of user.It is accordingly, it is desirable to provide a kind of quick
Identify the scheme of user view.
The content of the invention
This specification one or more embodiment describes a kind of user view recognition methods and device, with rapidly to
The problem of family, is answered.
First aspect, there is provided a kind of user view recognition methods, including：
Obtain the dialog text of user；
Associate angle value of the dialog text with multiple predefined key elements is determined, the multiple predefined key element is from knowledge
Extracted in the knowledge point in storehouse, and the multiple predefined key element is respectively belonging to N number of classification, N is positive integer；
According to the association angle value and predetermined threshold value, the first element is chosen from the multiple predefined key element；
According to the first element, the knowledge point corresponding to matching from the knowledge base；
According to the knowledge point, user view recognition result is determined.
Second aspect, there is provided a kind of user view identification device, including：
Acquiring unit, for obtaining the dialog text of user；
Determining unit, for determining the dialog text of the acquiring unit acquisition and associating for multiple predefined key elements
Angle value, the multiple predefined key element are extracted from the knowledge point of knowledge base, and the multiple predefined key element is returned respectively
Belong to N number of classification, N is positive integer；
Unit is chosen, for the association angle value and predetermined threshold value determined according to the determining unit, from described more
The first element is chosen in individual predefined key element；
Matching unit, for the first element according to the selection unit selection, the matching pair from the knowledge base
The knowledge point answered；
The determining unit, the knowledge point matched according to the matching unit is additionally operable to, determines that user view identifies
As a result.
The user view recognition methods and device that this specification one or more embodiment provides, pair of user is obtained first
Text is talked about, and determine the dialog text and multiple predefined key elements associates angle value.Afterwards according to the association angle value of above-mentioned determination,
Choose the first element to match with dialog text.Finally according to the first element of selection, know from knowledge base corresponding to matching
Know point.After knowledge point is matched, user view recognition result is determined according to the knowledge point.Thus, it is possible to rapidly identify
User view.
Brief description of the drawings
In order to illustrate the technical solution of the embodiments of the present invention more clearly, required use in being described below to embodiment
Accompanying drawing be briefly described, it should be apparent that, drawings in the following description are only some embodiments of the present invention, for this
For the those of ordinary skill of field, on the premise of not paying creative work, it can also be obtained according to these accompanying drawings other
Accompanying drawing.
Fig. 1 is the application scenarios schematic diagram for the user view recognition methods that this specification one embodiment provides；
Fig. 2 is the user view recognition methods flow chart that this specification one embodiment provides；
Fig. 3 is that this illustrates the user view recognition methods schematic diagram that another embodiment provides；
Fig. 4 is the user view identification device schematic diagram that this specification one embodiment provides.
Embodiment
Below in conjunction with the accompanying drawings, the scheme provided this specification is described.
The user view recognition methods that this specification one embodiment provides can apply in scene as shown in Figure 1,
In Fig. 1, hot line system can be any system that can provide " intelligent robot " service.Specifically, it can receive user's hair
The various problems sent, and the problem of reception is analyzed.The problem of by receiving, is analyzed, and identifies the meaning of user
Figure.It should be noted that during user view is identified, hot line system can carry out more wheel sessions with user.Identifying
Go out after the intention of user, the problem of user is answered.
In Fig. 1, knowledge base can be stored with hot line system, the knowledge base is made up of one or more knowledge points.Wherein,
Each knowledge point has corresponding answer scheme.Knowledge point herein can be understood as the rule of the problem of user to collecting in advance
Model is stated.Specifically, hot line system can be carried out the knowledge point in problem and knowledge base after the problem of receiving user
Matching.Using scheme is answered corresponding to the knowledge point to match user is sent to as answer the problem of user.
Fig. 2 is the user view recognition methods flow chart that this specification one embodiment provides.The execution master of methods described
Body can be the equipment with disposal ability：Server either system or device, e.g., the hot line system in Fig. 1.Such as Fig. 2 institutes
Show, methods described can specifically include：
Step 210, the dialog text of user is obtained.
Dialog text herein can be any word that can state the problem of user wants to ask.Such as, Ke Yiwei：" flower
Also not upper money, what if", " flower how to refund" and " how is Yuebao income" etc..
Step 220, determine dialog text and multiple predefined key elements associates angle value.
It is understood that because the dialog text of the user directly obtained compares colloquial style, machine is not allowed easy to identify.Cause
This, after the dialog text of user is got, generally will first be pre-processed, e.g., word segmentation processing etc..Word segmentation processing herein
It is traditional routine techniques Deng preprocessing process, does not repeat again herein.
Predefined key element herein can extract from knowledge point.Such as, for " flower refund fails " this knowledge
Point, " flower ", " refunds " and " failure " can serve as predefining key element.Above-mentioned predefined key element can have corresponding class
Not, a predefined key element can uniquely belong to a classification, namely can not intersect presence between predefined key element.
In one example, the predefined key element of above-mentioned extraction can belong to following three classification：" type of service ", " framework verb " with
And " problem types ".Such as previous example, " flower " can belong to " type of service ", and " refund " can belong to that " framework moves
Word ", " failure " can belong to " problem types ".It should be noted that " type of service " be one than broad classification, return
Belonging to the predefined key element of the category can also be：" account ", " password ", " Yuebao ", " flower ", " borrow ", " periodically reason
Wealth " etc..The predefined key element for belonging to " framework verb " is typically verb or verb phrase most crucial in knowledge point, e.g.,
Can be：" login ", " forgetting ", " payment ", " refund ", " loaning bill " etc..Belonging to " problem types " can be：" what ", " what
When ", " where ", " whether ", " failure ", " what meaning " etc..
In an example of this specification, the predefined key element of some knowledge points in knowledge base can be allowed to be belonged to
Class number be less than total class number.As an example it is assumed that total class that the predefined key element in certain knowledge base is belonged to
Other number is 3, is respectively：" type of service ", " framework verb " and " problem types ".Some knowledge point for " flower also
Money ", that is, it is respectively " type of service " and " framework verb " to form the classification that the predefined key element of the knowledge point is belonged to, namely class
Other number is 2.
In another example of this specification, above-mentioned predefined key element can also have corresponding extension key element, the expansion
The number for opening up key element can be multiple.Its effect can be to preferably identify the predefined key element in dialog text.
In one example, the extension key element can be the alias of corresponding predefined key element, and the alias is the title for facilitating user to understand.
Such as, when predefined key element is " expense transformation ", corresponding extension key element can be " Alipay " etc..
In addition, the predefined key element of this specification generally can accurately express implication, while there is generalization.It is appreciated that
, preferably predefine key element and typically occur in more than two knowledge points.In the further example of this specification, know
Attached description section can also be included by knowing point, and the part can form rhetorical question question sentence during more wheel sessions are carried out with user
When use.It is respectively with classification：For exemplified by " type of service ", " framework verb " and " problem types ", one is completely known
Knowing the included content part of point can be as shown in table 1.
Table 1
It is understood that in table 1, predefined key element 1, predefined key element 2 and predefined key element 3 may be constructed one
Knowledge point.
Certainly, in actual applications, the knowledge point in table 1 can also include other parts content, e.g., answer scheme etc.,
This specification is not construed as limiting to this.
In step 220, determine that dialog text and the method for associating angle value of multiple predefined key elements can include：Model is known
Method for distinguishing and/or the method for calculating text matches angle value.The method of Model Identification is specially：By pretreated dialog text
Key element identification model is inputted, the key element identification model can be the good disaggregated model of training in advance, can be with by the disaggregated model
Directly obtain the probable value that dialog text is categorized into each predefined key element.Using above-mentioned probable value as above-mentioned association angle value.
Calculate text matches angle value method be specially：By similarity algorithm, dialog text and the phase of each predefined key element are calculated
Like angle value, using the Similarity value as above-mentioned association angle value.
When determining above-mentioned association angle value by two methods, the results added that directly can obtain two methods or
Person can also be merged by the result that method for distinguishing obtains two methods.
In one example, each predefined key element and corresponding association angle value can be as shown in table 2.
Table 2
It is understood that often the predefined key element of row may be constructed a knowledge point in table 1, namely these predefine and wanted
Element extracts from corresponding knowledge point.It should be noted that table 2 is only to facilitate understand the present embodiment and showing for providing
Example property explanation, is not intended as the limitation of the present embodiment.Such as, predefined key element of other classifications etc. can also be included in table 2.
Step 230, according to association angle value and predetermined threshold value, the first element is chosen from multiple predefined key elements.
In one implementation, before step 230 is performed, the grade letter of each predefined key element can first be determined
Breath, afterwards according to class information, the first element is chosen from multiple predefined key elements.Its detailed process can be：According to association
Angle value and predetermined threshold value, determine class information corresponding to each predefined key element., will from multiple predefine according to class information
The first element is chosen in element.
For by taking table 2 as an example, two threshold values can be preset：0.8 and 0.6.Specifically, when the association of predefined key element
When angle value is more than or equal to 0.8, the class information of predefined key element can be defined as it is high-grade (e.g., in table 2 " flower ", " also
Money " and the class information of " failure " are high-grade).When the association angle value of predefined key element is between 0.6 and 0.8, can incite somebody to action
The class information of predefined key element is defined as middle grade (class information e.g., " forgotten " in table 2 is middle grade).Wanted when predefined
When the association angle value of element is less than 0.6, the class information of predefined key element can be defined as inferior grade.It should also be noted that,
In this manual, high-grade predefined key element is properly termed as credible key element, and the predefined key element of middle grade is properly termed as waiting
Key element is selected, the predefined key element of inferior grade is properly termed as unfamiliar feature.
A kind of it should be noted that above-mentioned simply mode of given threshold.In other implementations, to belonging to difference
The predefined key element of classification, can set different threshold values, and this specification is not construed as limiting to this.
Can be the by high-grade predefined select factors after the class information of each predefined key element is determined
One key element, also can be the first element by credible select factors.Such as, can be by " flower " in table 2, " refund " and " failure "
It is chosen for the first element.
Step 240, according to the first element, the knowledge point corresponding to matching from knowledge base.
From table 1 it follows that Radix Angelicae Sinensis belong to same category of predefined key element it is unique when, can just match unique
Knowledge point.Therefore, it is unique to belong to the same category of first element for this specification requirement, namely in this specification under same category
Credible key element be it is unique, and candidate's key element can then have it is multiple.
Such as previous example, can match knowledge point is：" flower refund failure ".
Step 250, according to knowledge point, user view recognition result is determined.
After unique knowledge point is matched, it is possible to determine user view recognition result.
Certainly, in actual applications, it is possible to the situation for the credible key element that can not be selected under all categories occurs, this
When can match multiple knowledge points.Such as previous example, credible key element may be only selected：" flower " and " refund ", namely it is unselected
Take out the credible key element under " problem types ".Two knowledge points can be then matched from table 2：" flower refund failure " and " flower
How to refund ".
Interacted when matching multiple knowledge points, it is necessary to carry out more wheels with user, to determine user view.Enter with user
During the more wheel interactions of row, how to determine to ask in reply question sentence, turn into key to quickly determine user view.In this specification
In, it is proposed that the method for rhetorical question question sentence is identified below：
Step A, according to class information, the second key element is chosen from multiple predefined key elements.
In one implementation, it can be the second key element by the predefined select factors of middle grade, will can also wait
It is the second key element to select select factors.Such as, " failure " in table 2 can be chosen for the second key element.
Step B, the classification belonged to according to the first element, the second key element and class information, read from rhetorical question ATL
Take the rhetorical question template to match.
Rhetorical question ATL herein is used to record between the class information for belonging to different classes of key element and rhetorical question template
Corresponding relation.In one example, asking in reply ATL can be as shown in table 3.
Table 3
In table 3, " √ " represents high-grade, and " √ " represents middle grade, and "×" represents inferior grade.As an example it is assumed that choose
The first element (i.e. credible key element) be：" flower " and " refund ", the second key element (i.e. candidate's key element) of selection are：" failure ".
Because the classification that " flower " is belonged to is：" type of service ", the classification that " refund " is belonged to are：" framework verb ", " failure " institute
The classification of ownership is " problem types ".In addition, the class information corresponding to " flower " and " refund " is：" high-grade ", " failure "
Corresponding class information is：" middle grade ".So the 5th row in table 3 can be matched, namely the rhetorical question template read is：
" may I ask you in #business##frame#, #type# also what happened to”.
Step C, according to the first element, the second key element and rhetorical question template, it is determined that corresponding rhetorical question question sentence.
Specifically, information different classes of in rhetorical question template can be replaced with to the first element or second under the category
Key element.Such as previous example, it can be assumed that " #business# " is the information of " type of service ", and " #frame# " is " framework verb "
Information, " #type# " is the information of " problem types ".The corresponding rhetorical question question sentence then determined is：" you is may I ask in flower refund
When, it have failed what happened to going back”.
It is determined that when rhetorical question question sentence, when the title of the first element or the second key element, to compare special user not understandable
, then corresponding attached description is could alternatively be, a kind of purpose of expression way is changed to reach.If the in addition, rhetorical question of generation
Question sentence is not clear and coherent enough, and needs to interact again, then can be realized by configuring specific rhetorical question question sentence.
Step D, rhetorical question question sentence is sent to user.
Step E, according to the answer of the rhetorical question question sentence of reception, determine user view recognition result.
Herein, answer of the user to rhetorical question question sentence is being received, it is possible to final to determine user view recognition result.
To sum up, by the first element and the second key element in this specification above-described embodiment, with reference to rhetorical question masterplate, adaptation there emerged a
The rhetorical question question sentence of property, reduces interactive difficulty as much as possible.
It should be noted that during above-mentioned steps 210- steps 250 are performed, it is possible to which choosing will less than first
Element, it is also possible to the first element can be chosen, but matching can now be sent less than knowledge point to user from knowledge base
Default rhetorical question question sentence, e.g., " it may I ask what problem you encounter”.It should also be noted that, in this manual, for certain
A little special dialog texts, can there is specific way to put questions, and it can be completed by configuring.Such as, in the first element and default set
In key element it is identical when, obtain the default rhetorical question question sentence corresponding with key element.According to answering for the default rhetorical question question sentence of reception
Case, determine user view recognition result.
It our experiments show that, the user view recognition methods proposed by this specification above-described embodiment, user can be anticipated
Figure recognition accuracy and recall rate all lift 5%.In addition, by the above-mentioned interaction for targetedly asking in reply question sentence, user is more willing to
More wheels are carried out with robot to talk with.
Fig. 3 is the user view recognition methods schematic diagram of this specification another embodiment offer., can be advance in Fig. 3
Extract three elements from the knowledge point in knowledge base, i.e., extracted from the knowledge point of knowledge base be respectively belonging to " type of service ",
The key element of " framework verb " and " problem types ".When receiving the dialog text of user, by calculate dialog text with it is pre-
The identification probability value of the key element first extracted, to identify the key element in dialog text.Afterwards, according to the identification probability and threshold of calculating
Value, credible key element, candidate's key element and unfamiliar feature are divided into by the key element of identification.Finally, by credible key element from knowledge base
Knowledge point corresponding to middle matching.If matching corresponding knowledge point, user view is directly exported.If match multiple knowledge
Point, then template is asked in reply corresponding to the matching from rhetorical question ATL according to credible key element and candidate's key element.Afterwards according to the anti-of matching
Ask that template to export rhetorical question question sentence to user, and according to the answer of rhetorical question question sentence, identify user view.
With above-mentioned user view recognition methods accordingly, a kind of user view that this specification one embodiment also provides is known
Other device, as shown in figure 4, the device includes：
Acquiring unit 401, for obtaining the dialog text of user.
Determining unit 402, for determining the dialog text and the degree of association of multiple predefined key elements of the acquisition of acquiring unit 401
Value, multiple predefined key elements are extracted from the knowledge point of knowledge base, and multiple predefined key elements are respectively belonging to N number of class
Not, N is positive integer.
Classification herein can include：Type of service, framework verb and problem types etc..
Optionally it is determined that unit 402 specifically can be used for：
Dialog text is inputted into key element identification model, exports probable value corresponding to multiple predefined key elements, wherein key element is known
Other model is the predefined machine learning mould for being used to calculate the probable value that the text inputted matches with multiple predefined key elements
Type.
And/or
Calculate dialog text and the matching angle value of multiple predefined key elements.
According to probable value and/or matching angle value, determine dialog text and multiple predefined key elements associates angle value.
Unit 403 is chosen, for the association angle value and predetermined threshold value determined according to determining unit 402, is made a reservation for from multiple
The first element is chosen in adopted key element.
Alternatively, unit 403 is chosen specifically to can be used for：
According to association angle value and predetermined threshold value, class information corresponding to multiple predefined key elements is determined.
According to class information, the first element is chosen from multiple predefined key elements.
Matching unit 404, for the first element chosen according to unit 403 is chosen, know from knowledge base corresponding to matching
Know point.
Determining unit 402, the knowledge point matched according to matching unit 404 is additionally operable to, determines user view recognition result.
Optionally it is determined that unit 402 can be also used for：
According to class information, the second key element is chosen from multiple predefined key elements.
The classification and class information belonged to according to the first element, the second key element, phase is read from rhetorical question ATL
The rhetorical question template matched somebody with somebody, rhetorical question ATL are used to record between the class information for belonging to different classes of key element and rhetorical question template
Corresponding relation.
According to the first element, second key element and rhetorical question template, it is determined that corresponding rhetorical question question sentence.
Rhetorical question question sentence is sent to user.
According to the answer of the rhetorical question question sentence of reception, user view recognition result is determined.
Alternatively, the device can also include：
Transmitting element 405, for sending default rhetorical question question sentence to user.
Determining unit 402, the answer of the default rhetorical question question sentence according to reception is additionally operable to, determines user view identification knot
Fruit.
Alternatively, acquiring unit 401, it is additionally operable to, when the first element is identical with the key element in default set, obtain and want
The corresponding default rhetorical question question sentence of element.
Determining unit 402, the answer of the default rhetorical question question sentence according to reception is additionally operable to, determines user view identification knot
Fruit.
The function of each functional module of this specification above-described embodiment device, each step of above method embodiment can be passed through
It is rapid to realize, therefore, the specific work process for the device that this specification one embodiment provides, do not repeat again herein.
The user view identification device that this specification one embodiment provides, acquiring unit 401 obtain the dialogue text of user
This.The dialog text that determining unit 402 determines to obtain associates angle value with multiple predefined key elements.Unit 403 is chosen according to pass
Join angle value and predetermined threshold value, the first element is chosen from multiple predefined key elements.Matching unit 404 will according to the first of selection
Element, the knowledge point corresponding to matching from knowledge base.Determining unit 402 determines user view identification knot according to the knowledge point of matching
Fruit.Thus, it is possible to rapidly identify user view.
Those skilled in the art are it will be appreciated that in said one or multiple examples, work(described in the invention
It is able to can be realized with hardware, software, firmware or their any combination.When implemented in software, can be by these functions
It is stored in computer-readable medium or is transmitted as one or more instructions on computer-readable medium or code.
Above-described embodiment, the purpose of the present invention, technical scheme and beneficial effect are carried out further
Describe in detail, should be understood that the embodiment that the foregoing is only the present invention, be not intended to limit the present invention
Protection domain, all any modification, equivalent substitution and improvements on the basis of technical scheme, done etc., all should
It is included within protection scope of the present invention.
Claims (14)
- A kind of 1. user view recognition methods, it is characterised in that including：Obtain the dialog text of user；Associate angle value of the dialog text with multiple predefined key elements is determined, the multiple predefined key element is from knowledge base Extracted in knowledge point, and the multiple predefined key element is respectively belonging to N number of classification, N is positive integer；According to the association angle value and predetermined threshold value, the first element is chosen from the multiple predefined key element；According to the first element, the knowledge point corresponding to matching from the knowledge base；According to the knowledge point, user view recognition result is determined.
- 2. according to the method for claim 1, it is characterised in that described to determine the dialog text and multiple predefined key elements Association angle value, including：The dialog text is inputted into key element identification model, exports probable value corresponding to the multiple predefined key element, wherein will Plain identification model is the predefined engineering for being used to calculate the probable value that the text inputted matches with multiple predefined key elements Practise model；And/orCalculate the matching angle value of the dialog text and the multiple predefined key element；According to the probable value and/or the matching angle value, the degree of association of the dialog text and multiple predefined key elements is determined Value.
- 3. method according to claim 1 or 2, it is characterised in that it is described to associate angle value and predetermined threshold value according to described, The first element is chosen from the multiple predefined key element, including：According to the association angle value and the predetermined threshold value, class information corresponding to the multiple predefined key element is determined；According to the class information, the first element is chosen from the multiple predefined key element.
- 4. according to the method for claim 3, it is characterised in that when the knowledge point of matching is multiple, described in the basis Knowledge point, user view recognition result is determined, including：According to the class information, the second key element is chosen from the multiple predefined key element；The classification and class information belonged to according to the first element, second key element, read from rhetorical question ATL The rhetorical question template to match, the rhetorical question ATL are used to record the class information for belonging to different classes of key element and rhetorical question mould Corresponding relation between plate；According to the first element, second key element and the rhetorical question template, it is determined that corresponding rhetorical question question sentence；The rhetorical question question sentence is sent to user；According to the answer of the rhetorical question question sentence of reception, user view recognition result is determined.
- 5. according to the method for claim 3, it is characterised in that ought not match when the class information is inferior grade or During to knowledge point, in addition to：Default rhetorical question question sentence is sent to user；According to the answer of the default rhetorical question question sentence of reception, user view recognition result is determined.
- 6. according to the method for claim 1, it is characterised in that also include：When the first element is identical with the key element in default set, obtains the default rhetorical question corresponding with the key element and ask Sentence；According to the answer of the default rhetorical question question sentence of reception, user view recognition result is determined.
- 7. according to the method described in claim any one of 1-6, it is characterised in that the classification includes：Type of service, framework move Word and problem types.
- A kind of 8. user view identification device, it is characterised in that including：Acquiring unit, for obtaining the dialog text of user；Determining unit, for determining the dialog text and the degree of association of multiple predefined key elements of the acquiring unit acquisition Value, the multiple predefined key element is extracted from the knowledge point of knowledge base, and the multiple predefined key element belongs to respectively In N number of classification, N is positive integer；Unit is chosen, for the association angle value and predetermined threshold value determined according to the determining unit, from the multiple pre- Define in key element and choose the first element；Matching unit, for according to it is described selection unit selection the first element, from the knowledge base matching corresponding to Knowledge point；The determining unit, the knowledge point matched according to the matching unit is additionally operable to, determines user view recognition result.
- 9. device according to claim 8, it is characterised in that the determining unit is specifically used for：The dialog text is inputted into key element identification model, exports probable value corresponding to the multiple predefined key element, wherein will Plain identification model is the predefined engineering for being used to calculate the probable value that the text inputted matches with multiple predefined key elements Practise model；And/orCalculate the matching angle value of the dialog text and the multiple predefined key element；According to the probable value and/or the matching angle value, the degree of association of the dialog text and multiple predefined key elements is determined Value.
- 10. device according to claim 8 or claim 9, it is characterised in that the selection unit is specifically used for：According to the association angle value and the predetermined threshold value, class information corresponding to the multiple predefined key element is determined；According to the class information, the first element is chosen from the multiple predefined key element.
- 11. device according to claim 10, it is characterised in that the determining unit is additionally operable to：According to the class information, the second key element is chosen from the multiple predefined key element；The classification and class information belonged to according to the first element, second key element, read from rhetorical question ATL The rhetorical question template to match, the rhetorical question ATL are used to record the class information for belonging to different classes of key element and rhetorical question mould Corresponding relation between plate；According to the first element, second key element and the rhetorical question template, it is determined that corresponding rhetorical question question sentence；The rhetorical question question sentence is sent to user；According to the answer of the rhetorical question question sentence of reception, user view recognition result is determined.
- 12. device according to claim 10, it is characterised in that also include：Transmitting element, for sending default rhetorical question question sentence to user；The determining unit, the answer of the default rhetorical question question sentence according to reception is additionally operable to, determines user view identification knot Fruit.
- 13. device according to claim 8, it is characterised in thatThe acquiring unit, it is additionally operable to, when the first element is identical with the key element in default set, obtain and the key element Corresponding default rhetorical question question sentence；The determining unit, the answer of the default rhetorical question question sentence according to reception is additionally operable to, determines user view identification knot Fruit.
- 14. according to the device described in claim any one of 8-13, it is characterised in that the classification includes：Type of service, framework Verb and problem types.
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711005661.3A CN107862005A (en) | 2017-10-25 | 2017-10-25 | User view recognition methods and device |
| TW107129570A TWI700632B (en) | 2017-10-25 | 2018-08-24 | User intention recognition method and device |
| PCT/CN2018/105192 WO2019080661A1 (en) | 2017-10-25 | 2018-09-12 | Method and device for identifying intention of user |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711005661.3A CN107862005A (en) | 2017-10-25 | 2017-10-25 | User view recognition methods and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107862005A true CN107862005A (en) | 2018-03-30 |
Family
ID=61696342
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711005661.3A Pending CN107862005A (en) | 2017-10-25 | 2017-10-25 | User view recognition methods and device |
Country Status (3)
| Country | Link |
|---|---|
| CN (1) | CN107862005A (en) |
| TW (1) | TWI700632B (en) |
| WO (1) | WO2019080661A1 (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108597503A (en) * | 2018-05-09 | 2018-09-28 | 科大讯飞股份有限公司 | Testing material generation method, device, equipment and read-write storage medium |
| CN109635271A (en) * | 2018-10-22 | 2019-04-16 | 阿里巴巴集团控股有限公司 | A kind of user's intension recognizing method, customer service system, device and electronic equipment |
| WO2019080661A1 (en) * | 2017-10-25 | 2019-05-02 | 阿里巴巴集团控股有限公司 | Method and device for identifying intention of user |
| CN110008308A (en) * | 2019-01-24 | 2019-07-12 | 阿里巴巴集团控股有限公司 | For the method and apparatus of user's question sentence supplemental information |
| CN110231985A (en) * | 2019-06-17 | 2019-09-13 | 三角兽（北京）科技有限公司 | Operation flow data processing method, device, electronic equipment and storage medium |
| CN110489635A (en) * | 2018-05-15 | 2019-11-22 | 阿里巴巴集团控股有限公司 | Data object search control method, apparatus and system |
| CN110517672A (en) * | 2018-05-03 | 2019-11-29 | 阿里巴巴集团控股有限公司 | User's intension recognizing method, method for executing user command, system and equipment |
| CN110765759A (en) * | 2019-10-21 | 2020-02-07 | 普信恒业科技发展（北京）有限公司 | Intention identification method and device |
| CN111368045A (en) * | 2020-02-21 | 2020-07-03 | 平安科技（深圳）有限公司 | User intention identification method, device, equipment and computer readable storage medium |
| CN111553162A (en) * | 2020-04-28 | 2020-08-18 | 腾讯科技（深圳）有限公司 | Intention identification method and related device |
| WO2020211008A1 (en) * | 2019-04-17 | 2020-10-22 | 深圳市欢太科技有限公司 | Speech recognition method and apparatus, storage medium and electronic device |
| CN111930854A (en) * | 2020-10-10 | 2020-11-13 | 北京福佑多多信息技术有限公司 | Intention prediction method and device |
| CN112328763A (en) * | 2020-11-04 | 2021-02-05 | 北京京东尚科信息技术有限公司 | Intention recognition method, device, dialogue method and system |
| WO2021077834A1 (en) * | 2019-10-23 | 2021-04-29 | 支付宝(杭州)信息技术有限公司 | Dialogue system-based method for asking follow-up question in response to user question, and device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8903924B2 (en) * | 2011-12-09 | 2014-12-02 | International Business Machines Corporation | Aggregating data in electronic communications |
| CN105868179A (en) * | 2016-04-08 | 2016-08-17 | 上海智臻智能网络科技股份有限公司 | Intelligent asking-answering method and device |
| CN106528759A (en) * | 2016-11-04 | 2017-03-22 | 上海智臻智能网络科技股份有限公司 | Information processing method and device for intelligent question-answering system |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103049433B (en) * | 2012-12-11 | 2015-10-28 | 微梦创科网络科技（中国）有限公司 | The method of automatic question-answering method, automatically request-answering system and structure question and answer case library |
| CN103049548B (en) * | 2012-12-27 | 2016-08-10 | 科大讯飞股份有限公司 | FAQ in electronic channel application identifies system and method |
| CN104050256B (en) * | 2014-06-13 | 2017-05-24 | 西安蒜泥电子科技有限责任公司 | Initiative study-based questioning and answering method and questioning and answering system adopting initiative study-based questioning and answering method |
| US9959355B2 (en) * | 2015-08-31 | 2018-05-01 | International Business Machines Corporation | Associating related threads in a question and answer session |
| CN107862005A (en) * | 2017-10-25 | 2018-03-30 | 阿里巴巴集团控股有限公司 | User view recognition methods and device |
-
2017
- 2017-10-25 CN CN201711005661.3A patent/CN107862005A/en active Pending
-
2018
- 2018-08-24 TW TW107129570A patent/TWI700632B/en active
- 2018-09-12 WO PCT/CN2018/105192 patent/WO2019080661A1/en active Application Filing
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8903924B2 (en) * | 2011-12-09 | 2014-12-02 | International Business Machines Corporation | Aggregating data in electronic communications |
| CN105868179A (en) * | 2016-04-08 | 2016-08-17 | 上海智臻智能网络科技股份有限公司 | Intelligent asking-answering method and device |
| CN106528759A (en) * | 2016-11-04 | 2017-03-22 | 上海智臻智能网络科技股份有限公司 | Information processing method and device for intelligent question-answering system |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019080661A1 (en) * | 2017-10-25 | 2019-05-02 | 阿里巴巴集团控股有限公司 | Method and device for identifying intention of user |
| CN110517672B (en) * | 2018-05-03 | 2023-08-15 | 阿里巴巴集团控股有限公司 | User intention recognition method, user instruction execution method, system and equipment |
| CN110517672A (en) * | 2018-05-03 | 2019-11-29 | 阿里巴巴集团控股有限公司 | User's intension recognizing method, method for executing user command, system and equipment |
| CN108597503A (en) * | 2018-05-09 | 2018-09-28 | 科大讯飞股份有限公司 | Testing material generation method, device, equipment and read-write storage medium |
| CN108597503B (en) * | 2018-05-09 | 2021-04-30 | 科大讯飞股份有限公司 | Test corpus generation method, device and equipment and readable and writable storage medium |
| CN110489635A (en) * | 2018-05-15 | 2019-11-22 | 阿里巴巴集团控股有限公司 | Data object search control method, apparatus and system |
| CN110489635B (en) * | 2018-05-15 | 2023-05-05 | 阿里巴巴集团控股有限公司 | Data object search control method, device and system |
| CN109635271A (en) * | 2018-10-22 | 2019-04-16 | 阿里巴巴集团控股有限公司 | A kind of user's intension recognizing method, customer service system, device and electronic equipment |
| CN110008308A (en) * | 2019-01-24 | 2019-07-12 | 阿里巴巴集团控股有限公司 | For the method and apparatus of user's question sentence supplemental information |
| CN110008308B (en) * | 2019-01-24 | 2023-06-02 | 创新先进技术有限公司 | Method and device for supplementing information for user question |
| WO2020211008A1 (en) * | 2019-04-17 | 2020-10-22 | 深圳市欢太科技有限公司 | Speech recognition method and apparatus, storage medium and electronic device |
| CN113348502A (en) * | 2019-04-17 | 2021-09-03 | 深圳市欢太科技有限公司 | Voice recognition method, voice recognition device, storage medium and electronic equipment |
| CN110231985A (en) * | 2019-06-17 | 2019-09-13 | 三角兽（北京）科技有限公司 | Operation flow data processing method, device, electronic equipment and storage medium |
| CN110765759A (en) * | 2019-10-21 | 2020-02-07 | 普信恒业科技发展（北京）有限公司 | Intention identification method and device |
| WO2021077834A1 (en) * | 2019-10-23 | 2021-04-29 | 支付宝(杭州)信息技术有限公司 | Dialogue system-based method for asking follow-up question in response to user question, and device |
| CN111368045A (en) * | 2020-02-21 | 2020-07-03 | 平安科技（深圳）有限公司 | User intention identification method, device, equipment and computer readable storage medium |
| CN111368045B (en) * | 2020-02-21 | 2024-05-07 | 平安科技（深圳）有限公司 | User intention recognition method, device, equipment and computer readable storage medium |
| CN111553162A (en) * | 2020-04-28 | 2020-08-18 | 腾讯科技（深圳）有限公司 | Intention identification method and related device |
| CN111553162B (en) * | 2020-04-28 | 2023-09-22 | 腾讯科技（深圳）有限公司 | Intention recognition method and related device |
| CN111930854A (en) * | 2020-10-10 | 2020-11-13 | 北京福佑多多信息技术有限公司 | Intention prediction method and device |
| CN112328763A (en) * | 2020-11-04 | 2021-02-05 | 北京京东尚科信息技术有限公司 | Intention recognition method, device, dialogue method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| TW201917601A (en) | 2019-05-01 |
| WO2019080661A1 (en) | 2019-05-02 |
| TWI700632B (en) | 2020-08-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107862005A (en) | User view recognition methods and device | |
| WO2021120631A1 (en) | Intelligent interaction method and apparatus, and electronic device and storage medium | |
| CN107492379B (en) | Voiceprint creating and registering method and device | |
| CN113591902B (en) | Cross-modal understanding and generating method and device based on multi-modal pre-training model | |
| JP6793975B2 (en) | Video-based Jobs Job Matching Servers and methods and computer-readable recording media containing programs to perform those methods | |
| US11106983B2 (en) | Intelligent interaction method and intelligent interaction system | |
| CN106373575B (en) | User voiceprint model construction method, device and system | |
| CN109101545A (en) | Natural language processing method, apparatus, equipment and medium based on human-computer interaction | |
| CN107329967A (en) | Question answering system and method based on deep learning | |
| CN108021934B (en) | Method and device for recognizing multiple elements | |
| CN107704453A (en) | A kind of word semantic analysis, word semantic analysis terminal and storage medium | |
| CN110222145A (en) | A kind of intelligence law appraisal procedure and system | |
| CN113515988B (en) | Palm print recognition method, feature extraction model training method, device and medium | |
| CN111666400B (en) | Message acquisition method, device, computer equipment and storage medium | |
| CN109543012A (en) | A kind of user's intension recognizing method and device based on Word2Vec and RNN | |
| CN110020059A (en) | System and method for inclusiveness CAPTCHA | |
| CN110399473B (en) | Method and device for determining answers to user questions | |
| CN109447273A (en) | Model training method, advertisement recommended method, relevant apparatus, equipment and medium | |
| CN112434501A (en) | Work order intelligent generation method and device, electronic equipment and medium | |
| CN110347786A (en) | A kind of tuning method and system of semantic model | |
| CN111128182A (en) | Intelligent voice recording method and device | |
| CN106844344A (en) | For the contribution degree computational methods and subject extraction method and system talked with | |
| CN110704618A (en) | Method and device for determining standard problem corresponding to dialogue data | |
| CN111625636B (en) | Method, device, equipment and medium for rejecting man-machine conversation | |
| CN110047473B (en) | Man-machine cooperative interaction method and system |
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 | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1249785 Country of ref document: HK |
|
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180330 |
|
| RJ01 | Rejection of invention patent application after publication |