CN109544195A - A kind of information processing method and electronic equipment - Google Patents
A kind of information processing method and electronic equipment Download PDFInfo
- Publication number
- CN109544195A CN109544195A CN201811581667.XA CN201811581667A CN109544195A CN 109544195 A CN109544195 A CN 109544195A CN 201811581667 A CN201811581667 A CN 201811581667A CN 109544195 A CN109544195 A CN 109544195A
- Authority
- CN
- China
- Prior art keywords
- session
- information
- corresponding relationship
- classification
- attribute
- 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
- G06Q30/01—Customer relationship services
Abstract
This application provides a kind of information processing method and electronic equipments, applied in intelligent customer service conversational system, the intelligent customer service conversational system can respond to automatically provide feedback information the input information received, wherein, the information processing method comprises determining that dialogue state round to intelligent customer service conversational system end, obtains session information；Determine that session resets degree based on the session information；Determine that the session resetting degree meets the first preset condition, respond the reset operation to current sessions, it can be seen that, intelligent customer service conversational system in the application can dialogue-based information automated to respond to the reset operation to current sessions, it is reset manually without user, the operating burden for alleviating user, improves user experience.
Description
Technical field
The present invention relates to technical field of information processing, more particularly to a kind of information processing method and electronics are set
It is standby.
Background technique
In intelligent conversational system, system can engage in the dialogue with user, and provide relevant feedback.
And in system and during user session, often will appear system can not correct understanding user input the standard of information
The problem of being really intended to will lead to session termination or user caused to dislike if systems stay issues inquiry message to user.
Under in response to this, in existing intelligence conversational system, can be conversated reset operation by user, as user clicks " weight
Setting " button or user input resetting session sentence, and the session between system and user can be allowed to open again in this way
Begin.
But by user conversate reset operation mode will increase user's operation burden, reduce user experience.
Summary of the invention
In view of this, the present invention provides a kind of information processing method and electronic equipment, to solve the above technical problems.
To achieve the above object, the invention provides the following technical scheme:
A kind of information processing method is applied in intelligent customer service conversational system, and the intelligent customer service conversational system can be right
The input information received is responded to automatically provide feedback information, and the information processing method includes:
Determine that dialogue state round to intelligent customer service conversational system end, obtains session information；
Determine that session resets degree based on the session information；
It determines that the session resetting degree meets the first preset condition, responds the reset operation to current sessions.
It is preferably, described to determine that session resets degree based on the session information, comprising:
Current sessions attribute possessed by different sessions classification is determined based on the session information；
Determine that session resets degree based on the current sessions attribute.
It is preferably, described to determine that session resets degree based on the current sessions attribute, comprising:
Target information numerical value corresponding with current sessions attribute is determined based on the first corresponding relationship；Described first corresponding pass
System is for characterizing the different sessions attribute under same session classification and the corresponding relationship of different information values；
Weighted value corresponding with different sessions classification is determined based on the second corresponding relationship；Second corresponding relationship is used
In the corresponding relationship of characterization different sessions classification and different weighted values；
The information value of weighted value and the current sessions attribute based on different sessions classification determines that session is reset
Numerical value.
Preferably, the determination session resetting degree meets the first preset condition, comprising:
Determine that the session resetting numerical value is more than or equal to first threshold.
Preferably, further includes:
It determines that the session resetting numerical value is less than first threshold, is greater than second threshold, forbids the weight to the current sessions
Set operation；Wherein, the first threshold is greater than the second threshold.
Preferably, the first threshold is determined as follows:
Determine current sessions attribute possessed by the different sessions classification under preset session threshold value；
Target information numerical value corresponding with current sessions attribute is determined based on third corresponding relationship；The third is corresponding to close
System is for characterizing the different sessions attribute under same session classification and the corresponding relationship of different information values；
Weighted value corresponding with different sessions classification is determined based on the 4th corresponding relationship；4th corresponding relationship is used
In the corresponding relationship of characterization different sessions classification and different weighted values；
The information value and preset first of weighted value, the current sessions attribute based on different sessions classification
Coefficient determines the first threshold.
A kind of electronic equipment is applied in intelligent customer service conversational system, and the intelligent customer service conversational system can be to reception
To input information responded to automatically provide feedback information, the electronic equipment includes:
Memory, memory, for storing program；
Processor is for executing described program, and described program is for determining dialogue state round to the intelligent customer service meeting
Telephone system end obtains session information, determines that session resets degree based on the session information, determines the session resetting degree
Meet the first preset condition, responds the reset operation to current sessions.
Preferably, the memory is stored with the first corresponding relationship and the second corresponding relationship, and first corresponding relationship is used
Different sessions attribute and the corresponding relationship of different information values, second corresponding relationship under the same session classification of characterization
For characterizing the corresponding relationship of different sessions classification from different weighted values；
The processor executes described program and is specifically used for determining that different sessions classification is had based on the session information
Current sessions attribute, corresponding with current sessions attribute target information numerical value is determined based on first corresponding relationship, is based on
Second corresponding relationship determines weighted value corresponding with different sessions classification, the weighted value based on different sessions classification
And the information value of the current sessions attribute determines that session resets numerical value.
Preferably, the processor executes described program for determining that the session resetting degree meets the first default item
Part, comprising: determine that the session resetting numerical value is more than or equal to first threshold；
Wherein, the memory is also stored with third corresponding relationship and the 4th corresponding relationship, and the third corresponding relationship is used
Different sessions attribute and the corresponding relationship of different information values, the 4th corresponding relationship under the same session classification of characterization
For characterizing the corresponding relationship of different sessions classification from different weighted values；
The processor executes described program and is determined as follows the first threshold:
Determine current sessions attribute possessed by the different sessions classification under preset session threshold value, it is corresponding based on third
Relationship determines target information numerical value corresponding with current sessions attribute, based on the determination of the 4th corresponding relationship and different sessions classification
Corresponding weighted value, the information value of weighted value, the current sessions attribute based on different sessions classification and default
The first coefficient determine the first threshold.
A kind of electronic equipment is applied in intelligent customer service conversational system, and the intelligent customer service conversational system can be to reception
To input information responded to automatically provide feedback information, the electronic equipment includes:
Information unit is obtained, for determining that dialogue state round to intelligent customer service conversational system end, obtains session letter
Breath；
First determination unit, for determining that session resets degree based on the session information；
First response unit is responded for determining that the session resetting degree meets the first preset condition to current sessions
Reset operation.
It can be seen via above technical scheme that compared with prior art, the embodiment of the invention provides a kind of information processings
Method is applied in intelligent customer service conversational system, which can ring the input information received
It should be to automatically provide feedback information, specifically, determining dialogue state round to intelligent customer service session by obtaining session information
System end determines that session resets degree based on the session information, determines that the session resetting degree meets the first default item
Part responds the reset operation to current sessions, it can be seen that, the intelligent customer service conversational system in the application can be dialogue-based
Information automated tos respond to the reset operation to current sessions, resets manually without user, alleviates the operating burden of user,
Improve user experience.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, below will to embodiment or
Attached drawing needed to be used in the description of the prior art is briefly described, it should be apparent that, the accompanying drawings in the following description is only
The embodiment of the present invention for those of ordinary skill in the art without creative efforts, can be with
Other attached drawings are obtained according to the attached drawing of offer.
Fig. 1 is a kind of flow diagram for information processing method that the application embodiment of the method one provides；
Fig. 2 is a kind of flow diagram for information processing method that the application embodiment of the method two provides；
Fig. 3 is a kind of flow diagram for information processing method that the application embodiment of the method three provides；
Fig. 4 is the flow diagram that the first threshold that the application embodiment of the method four provides determines method；
Fig. 5 is the structural schematic diagram for a kind of electronic equipment that the application Installation practice one provides；
Fig. 6 is the structural schematic diagram for a kind of electronic equipment that the application Installation practice five provides.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts it is all its
His embodiment, shall fall within the protection scope of the present invention.
The application embodiment of the method one provides a kind of information processing method, is applied in intelligent customer service conversational system, should
Intelligent customer service conversational system can respond to automatically provide feedback information the input information received.That is, working as
After user inputs information to intelligent customer service conversational system, intelligent customer service conversational system can be rung based on the user's input information
It answers, to provide the feedback information about the input information.As it can be seen that intelligent customer service conversational system can simulate true people, so that with
Family can with there is the impression chatted with true people in intelligent customer service conversational system.
As shown in Figure 1, this method comprises the following steps:
Step 101: determining that dialogue state round to intelligent customer service conversational system end, obtains session information；
For user during engaging in the dialogue with intelligent customer service conversational system, dialogue state will do it round, for example, user
First propose one information of a problem or description, then intelligent customer service system can receive the input information of user, to the input
Information carries out response to export feedback information, waits user for the response of the feedback information or inputting again for user.
So, determine that dialogue state round can be characterized as intelligent customer service conversational system termination to intelligent customer service conversational system end and receive
State or intelligent customer service conversational system end when inputting information export the state before feedback information.
Intelligent customer service conversational system obtains the session information during conversating with user, which at least wraps
When including dialogue state round to the intelligent customer service conversational system end, currently received input information can also include pair certainly
All session informations before speech phase round to intelligent customer service conversational system end.
Step 102: determining that session resets degree based on the session information；
Session resetting degree is for characterizing the degree for needing to be reset to current sessions, specifically, being based on the session
Information determines that session resets degree, may include following process:
(1) current sessions attribute possessed by different sessions classification is determined based on the session information；
Dividing in advance in intelligent customer service conversational system has for multiple session classifications of session resetting degree and per for a moment
Multiple session attributes under classification are talked about, so that carrying out processing to session information based on natural language understanding algorithm determines different meetings
Talk about possessed current sessions attribute under classification.
As a kind of implementation, session classification may include the other at least one of following conversation class:
The current sessions stage；
Active user's behavior；
Active user's feedback；
Active user's emotion；
User's history feedback；
State randomness.
Wherein, the session attribute that the current session stage has includes but is not limited to: starting session attribute, orientation problem
Attribute solves question attributes, terminates session attribute, greeting and chat attribute.
The session attribute that active user's behavior has includes but is not limited to: greeting attribute, chats attribute, provides information
Attribute proposes question attributes, interpretation problems attribute.
The session attribute that active user's feedback has includes but is not limited to: agreeing to attribute, unknown properties, opposes attribute, is strong
Strong opposition attribute.
The session attribute that active user's emotion has includes but is not limited to: glad, normal, disappointed, indignation.
The session attribute that user's history feedback has includes but is not limited to: agreeing to, opposes.
State randomness is used to characterize session confusion degree between user and system, specifically, from initial state to complete
The path passed through at state is longer, repeat mode is more, so that state randomness is higher.Assuming that under certain session context, from proposition
Problem is to solving the problems, such as normally only to need d state transition, and during the real dialog of user and intelligent customer service conversational system
State transition y times, if that y<d, the session attribute of state randomness be it is normal, if y>=d, the meeting of state randomness
It is abnormal for talking about attribute.
It is currently received defeated when due to session information including dialogue state round to the intelligent customer service conversational system end
Enter information, therefore determines the current sessions attribute of different sessions classification based on currently received input information.And user's history is anti-
Feedback and state randomness need to believe using session all before dialogue state round to intelligent customer service conversational system end
Breath, if that the session letter that session information is all before including dialogue state round to intelligent customer service conversational system end
Breath can then determine the current sessions attribute of user's history feedback and the current sessions attribute of state randomness.
For example, user and the dialogue of intelligent customer service conversational system are as follows:
It is defeated that user inputs input1- system feedback intent1- user input input2- system feedback intent2- user
Entering input3- determines dialogue state round to intelligent customer service conversational system end-....
Session information includes at least the input3 of active user's input, can determine different sessions classification based on input3
Possessed current sessions attribute.Specifically, can determine the current sessions attribute in current sessions stage based on input3, such as
Fruit input3 is the chat information with intelligent customer service conversational system, then the current sessions attribute in current sessions stage is to greet
Chat attribute.It is also based on the current sessions attribute that input3 determines active user's behavior, as determined active user's behavior
Current sessions attribute is to chat attribute.It is also based on the current sessions attribute that input3 determines active user's feedback, if
Comprising the approval information for system feedback intent2 in input3, then can determine what input3 was fed back in active user
Current sessions attribute is to agree to attribute.It is also based on the current sessions attribute that input3 determines active user's emotion, if
Emotion when user inputs input3 is happiness, then can determine that the current sessions attribute of active user's emotion belongs to be glad
Property.
If session information session letter all before further including dialogue state round to intelligent customer service conversational system end
Breath, then be also based on session information determine user's history feedback current sessions attribute and state randomness it is current
Session attribute.
(2) determine that session resets degree based on the current sessions attribute.
After dialogue-based information has determined possessed current sessions attribute under different sessions classification, it can be based on working as
Preceding session attribute come determine session reset degree, specific method of determination the application without limitation, for example in advance be each session
Different sessions attribute under classification assigns different numerical value, and possessed current sessions attribute is total under calculating different sessions classification
The total value is reset degree by numerical value.
Step 103: determining that the session resetting degree meets the first preset condition, respond the resetting to current sessions and grasp
Make.
First preset condition can be set based on actual conditions, a threshold value such as be preset, if calculated
Total value is greater than preset threshold value, then responds the reset operation to current sessions.
It can be seen that, by obtaining session information, determining that dialogue state round arrives in the application embodiment of the method one
Intelligent customer service conversational system end determines that session resets degree based on the session information, determines that the session resetting degree meets
First preset condition responds the reset operation to current sessions, automated tos respond to realize dialogue-based information to current meeting
The reset operation of words resets manually without user, alleviates the operating burden of user, improve user experience.
The application embodiment of the method two provides a kind of information processing method, is based on current sessions attribute with detailed description
A kind of implementation for determining session resetting degree, as shown in Fig. 2, a kind of information processing method includes the following steps:
Step 201: determining that dialogue state round to intelligent customer service conversational system end, obtains session information；
Step 202: current sessions attribute possessed by different sessions classification is determined based on the session information；
Step 203: target information numerical value corresponding with current sessions attribute is determined based on the first corresponding relationship；
Step 204: weighted value corresponding with different sessions classification is determined based on the second corresponding relationship；
Wherein, the first corresponding relationship and the second corresponding relationship are pre-established in intelligent customer service conversational system.First is corresponding
Relationship is for characterizing the different sessions attribute under same session classification and the corresponding relationship of different information values, that is to say, that the
Record has information value corresponding with each session attribute under each session classification in one corresponding relationship.
Second corresponding relationship is used to characterize the corresponding relationship of different sessions classification from different weighted values, that is to say, that the
Record has the corresponding weighted value of each session classification in two corresponding relationships.
For example, the first corresponding relationship can be such that
The current session stage: starting session attribute -5, solves question attributes -3, terminates session category orientation problem attribute -4
Property -2, greet chat attribute -1.Wherein, the bigger expression of information value is more likely to cause confusion, is more likely to require resetting.
Active user's behavior: attribute -5 is greeted, attribute -4 is chatted, information attribute -3 is provided, proposes question attributes -2, solution
Release question attributes -1.
Active user's feedback: agree to attribute -0, unknown properties -1, oppose attribute -2, be strongly opposed to attribute -6.
Active user's emotion: happiness -0, normal -1, disappointed -2, angry -6.
User's history feedback: agree to number, oppose number.Wherein, the opinion if user's continuous several times are lodged an objection, says
Intention before bright system understands that problematic or dialogue state is chaotic, can specifically set according to the number continuously opposed
Corresponding information numerical value opposes that number is more, and information value is bigger, agrees to that number is more, information value is smaller.
State randomness: set from proposition problem to solve the problems, such as the state transition of normal need as d, user and system
State transition during real dialog is y, then state randomness s is as follows:
Second corresponding relationship can be such that
The current sessions stage -1
Active user's behavior -2
Active user's feedback -2
Active user's emotion -10
User's history feedback -5
State randomness -5
It should be noted that above-mentioned first corresponding relationship and the second corresponding relationship are only easy for the concrete example understood, this Shen
It please be not limited thereto.
Step 205: the information value of weighted value and the current sessions attribute based on different sessions classification determines
Session resets numerical value；
Specifically, the first calculation formula can be used to calculate session resetting numerical value, wherein the first calculation formula is as follows
It is shown:
Wherein, R indicates that session resets numerical value, aiIndicate the other weighted value of i-th of conversation class, piIndicate i-th of classification
Information value, n indicate the other number of conversation class.
For ease of understanding, it is briefly described with a specific example, wherein determining and current based on the first corresponding relationship
The corresponding target information numerical value of session attribute, and weight corresponding with different sessions classification is determined based on the second corresponding relationship
Numerical value is as shown in table 1 below:
Table 1
So, by above-mentioned first calculation formula it can be concluded that,
Session resets numerical value R=5*1+2*2+1*2+2*10+2*5+3*5=56
Step 206: determining that the session resetting numerical value is more than or equal to first threshold, respond the resetting to current sessions and grasp
Make.
Wherein, in the present embodiment, system can preset the specific value of first threshold.And the application its
He refers to the present processes example IV embodiments also describe other methods of determination of first threshold.
It can be seen that can reset numerical value in the application embodiment of the method two by session and reset journey to characterize session
Degree is responded when session resetting numerical value is greater than first threshold to the reset operations of current sessions, is automated toed respond to pair to realize
The reset operation of current sessions resets manually without user, alleviates the operating burden of user, improve user experience.
The application embodiment of the method three provides a kind of information processing method, as shown in figure 3, this method includes following step
It is rapid:
Step 301: determining that dialogue state round to intelligent customer service conversational system end, obtains session information；
Step 302: current sessions attribute possessed by different sessions classification is determined based on the session information；
Step 303: target information numerical value corresponding with current sessions attribute is determined based on the first corresponding relationship；
Step 304: weighted value corresponding with different sessions classification is determined based on the second corresponding relationship；
Step 305: the information value of weighted value and the current sessions attribute based on different sessions classification determines
Session resets numerical value；
Step 306: determining that the session resetting numerical value is more than or equal to first threshold, respond the resetting to current sessions and grasp
Make；
Step 307: determining that the session resetting numerical value is less than first threshold, be greater than second threshold, forbid to described current
The reset operation of session；
Wherein, the first threshold is greater than the second threshold.
Wherein, in the present embodiment, system can preset the specific value of first threshold and second threshold.And
The other embodiments of the application also describe other methods of determination of first threshold and second threshold, refer to the application
Embodiment of the method four.
It can be seen that can reset numerical value in the application embodiment of the method three by session and reset journey to characterize session
Degree is responded when session resetting numerical value is greater than first threshold to the reset operations of current sessions, is automated toed respond to pair to realize
The reset operation of current sessions resets manually without user, alleviates the operating burden of user, improve user experience；
And when session resetting numerical value is less than first threshold, greater than second threshold, forbid the reset operation to current sessions, to reduce
The error rate of reset operation.
In the present processes example IV, as shown in figure 4, first threshold is determined as follows:
Step 401: determining current sessions attribute possessed by the different sessions classification under preset session threshold value；
Not only dividing in advance in intelligent customer service conversational system has for multiple session classifications of session resetting degree and every
Multiple session attributes under one session classification, also dividing in advance has for multiple session classifications of session threshold value and per for a moment
Talk about multiple session attributes under classification.
It should be noted that resetting multiple session classifications of degree and the multiple conversation class for being directed to session threshold value for session
It is not exactly the same.
As a kind of implementation, the session classification for meeting threshold value may include following conversation class other at least one
Kind:
The current session stage；
User's tolerance；
User resets preference.
Wherein, the session attribute that the current session stage has includes but is not limited to: starting session attribute, orientation problem
Attribute solves question attributes, terminates session attribute, greeting and chat attribute.
The session attribute that user's tolerance has includes but is not limited to: very weak, weaker, general, relatively strong, very strong.
The session attribute that user resets preference and has includes but is not limited to: not liking, it doesn't matter, likes.
Wherein, the current sessions attribute that the current sessions attribute of user's tolerance and user reset preference can be based on
The historical session data statistics of the user to engage in the dialogue with intelligent customer service conversational system obtains.The current meeting in current session stage
Words attribute can use session information to determine.
Step 402: target information numerical value corresponding with current sessions attribute is determined based on third corresponding relationship；
Step 403: weighted value corresponding with different sessions classification is determined based on the 4th corresponding relationship；
Wherein, third corresponding relationship and the 4th corresponding relationship are pre-established in intelligent customer service conversational system.Third is corresponding
Relationship is for characterizing the different sessions attribute under same session classification and the corresponding relationship of different information values, that is to say, that the
Record has information value corresponding with each session attribute under each session classification in three corresponding relationships.
4th corresponding relationship is used to characterize the corresponding relationship of different sessions classification from different weighted values, that is to say, that the
Record has the corresponding weighted value of each session classification in four corresponding relationships.
For example, third corresponding relationship can be such that
The current session stage: starting session attribute -5, solves question attributes -3, terminates session category orientation problem attribute -4
Property -2, greet chat attribute -1.
Very weak -0, weaker -1, general -2, relatively strong by -3, very strong -4 user's tolerance:.
User restarts preference: not liking -0, it doesn't matter -1, likes -2.
4th corresponding relationship is as follows:
The current session stage -2
User's tolerance -2
User restarts preference -2.
It should be noted that above-mentioned third corresponding relationship and the 4th corresponding relationship are only easy for the concrete example understood, this Shen
It please be not limited thereto.
Step 404: the information value of weighted value, the current sessions attribute based on different sessions classification and pre-
If the first coefficient determine the first threshold.
First coefficient is preset, the coefficient used for calculating first threshold.
Specifically, the second calculation formula can be used to calculate first threshold, wherein the second calculation formula is as follows:
Wherein, U indicates first threshold, fiIndicate the other weighted value of i-th of conversation class, xiIndicate the letter of i-th classification
Numerical value is ceased, m indicates the other number of conversation class, and k is upper limit coefficient.It should be noted that E is chaotic coefficient, it is empirical value,
E can be added such as the second calculation formula when calculating first threshold, naturally it is also possible to not add E.
It should be noted that the method for determination of second threshold is referred to the method for determination of first threshold, with first threshold
Method of determination the difference is that only that the numerical value of k is different.When calculating second threshold, k is lower limit coefficient.Wherein, upper limit system
Several and lower limit coefficient is empirical value, for different conversational systems, different users, it will usually there is different setting.
Specific intelligence conversational system can be according to user behavior feedback correction coefficient k, to meet corresponding user.
For ease of understanding, it is briefly described with a specific example, referring to table 2:
Table 2
So, it can be derived that based on the second calculation formula,
First threshold U1=50+1.5* (5*1+2*2+1*2)=66.5
Second threshold U2=50+0.8* (5*1+2*2+1*2)=58.8
When session resets numerical value R >=U1, reset operation is carried out immediately；As U1 > R > U2, forbid to the current meeting
The reset operation of words.
Corresponding with a kind of above-mentioned information processing method, the embodiment of the present application also provides a kind of electronic equipment, lead to below
Several Installation practices are crossed to be described.
The application Installation practice one provides a kind of electronic equipment, is applied in intelligent customer service conversational system, the intelligence
Customer service conversational system can respond to automatically provide feedback information the input information received.That is, working as user
After inputting information to intelligent customer service conversational system, intelligent customer service conversational system can be responded based on the user's input information, with
Provide the feedback information about the input information.As it can be seen that intelligent customer service conversational system can simulate true people, enable a user to
It is enough with there is the impression chatted with true people in intelligent customer service conversational system.
As shown in figure 5, a kind of electronic equipment, including memory 100 and processor 200；
Wherein, memory 100, for storing program；
Processor 200 is for executing described program, and described program is for determining dialogue state round to the intelligent customer service
Conversational system end obtains session information, determines that session resets degree based on the session information, determines the session resetting journey
Degree meets the first preset condition, responds the reset operation to current sessions.
For user during engaging in the dialogue with intelligent customer service conversational system, dialogue state will do it round, for example, user
First propose one information of a problem or description, then intelligent customer service system can receive the input information of user, to the input
Information carries out response to export feedback information, waits user for the response of the feedback information or inputting again for user.
So, determine that dialogue state round can be characterized as intelligent customer service conversational system termination to intelligent customer service conversational system end and receive
State or intelligent customer service conversational system end when inputting information export the state before feedback information.
Intelligent customer service conversational system obtains the session information during conversating with user, which at least wraps
When including dialogue state round to the intelligent customer service conversational system end, currently received input information can also include pair certainly
All session informations before speech phase round to intelligent customer service conversational system end.
Session resetting degree is for characterizing the degree for needing to be reset to current sessions, specifically, processor executes institute
Program is stated for determining that session resets degree based on the session information, may include:
(1) current sessions attribute possessed by different sessions classification is determined based on the session information；
There is natural language understanding algorithm, divide has for session resetting degree in advance in intelligent customer service conversational system
Multiple session attributes under multiple session classifications and each session classification, thus based on natural language understanding algorithm to session
Information handle possessed current sessions attribute under determining different sessions classification.
As a kind of implementation, session classification may include the other at least one of following conversation class:
The current sessions stage；
Active user's behavior；
Active user's feedback；
Active user's emotion；
User's history feedback；
State randomness.
Wherein, the session attribute that the current session stage has includes but is not limited to: starting session attribute, orientation problem
Attribute solves question attributes, terminates session attribute, greeting and chat attribute.
The session attribute that active user's behavior has includes but is not limited to: greeting attribute, chats attribute, provides information
Attribute proposes question attributes, interpretation problems attribute.
The session attribute that active user's feedback has includes but is not limited to: agreeing to attribute, unknown properties, opposes attribute, is strong
Strong opposition attribute.
The session attribute that active user's emotion has includes but is not limited to: glad, normal, disappointed, indignation.
The session attribute that user's history feedback has includes but is not limited to: agreeing to, opposes.
State randomness is used to characterize session confusion degree between user and system, specifically, from initial state to complete
The path passed through at state is longer, repeat mode is more, so that status wheel unrest degree is higher.Assuming that under certain session context, from proposition
Problem is to solving the problems, such as normally only to need d state transition, and during the real dialog of user and intelligent customer service conversational system
State transition y times, if that y<d, the session attribute of state randomness be it is normal, if y>=d, the meeting of state randomness
It is abnormal for talking about attribute.
It is currently received defeated when due to session information including dialogue state round to the intelligent customer service conversational system end
Enter information, therefore determines the current sessions attribute of different sessions classification based on currently received input information.And user's history is anti-
Feedback and state randomness need to believe using session all before dialogue state round to intelligent customer service conversational system end
Breath, if that the session letter that session information is all before including dialogue state round to intelligent customer service conversational system end
Breath can then determine the current sessions attribute of user's history feedback and the current sessions attribute of state randomness.
(2) determine that session resets degree based on the current sessions attribute.
After dialogue-based information has determined possessed current sessions attribute under different sessions classification, it can be based on working as
Preceding session attribute come determine session reset degree, specific method of determination the application without limitation, for example in advance be each session
Different sessions attribute under classification assigns different numerical value, and possessed current sessions attribute is total under calculating different sessions classification
The total value is reset degree by numerical value.
First preset condition can be set based on actual conditions, a threshold value such as be preset, if calculated
Total value is greater than preset threshold value, then responds the reset operation to current sessions.
It can be seen that, by obtaining session information, determining that dialogue state round arrives in the application Installation practice one
Intelligent customer service conversational system end determines that session resets degree based on the session information, determines that the session resetting degree meets
First preset condition responds the reset operation to current sessions, automated tos respond to realize dialogue-based information to current meeting
The reset operation of words resets manually without user, alleviates the operating burden of user, improve user experience.
In the application Installation practice two, memory is stored with the first corresponding relationship and the second corresponding relationship.
Wherein, the first corresponding relationship is for characterizing the different sessions attribute under same session classification and different information values
Corresponding relationship, that is to say, that in the first corresponding relationship record have letter corresponding with each session attribute under each session classification
Cease numerical value.
Second corresponding relationship is used to characterize the corresponding relationship of different sessions classification from different weighted values, that is to say, that the
Record has the corresponding weighted value of each session classification in two corresponding relationships.
Processor is executed described program and worked as possessed by different sessions classification specifically for being determined based on the session information
Preceding session attribute determines target information numerical value corresponding with current sessions attribute based on first corresponding relationship, based on described
Second corresponding relationship determines corresponding with different sessions classification weighted value, weighted value based on different sessions classification and
The information value of the current sessions attribute determines that session resets numerical value.
Specifically, the first calculation formula can be used to calculate session resetting numerical value, wherein the first calculation formula is as follows
It is shown:
Wherein, R indicates that session resets numerical value, aiIndicate the other weighted value of i-th of conversation class, piIndicate i-th of classification
Information value, n indicate the other number of conversation class.
Correspondingly, processor executes described program for determining that the session resetting degree meets the first preset condition, packet
It includes: determining that the session resetting numerical value is more than or equal to first threshold.
In the Installation practice three of the application, processor executes described program and is also used to determine the session resetting number
Value is less than first threshold, is greater than second threshold, forbids the reset operation to the current sessions.
Wherein, the first threshold is greater than the second threshold.
In the Installation practice four of the application, memory is also stored with third corresponding relationship and the 4th corresponding relationship.
Wherein, third corresponding relationship is for characterizing the different sessions attribute under same session classification and different information values
Corresponding relationship, that is to say, that in third corresponding relationship record have letter corresponding with each session attribute under each session classification
Cease numerical value.
4th corresponding relationship is used to characterize the corresponding relationship of different sessions classification from different weighted values, that is to say, that the
Record has the corresponding weighted value of each session classification in four corresponding relationships.
The processor executes described program and is determined as follows the first threshold:
Determine current sessions attribute possessed by the different sessions classification under preset session threshold value, it is corresponding based on third
Relationship determines target information numerical value corresponding with current sessions attribute, based on the determination of the 4th corresponding relationship and different sessions classification
Corresponding weighted value, the information value of weighted value, the current sessions attribute based on different sessions classification and default
The first coefficient determine the first threshold.
Not only dividing in advance in intelligent customer service conversational system has for multiple session classifications of session resetting degree and every
Multiple session attributes under one session classification, also dividing in advance has for multiple session classifications of session threshold value and per for a moment
Talk about multiple session attributes under classification.
It should be noted that resetting multiple session classifications of degree and the multiple conversation class for being directed to session threshold value for session
It is not exactly the same.
First coefficient is preset, the coefficient used for calculating first threshold.
Specifically, the second calculation formula can be used to calculate first threshold, wherein the second calculation formula is as follows:
Wherein, U indicates first threshold, fiIndicate the other weighted value of i-th of conversation class, xiIndicate the letter of i-th classification
Numerical value is ceased, m indicates the other number of conversation class, and k is upper limit coefficient.It should be noted that E is chaotic coefficient, it is empirical value,
E can be added such as the second calculation formula when calculating first threshold, naturally it is also possible to not add E.
It should be noted that the method for determination of second threshold is referred to the method for determination of first threshold, with first threshold
Method of determination the difference is that only that the numerical value of k is different.When calculating second threshold, k is lower limit coefficient.Wherein, upper limit system
Several and lower limit coefficient is empirical value, for different conversational systems, different users, it will usually there is different setting.
Specific intelligence conversational system can be according to user behavior feedback correction coefficient k, to meet corresponding user.
The application Installation practice five additionally provides a kind of electronic equipment, is applied in intelligent customer service conversational system, the intelligence
Energy customer service conversational system can respond to automatically provide feedback information the input information received.That is, when using
After family inputs information to intelligent customer service conversational system, intelligent customer service conversational system can be responded based on the user's input information,
To provide the feedback information about the input information.As it can be seen that intelligent customer service conversational system can simulate true people, so that user
Can with there is the impression chatted with true people in intelligent customer service conversational system.
As shown in fig. 6, a kind of electronic equipment includes: to obtain information unit 601, the response of the first determination unit 602, first
Unit 603；Wherein:
Information unit 601 is obtained, for determining that dialogue state round to intelligent customer service conversational system end, obtains meeting
Talk about information；
First determination unit 602, for determining that session resets degree based on the session information；
Wherein, the first determination unit includes: the first determining module and the second determining module.
First determining module, for determining current sessions category possessed by different sessions classification based on the session information
Property；
Second determining module, for determining that session resets degree based on the current sessions attribute.
First response unit 603 is responded for determining that the session resetting degree meets the first preset condition to current
The reset operation of session.
In the application Installation practice six, the second determining module is specifically used for being determined and being worked as based on the first corresponding relationship
The corresponding target information numerical value of preceding session attribute determines weight number corresponding with different sessions classification based on the second corresponding relationship
Value, the information value of weighted value and the current sessions attribute based on different sessions classification determine that session resets numerical value.
Wherein, the first corresponding relationship is for characterizing the different sessions attribute under same session classification and different information values
Corresponding relationship.
Second corresponding relationship is used to characterize the corresponding relationship of different sessions classification from different weighted values.
Specifically, the first calculation formula can be used to calculate session resetting numerical value in the second determining module, wherein first
Calculation formula is as follows:
Wherein, R indicates that session resets numerical value, aiIndicate the other weighted value of i-th of conversation class, piIndicate i-th of classification
Information value, n indicate the other number of conversation class.
Correspondingly, the first response unit is specifically used for determining that the session resetting numerical value is more than or equal to first threshold, response
To the reset operation of current sessions.
In the application Installation practice seven, electronic equipment further includes first forbidding unit, for determining the session weight
Setting value is less than first threshold, is greater than second threshold, forbids the reset operation to the current sessions；Wherein, first threshold
Value is greater than the second threshold.
In the application Installation practice eight, electronic equipment further includes threshold value determination unit, for determining first threshold,
It is corresponding based on third specifically for current sessions attribute possessed by the different sessions classification under the preset session threshold value of determination
Relationship determines target information numerical value corresponding with current sessions attribute, based on the determination of the 4th corresponding relationship and different sessions classification
Corresponding weighted value；4th corresponding relationship is used to characterize the corresponding relationship of different sessions classification from different weighted values,
It is determined based on the weighted value of different sessions classification, the information value of the current sessions attribute and preset first coefficient
The first threshold.
Wherein, third corresponding relationship is for characterizing the different sessions attribute under same session classification and different information values
Corresponding relationship.4th corresponding relationship is used to characterize the corresponding relationship of different sessions classification from different weighted values.
Specifically, the second calculation formula can be used to calculate first threshold in threshold value determination unit, wherein second calculates
Formula is as follows:
Wherein, U indicates first threshold, fiIndicate the other weighted value of i-th of conversation class, xiIndicate the letter of i-th classification
Numerical value is ceased, m indicates the other number of conversation class, and k is upper limit coefficient.It should be noted that E is chaotic coefficient, it is empirical value,
E can be added such as the second calculation formula when calculating first threshold, naturally it is also possible to not add E.
It should be noted that threshold value determination unit can also determine second threshold by such as upper type, really with first threshold
Determining mode the difference is that only, the numerical value of k is different.When calculating second threshold, k is lower limit coefficient.Wherein, upper limit coefficient
It is empirical value with lower limit coefficient, for different conversational systems, different users, it will usually there is different setting.Tool
The intelligent conversational system of body can be according to user behavior feedback correction coefficient k, to meet corresponding user.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with its
The difference of his embodiment, the same or similar parts in each embodiment may refer to each other.For being filled disclosed in embodiment
For setting, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is referring to method portion
It defends oneself bright.
The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention.
Various modifications to these embodiments will be readily apparent to those skilled in the art, defined herein
General Principle can realize in other embodiments without departing from the spirit or scope of the present invention.Therefore, originally
Invention is not intended to be limited to the embodiments shown herein, and is to fit to special with principles disclosed herein and novelty
The consistent widest scope of point.
Claims (10)
1. a kind of information processing method is applied in intelligent customer service conversational system, the intelligent customer service conversational system can be docked
The input information received is responded to automatically provide feedback information, wherein the information processing method includes:
Determine that dialogue state round to intelligent customer service conversational system end, obtains session information；
Determine that session resets degree based on the session information；
It determines that the session resetting degree meets the first preset condition, responds the reset operation to current sessions.
2. described to determine that session resets degree based on the session information according to the method described in claim 1, wherein, comprising:
Current sessions attribute possessed by different sessions classification is determined based on the session information；
Determine that session resets degree based on the current sessions attribute.
3. it is described to determine that session resets degree based on the current sessions attribute according to the method described in claim 2, wherein,
Include:
Target information numerical value corresponding with current sessions attribute is determined based on the first corresponding relationship；First corresponding relationship is used for
Characterize different sessions attribute under same session classification and the corresponding relationship of different information values；
Weighted value corresponding with different sessions classification is determined based on the second corresponding relationship；Second corresponding relationship is for characterizing
The corresponding relationship of different sessions classification and different weighted values；
The information value of weighted value and the current sessions attribute based on different sessions classification determines that session resets numerical value.
4. the determination session resetting degree meets the first preset condition according to the method described in claim 3, wherein,
Include:
Determine that the session resetting numerical value is more than or equal to first threshold.
5. according to the method described in claim 3, wherein, further includes:
It determines that the session resetting numerical value is less than first threshold, is greater than second threshold, the resetting to the current sessions is forbidden to grasp
Make；Wherein, the first threshold is greater than the second threshold.
6. according to the method described in claim 4, wherein, the first threshold is determined as follows:
Determine current sessions attribute possessed by the different sessions classification under preset session threshold value；
Target information numerical value corresponding with current sessions attribute is determined based on third corresponding relationship；The third corresponding relationship is used for
Characterize different sessions attribute under same session classification and the corresponding relationship of different information values；
Weighted value corresponding with different sessions classification is determined based on the 4th corresponding relationship；4th corresponding relationship is for characterizing
The corresponding relationship of different sessions classification and different weighted values；
The information value of weighted value, the current sessions attribute based on different sessions classification and preset first coefficient are true
The fixed first threshold.
7. a kind of electronic equipment, it is applied in intelligent customer service conversational system, the intelligent customer service conversational system can be to receiving
Input information responded to automatically provide feedback information, wherein the electronic equipment includes:
Memory, memory, for storing program；
Processor is for executing described program, and described program is for determining dialogue state round to the intelligent customer service conversational system
End obtains session information, determines that session resets degree based on the session information, determines that the session resetting degree meets first
Preset condition responds the reset operation to current sessions.
8. according to the method described in claim 7, wherein, the memory is stored with the first corresponding relationship and the second corresponding pass
System, first corresponding relationship are used to characterize the different sessions attribute under same session classification and close from the corresponding of different information values
System, second corresponding relationship are used to characterize the corresponding relationship of different sessions classification from different weighted values；
The processor is executed described program and worked as possessed by different sessions classification specifically for being determined based on the session information
Preceding session attribute determines target information numerical value corresponding with current sessions attribute based on first corresponding relationship, based on described
Second corresponding relationship determines weighted value corresponding with different sessions classification, weighted value and institute based on different sessions classification
The information value for stating current sessions attribute determines that session resets numerical value.
9. according to the method described in claim 8, wherein, the processor executes described program for determining the session resetting
Degree meets the first preset condition, comprising: determines that the session resetting numerical value is more than or equal to first threshold；
Wherein, the memory is also stored with third corresponding relationship and the 4th corresponding relationship, and the third corresponding relationship is used for table
Different sessions attribute under same session classification and the corresponding relationship of different information values are levied, the 4th corresponding relationship is used for table
Levy the corresponding relationship of different sessions classification and different weighted values；
The processor executes described program and is determined as follows the first threshold:
Determine current sessions attribute possessed by the different sessions classification under preset session threshold value, it is true based on third corresponding relationship
Fixed target information numerical value corresponding with current sessions attribute determines power corresponding with different sessions classification based on the 4th corresponding relationship
Tuple value, the information value of weighted value, the current sessions attribute based on different sessions classification and preset first system
Number determines the first threshold.
10. a kind of electronic equipment, it is applied in intelligent customer service conversational system, the intelligent customer service conversational system can be to receiving
Input information responded to automatically provide feedback information, wherein the electronic equipment includes:
Information unit is obtained, for determining that dialogue state round to intelligent customer service conversational system end, obtains session information；
First determination unit, for determining that session resets degree based on the session information；
First response unit responds the weight to current sessions for determining that the session resetting degree meets the first preset condition
Set operation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811581667.XA CN109544195A (en) | 2018-12-24 | 2018-12-24 | A kind of information processing method and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811581667.XA CN109544195A (en) | 2018-12-24 | 2018-12-24 | A kind of information processing method and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109544195A true CN109544195A (en) | 2019-03-29 |
Family
ID=65856834
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811581667.XA Pending CN109544195A (en) | 2018-12-24 | 2018-12-24 | A kind of information processing method and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109544195A (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106685752A (en) * | 2016-06-28 | 2017-05-17 | 腾讯科技（深圳）有限公司 | Information processing method and terminal |
| CN107633042A (en) * | 2012-07-20 | 2018-01-26 | 韦韦欧股份有限公司 | The method and system of user view in search input is inferred in talking with interactive system |
-
2018
- 2018-12-24 CN CN201811581667.XA patent/CN109544195A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107633042A (en) * | 2012-07-20 | 2018-01-26 | 韦韦欧股份有限公司 | The method and system of user view in search input is inferred in talking with interactive system |
| CN106685752A (en) * | 2016-06-28 | 2017-05-17 | 腾讯科技（深圳）有限公司 | Information processing method and terminal |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105591882B (en) | A kind of intelligence machine person to person mixes the method and system of customer service | |
| CN110096191B (en) | Man-machine conversation method and device and electronic equipment | |
| US10477026B2 (en) | System for routing interactions using bio-performance attributes of persons as dynamic input | |
| US10554594B2 (en) | Method and system for automatic switching between chat windows | |
| US20190260873A1 (en) | Media forked application system in call centers | |
| US20080040431A1 (en) | Monitoring and responding to instant messaging user activity | |
| Tezcan et al. | Routing and staffing in customer service chat systems with impatient customers | |
| US8458252B2 (en) | Minimizing the time required to initiate and terminate an instant messaging session | |
| US20040062363A1 (en) | Third party coaching for agents in a communication system | |
| US20180225279A1 (en) | Contextual fit determination for proposed messages | |
| WO2019201202A1 (en) | Call handling method and apparatus, server, storage medium, and system | |
| US9077699B1 (en) | Text chat | |
| CN113810265B (en) | System and method for message insertion and guidance | |
| US9191218B1 (en) | Participant reputation in a group communication environment | |
| CN107092196A (en) | The control method and relevant device of intelligent home device | |
| CN110570215A (en) | Intelligent customer service system | |
| JP7399505B2 (en) | Communication support server, communication support system, communication support method, and communication support program | |
| US11776546B1 (en) | Intelligent agent for interactive service environments | |
| CN109544195A (en) | A kind of information processing method and electronic equipment | |
| CN110020007A (en) | Robot interactive control method and device, storage medium, server | |
| CN104579933B (en) | A kind of user management method and system | |
| CN111191019A (en) | Information processing method, electronic equipment and information processing system | |
| CN111047362A (en) | Statistical management method and system for use activity of intelligent sound box | |
| CN110852115A (en) | Synchronous translation system and method based on artificial intelligence and real person translation | |
| WO2022001161A1 (en) | Online interview 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 |