CN106126516A - Man-machine interaction method and system - Google Patents
Man-machine interaction method and system Download PDFInfo
- Publication number
- CN106126516A CN106126516A CN201610374251.5A CN201610374251A CN106126516A CN 106126516 A CN106126516 A CN 106126516A CN 201610374251 A CN201610374251 A CN 201610374251A CN 106126516 A CN106126516 A CN 106126516A
- Authority
- CN
- China
- Prior art keywords
- user
- answer
- knowledge base
- satisfaction
- feedback
- 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 OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2453—Query optimisation
- G06F16/24534—Query rewriting; Transformation
- G06F16/24537—Query rewriting; Transformation of operators
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of man-machine interaction method, including: obtain problem and the user type of user；All problems from knowledge base retrieves the problem maximum with first problem matching degree as Second Problem；According to the user feedback satisfaction answering corresponding first user type of Second Problem in knowledge base, selecting the answer of Second Problem in knowledge base, obtain answers the answer as first problem, and i.e. first answers；Answer first and send to described user.Above-mentioned man-machine interaction method, when machine is talked with user, relevant to customer problem and that user feedback is good answer is searched out from off-line or online corpus or chat data storehouse, take full advantage of the dissimilar user feedback to the answer of conventional same or like problem, achieve the machine prediction to user preferences, there is preferable experience property and the suitability.The invention also discloses a kind of man-machine interactive system.
Description
Technical field
The present invention relates to computer and field of artificial intelligence, particularly relate to a kind of man-machine interaction method and system.
Background technology
Along with the continuous expansion of the technology such as the Internet and sensing, machine function also becomes to become stronger day by day, meanwhile, and machine
Also become the most various with the demand of the exchange and interdynamic mode of user.In prior art, when machine and user talk, typically
It is to search out the answer relevant to customer problem from off-line or online corpus or chat data storehouse.
Although, it is relatively reasonable for the problem that user is asked by such answer, but is not necessarily in accordance with user's
Emotion, namely it is not necessarily the answer of user preferences.Because machine does not accounts for user and to answer is in prior art
No satisfaction, does not make full use of user type and the user feedback to the answer of former same or like problem.So, mesh
Before, machine of the prior art can not meet for user psychology, the answer of hobby, it is impossible to realizes pre-to user preferences of machine
Survey, more cannot realize the family inclination that comes into operation, not there is good experience property and the suitability.
Summary of the invention
Based on this, it is necessary to provide a kind of can be accurate according to user's feedback to the answer of former same or like problem
Really, man-machine interaction method and the system of customer problem is answered on efficient analysis ground.
A kind of man-machine interaction method, comprises the following steps:
Obtaining problem and the user type of user, wherein, the problem of the user of acquisition is as first problem, the user type of acquisition
As first user type；
All problems from knowledge base retrieves the problem maximum with described first problem matching degree as Second Problem；
According to the user feedback satisfaction answering corresponding described first user type of Second Problem described in knowledge base, to knowing
Knowing the answer of Second Problem described in storehouse to select, obtain answers as the answer to described first problem, i.e. first time
Answer；
Answer described first and send to described user.
Wherein in an embodiment, also include: be pre-created described knowledge base；
Wherein, described knowledge base includes: at least one problem；
At least one of each described problem is answered；
At least one user type corresponding with each described answer；And
The user feedback satisfaction of every kind of described user type.
Wherein in an embodiment, also include:
Obtain user's satisfaction to described first answer, as first user satisfaction feedback；
Described knowledge base is updated according to described first user satisfaction feedback；
By by no less than the described first problem of predetermined number, described first answer, described first user type and described the
One user feedback satisfaction is added into described knowledge base, forms big data knowledge storehouse.
Wherein in an embodiment, update described knowledge base according to described first user satisfaction feedback and specifically include:
When described first problem and described Second Problem Incomplete matching, by described first problem, described first answer, described
First user type, described first user satisfaction feedback are added into described knowledge base；
When described first problem mate completely with described Second Problem and when in knowledge base with described Second Problem corresponding described in
In the presence of first user type, by described first user satisfaction feedback, and it is stored in described in described knowledge base second and asks
Described the first of topic answers the described user feedback satisfaction of corresponding described first user type, is weighted average computation,
The result calculated answers the user of corresponding described first user type as described first of the described Second Problem after updating
Satisfaction feedback；Or
When described first problem mate completely with described Second Problem and when in knowledge base with described Second Problem corresponding described in
First user type not in the presence of, using described first user satisfaction feedback as described the of the described Second Problem after updating
The one user feedback satisfaction answering corresponding described first user type.
Wherein in an embodiment, described corresponding described first of answering according to Second Problem described in knowledge base is used
The user feedback satisfaction of family type, the answer to Second Problem described in knowledge base selects, and the answer obtained is as right
The answer of described first problem, the i.e. first step answered specifically includes:
In the presence of described first user type corresponding with described Second Problem in knowledge base, from described Second Problem at least
One answer in select the answer that the described user feedback satisfaction of described first user type is the highest to answer as described first；
Or
When described first user type corresponding with described Second Problem in knowledge base not in the presence of, from described Second Problem to
A few answer select the described user corresponding with the immediate user type of attribute information of described first user type anti-
The answer that feedback satisfaction is the highest is answered as described first.
Wherein in an embodiment, described corresponding described first of answering according to Second Problem described in knowledge base is used
The user feedback satisfaction of family type, the answer to Second Problem described in knowledge base selects, and the answer obtained is as right
The answer of described first problem, the i.e. first step answered specifically includes:
In the presence of described first user type corresponding with described Second Problem in knowledge base, by described in described knowledge base
Each described user feedback satisfaction answering corresponding described first user type of two problems is normalized and obtains probability,
Answer, to this each, the answer selecting to obtain according to this probability, answer as described first；Or
When described first user type corresponding with described Second Problem in knowledge base not in the presence of, will be with institute in described knowledge base
State each of Second Problem and answer the institute that the immediate user type of attribute information of corresponding described first user type is corresponding
State user feedback satisfaction to be normalized and obtain probability, answer, to this each, the answer selecting to obtain according to this probability,
Answer as described first.
A kind of man-machine interactive system, including:
First acquisition module, for obtaining problem and the user type of user, wherein, the problem of the user of acquisition is asked as first
Topic, the user type of acquisition is as first user type；
Matching module, retrieves the problem maximum with described first problem matching degree in all problems from knowledge base and makees
For Second Problem；
Select module, anti-for the user answering corresponding described first user type according to Second Problem described in knowledge base
Feedback satisfaction, the answer to Second Problem described in knowledge base selects, and obtain answers as to described first problem
Answering, i.e. first answers；
Sending module, sends to described user for answering described first.
Wherein in an embodiment, also include: creation module, be used for being pre-created described knowledge base；
Wherein, described knowledge base includes: at least one problem；
At least one of each described problem is answered；
At least one user type corresponding with each described answer；And
The user feedback satisfaction of every kind of described user type.
Wherein in an embodiment, also include:
Second acquisition module, for obtaining user's satisfaction to described first answer, as first user satisfaction feedback；
More new module, for updating described knowledge base according to described first user satisfaction feedback；
Add module, for by by no less than the described first problem of predetermined number, described first answer, described first user
Type and described first user satisfaction feedback are added into described knowledge base, form big data knowledge storehouse.
Wherein in an embodiment, described more new module includes:
Adding device, for during when described first problem and described Second Problem Incomplete matching, by described first problem, described
First answer, described first user type, described first user satisfaction feedback are added into described knowledge base；
Computing unit, for mating completely and when asking with described second in knowledge base with described Second Problem when described first problem
In the presence of the described first user type that topic is corresponding, by described first user satisfaction feedback, and it is stored in described knowledge base
Described in the described first described user feedback satisfaction answering corresponding described first user type of Second Problem, add
Weight average calculates, and the result of calculating answers corresponding described first user as described first of the described Second Problem after updating
The user feedback satisfaction of type；Or
First signal generating unit, for mating completely with described Second Problem and work as in knowledge base when described first problem with described the
Described first user type corresponding to two problems not in the presence of, using described first user satisfaction feedback as described in after updating
Described the first of Second Problem answers the user feedback satisfaction of corresponding described first user type.
Wherein in an embodiment, described selection module includes:
First selects unit, is used in the presence of described first user type corresponding with described Second Problem in knowledge base, from
At least one of described Second Problem selects the highest the returning of described user feedback satisfaction of described first user type in answering
Answer and answer as described first；Or
Second select unit, for when described first user type corresponding with described Second Problem in knowledge base not in the presence of,
The immediate user class of attribute information with described first user type is selected from least one answer of described Second Problem
The answer that described user feedback satisfaction that type is corresponding is the highest is answered as described first.
Wherein in an embodiment, described selection module includes:
3rd selects unit, is used in the presence of described first user type corresponding with described Second Problem in knowledge base, will
The described user feedback satisfaction of the described first user type of each answer correspondence of Second Problem described in described knowledge base
It is normalized and obtains probability, answer, to this each, the answer selecting to obtain according to this probability, answer as described first；
Or
4th select unit, for when described first user type corresponding with described Second Problem in knowledge base not in the presence of,
By closest for the attribute information of the described first user type corresponding with each answer of Second Problem described in described knowledge base
Described user feedback satisfaction corresponding to user type be normalized and obtain probability, according to this probability answer this each into
Row selects the answer obtained, and answers as described first.
Above-mentioned man-machine interaction method and system, first obtain problem and the user type of user, wherein, the user's of acquisition
Problem is as first problem, and the user type of acquisition is as first user type；All problems from knowledge base retrieves
The problem maximum with first problem matching degree, as Second Problem, is used according to corresponding first of answering of Second Problem in knowledge base
The user feedback satisfaction of family type, selects the answer of Second Problem in knowledge base, and the answer obtained is as to first
The answer of problem, i.e. first answers, and answers first and sends to user.Above-mentioned man-machine interaction method and system, by machine and use
During the talk of family, search out relevant to customer problem from off-line or online corpus or chat data storehouse and user feedback is good
Answer, this mode takes full advantage of the dissimilar user feedback to the answer of conventional same or like problem, it is achieved that
The machine prediction to user preferences, has preferable experience property and the suitability.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of man-machine interaction method in an embodiment；
Fig. 2 is the schematic flow sheet of man-machine interaction method in another embodiment；
Fig. 3 is the structural representation of man-machine interactive system in an embodiment；
Fig. 4 is the structural representation of man-machine interactive system in another embodiment；
Fig. 5 is the structural representation of more new module in man-machine interactive system；And
Fig. 6 is the structural representation selecting module in man-machine interactive system.
Detailed description of the invention
In order to make the purpose of the present invention, technical scheme and advantage clearer, by the following examples, and combine attached
Figure, is further elaborated to the man-machine interaction method of the present invention and the detailed description of the invention of system.Should be appreciated that herein
Described specific embodiment, only in order to explain the present invention, is not intended to limit the present invention.
Seeing Fig. 1, in an embodiment, man-machine interaction method may comprise steps of:
Step S100, obtains problem and the user type of user, and wherein, the problem of the user of acquisition, as first problem, obtains
User type as first user type.Concrete, the problem obtaining user can be to key in asking of user by terminating machine
Topic, or management system can also be exchanged with machine, by selecting on this system by running user on terminating machine
And the problem of upload user, it is also possible to the problem obtaining user by voice transfer instruction.Thus, improve acquisition problem can
Can property and multiformity.
Same principle, the user type obtaining user can be to key in user type by terminating machine；Or can also
Management system is exchanged with machine, by carrying out the user type selected on this system by running user on terminating machine；And
The problem of upload user；The user that can also run on terminating machine exchanges in management system with machine and passes through user's registration information
Or photographic head picture automatically identifies user type；User type can also be obtained by voice transfer instruction.Thus, improve and obtain
Take safety and the multiformity of user type.
It is understood that wherein, by the problem of terminating machine key entry user and the terminating machine of user type can be
Computer or other problems that can carry out user and user type are keyed in, typing and the electric terminal equipment uploaded, example
Such as smart mobile phone, wearable intelligent equipment, panel computer etc..
Step S200, retrieves the problem maximum with first problem matching degree as the in all problems from knowledge base
Two problems.Wherein, knowledge base includes at least one problem；At least one of each problem is answered；Corresponding with each answer
At least one user type；And with the user feedback satisfaction of every kind of user type.Thus, improve problem in knowledge base,
Answer the relatedness between user feedback satisfaction and user type.Need it is further noted that in invention, knowledge base
For be pre-created.Thus, improve the suitability that first problem is carried out to be mated with the multiple problems in knowledge base.
Step S300, satisfied according to the user feedback answering corresponding first user type of Second Problem in knowledge base
Degree, selects the answer of Second Problem in knowledge base, and obtain answers as the answer to first problem, i.e. first time
Answer.
Step S400, answers first and sends to user.
Further, seeing Fig. 2, in an embodiment, man-machine interaction method can also comprise the following steps:
Step S500, obtains user's satisfaction to the first answer, as first user satisfaction feedback.
Step S600, according to first user satisfaction feedback more new knowledge base.Wherein, according to first user satisfaction feedback
More new knowledge base specifically includes:
When first problem and Second Problem Incomplete matching, by first problem, the first answer, first user type, the first use
Family satisfaction feedback is added into knowledge base；When first problem mate completely with Second Problem and when in knowledge base with Second Problem pair
In the presence of the first user type answered, by first user satisfaction feedback, and it is stored in knowledge base the first of Second Problem
Answer the user feedback satisfaction of corresponding first user type, be weighted average computation, after the result of calculating is as updating
The satisfaction feedback of user of first user type corresponding to the first answer of Second Problem；Or when first problem and second is asked
Topic mate completely and when first user type corresponding with Second Problem in knowledge base not in the presence of, by satisfied for first user feedback
Spend the user feedback satisfaction of the first user type corresponding as the first answer of the Second Problem after updating.Thus, improve
Upgraded in time the probability of first problem satisfaction feedback and the suitability by feedback.
Step S700, by by no less than predetermined number first problem, first answer, first user type and first
User feedback satisfaction is added into knowledge base, forms big data knowledge storehouse.
Additionally, in step S300, according to the user answering corresponding first user type of Second Problem in knowledge base
Satisfaction feedback, selects the answer of Second Problem in knowledge base, and obtain answers as the answer to first problem, i.e.
First step answered specifically includes: in the presence of first user type corresponding with Second Problem in knowledge base, ask from second
The answer that at least one answer of topic, the user feedback satisfaction of selection first user type is the highest is answered as first；Or work as
First user type corresponding with Second Problem in knowledge base not in the presence of, from Second Problem at least one answer select with
The highest answer of user feedback satisfaction corresponding to the immediate user type of attribute information of first user type is as first
Answer.
It addition, in step S300, according to the user answering corresponding first user type of Second Problem in knowledge base
Satisfaction feedback, selects the answer of Second Problem in knowledge base, and obtain answers as the answer to first problem, i.e.
First step answered, specifically includes: in the presence of first user type corresponding with Second Problem in knowledge base, by knowledge base
Each user feedback satisfaction answering corresponding first user type of middle Second Problem is normalized and obtains probability, according to
This probability answers, to this each, the answer selecting to obtain, and answers as first；Or when corresponding with Second Problem in knowledge base
First user type not in the presence of, the attribute of corresponding first user type will be answered with each of Second Problem in knowledge base
User feedback satisfaction corresponding to the immediate user type of information is normalized and obtains probability, according to this probability to this each
Answer the answer carrying out selecting to obtain, answer as first.Thus, improve the answer of knowledge based storehouse acquisition first problem
Accuracy and multiformity.
Wherein, at least one problem in knowledge base and the one or more answers corresponding with at least one problem are permissible
Obtain from off-line or online corpus or chat data storehouse.Thus, improve acquisition matching problem and relative with problem
Accuracy, multiformity and the suitability that should answer.And further, the initial value of each user feedback satisfaction in knowledge base is such as
Fruit fails to obtain, and can be preset as identical numerical value.Thus, for subsequent calculations, update and obtain the answer for customer problem
Corresponding user feedback satisfaction improves convenience.
It should be understood that knowledge base is feedback knowledge base in the embodiment of the present invention, feedback knowledge base is to tie in knowledge engineering
Structure, easily operates, and easily utilizes, comprehensive organized knowledge cluster, is the needs solved for a certain (or some) field question,
Use that certain (or some) knowledge representation mode stores in computer storage, organizes, manages and use interknits
Knowledge sheet set.Such as, relevant in the artificial intelligence field in computer definition, theorem and algorithm and common-sense are known
Know.
Above-mentioned man-machine interaction method, first obtains problem and the user type of user, and wherein, the problem of the user of acquisition is made
For first problem, the user type of acquisition is as first user type；Again all problems from knowledge base retrieves and the
The problem of one problem matching degree maximum is as Second Problem；Then use according to corresponding first of answering of Second Problem in knowledge base
The user feedback satisfaction of family type, selects the answer of Second Problem in knowledge base, and the answer obtained is as to first
The answer of problem, i.e. first answers；First answer sends extremely and user the most at last.Above-mentioned man-machine interaction method, it is achieved that at machine
When device and user talk, from off-line or online corpus or chat data storehouse, search out and user relevant to customer problem
The answer fed back, this mode makes full use of the dissimilar user feedback to the answer of conventional same or like problem, real
Show the machine prediction to user preferences, there is preferable experience property and the suitability.
For a kind of man-machine interaction method being better understood from apply the present invention to propose, carry out the example below, need
Bright, the scope that the present invention is protected does not limits to the example below.
Such as, obtain the problem of user and user profile, wherein, the problem of the user of acquisition as first problem, i.e. the
One problem is: have a bath every day to whether health is beneficial to；Obtain user type as first user type, i.e. first user class
Type is: 29 years old age, women, foreign enterprise office worker, and fertility cycle is 8 weeks.Further, by first problem, (every day has a bath to health
Whether it is beneficial to) mate with the multiple problems being pre-stored in knowledge base.Wherein, in knowledge base pre-stored with first
The problem that problem is associated includes but not limited to: have a bath every day to are you fine, water temperature of having a bath how much spend more helpful to health,
How long the time every time having a bath must not exceed and have a bath every day the most beneficial etc. to health.
It should be noted that the key word in extraction first problem.Such as, every day, have a bath, health, benefit therein
Individual or several random conduct combinations, scan in the multiple problems in being pre-stored in knowledge base, finally search and first
The problem that in problem, all key words all mate or matching degree is the highest, as Second Problem.I.e. in this example, in knowledge base
The problem searched and the highest problem of first problem matching degree obtaining user, i.e. have a bath to are you fine every day, as the
Two problems.
Further, according to the user feedback satisfaction of the first user type of the answer of Second Problem, one is selected to return
Answering the answer of problem as acquired user, i.e. first answers.Such as, for Second Problem: have a bath every day to are you fine,
Answering one is: has a bath every day and is no advantage health, because having a bath the sebum to body skin and fat secretion the most frequently
Causing burden, and can wash drier and drier, the woman of period of gestation has a bath for every day, should be noted that water temperature, bathtime and
Safety etc. during having a bath.The user feedback satisfaction of above-mentioned answer is 98%；Answering two is: have a bath every day and health is not had benefit
Place, has a bath because out of season, such as, during health has wound or flu, is not suitable for all having a bath every day.On
The user feedback satisfaction stating answer is 60%；Answering three is: has a bath every day and benefits health, because good and clean is individual
People's hygienic habit needs to advocate and set an example by personally taking part to adhere to.The user feedback satisfaction of above-mentioned answer is 10%.
It is understood that the user satisfaction that answer one selected probability is answer one correspondence/(answer a correspondence
The user satisfaction of the user satisfaction of user satisfaction+answer two correspondences+answer three correspondences)；Answer two selected probability
For answering the user satisfaction of two correspondences/(answer the user satisfaction+answer of the user satisfaction of a correspondence+answer two correspondences
The user satisfaction of three correspondences)；Answering three selected probability is the user satisfaction/(answer a correspondence answering three correspondences
The user satisfaction of the user satisfaction of user satisfaction+answer two correspondences+answer three correspondences).
It is understood that as 29 years old age, women, foreign enterprise office worker, fertility cycle is that this kind of user type of 8 weeks exists
Time, select the answer that the user feedback satisfaction of above-mentioned user type is the highest as acquired from multiple answers of Second Problem
The answer of the problem of user, the user feedback satisfaction i.e. answering is 98%, the highest, i.e. selects to answer one as acquired use
The answer of the problem at family；When 29 years old age, women, foreign enterprise office worker, fertility cycle be 8 weeks this kind of user type not in the presence of,
The user feedback with the immediate user type of above-mentioned user type then can be selected from multiple answers of Second Problem satisfied
Spend the highest answer to answer as first.Thus, improve select with mate after the corresponding answer of problem accuracy with
Multiformity.Finally, send answering one to user as the first answer for first problem.
Based on above-mentioned the same principle, according to the user feedback satisfaction of the first user type of the answer of Second Problem,
Select an answer answering the problem as acquired user, also include: when 29 years old age, women, foreign enterprise office worker, conceived week
In the presence of phase is this kind of user type of 8 weeks, user feedback satisfaction based on above-mentioned user type is normalized and obtains generally
Rate, answers, to this each, the answer selecting to obtain according to this probability, answers as first；When 29 years old age, women, foreign enterprise
Office worker, fertility cycle be 8 weeks this kind of user type not in the presence of, enter based on the user feedback satisfaction with above-mentioned user type
Row normalization obtains probability, answers, to each, the answer selecting to obtain according to this probability, answers as first.
Additionally, obtain user to first answer satisfaction, such as 96%, as the satisfaction feedback for first problem, i.e.
For first user satisfaction feedback, and according to answer corresponding with first problem in first user satisfaction feedback more new knowledge base
The satisfaction feedback of the type user that corresponding first user type is corresponding.In addition to above-mentioned mode, it is also possible to when first
When problem and Second Problem Incomplete matching, by first problem, corresponding answer, corresponding first user type, corresponding use
Family satisfaction feedback adds knowledge base；When first problem mate completely with Second Problem and when first user type (29 years old age,
Women, foreign enterprise office worker, fertility cycle is 8 weeks) in the presence of this kind of user type, by user to the first satisfaction answered, and
It is stored in the feedback of the type user corresponding to first user type corresponding to the first answer corresponding to Second Problem in knowledge base
Satisfaction, is weighted average computation, the result of calculating as the first answer that the Second Problem after updating is corresponding corresponding the
The satisfaction feedback of the type user that one user type is corresponding；Or when first problem mates with Second Problem and completely when first
User type (29 years old age, women, foreign enterprise office worker, fertility cycle is 8 weeks) not in the presence of, by first user satisfaction feedback make
First feedback of answering the type user corresponding to corresponding first user type corresponding for the Second Problem after updating is satisfied with
Degree.
Above-mentioned man-machine interaction method, machine is when talking with user, from off-line or online corpus or chat data
Storehouse searches out relevant to customer problem and that user feedback is good answer, the problem that user is not only asked by such answer
For be relatively rational, and be the emotion in accordance with user, be i.e. the answer liked of user, because machine is examined in the present invention
Consider to whether answer is satisfied with by user, take full advantage of anti-to the answer of former same or like problem of dissimilar user
Feedback, thus in the present invention machine understand user psychology it is known that user preferences which type of answer, it is achieved that machine to user like
Good prediction, it is possible to achieve come into operation family inclination.
Based on same inventive concept, in one embodiment, it is also proposed that a kind of man-machine interactive system.Seeing Fig. 3, this is man-machine
Interactive system 10 can include the first acquisition module 110, matching module 120, select module 130 and sending module 140.
Wherein, the first acquisition module 110 is for obtaining problem and the user type of user, and wherein, the user of acquisition asks
Topic is as first problem, and the user type of acquisition is as first user type；Matching module 120 is for owning from knowledge base
Problem retrieves the problem maximum with first problem matching degree as Second Problem；；Select module 130 for according to knowledge base
The satisfaction feedback of the user answering corresponding first user type of middle Second Problem, to the answer of Second Problem in knowledge base
Selecting, obtain answers as the answer to first problem, and i.e. first answers；Sending module 140 is for answering first
Send to user.
Further, seeing Fig. 4, in one embodiment, a kind of man-machine interactive system 10 can also include: creation module
150.Wherein, creation module 150 is used for being pre-created knowledge base, and knowledge base includes: at least one problem；Each problem is at least
One answer；At least one corresponding user type is answered with each；And with the user feedback satisfaction of every kind of user type.
In the present embodiment, knowledge base be pre-created being suitable for of improve that first problem carries out with the multiple problems in knowledge base mating
Property.
Additionally, see Fig. 4, in one embodiment, a kind of man-machine interactive system 10 can also include: the second acquisition module
160, more new module 170 and interpolation module 180.Wherein, the first answer is expired by the second acquisition module 160 for obtaining user
Meaning degree, as first user satisfaction feedback；More new module 170 is for according to first user satisfaction feedback more new knowledge base；
Add module 180 for by by first problem, the first answer, first user type and the first use no less than predetermined number
Family satisfaction feedback is added into knowledge base, forms big data knowledge storehouse.Thus, improve and upgraded in time first problem by feedback
The probability of satisfaction feedback and the suitability.
It addition, see Fig. 5, in one embodiment, a kind of man-machine interactive system updates mould 170 to farther include: add
Add unit 1701, computing unit 1702 and the first signal generating unit 1703.
Wherein, adding device 1701 for during when first problem and Second Problem Incomplete matching, by first problem, first
Answer, first user type, first user satisfaction feedback are added into knowledge base；Computing unit 1701 for when first problem with
Second Problem mates and completely in the presence of first user type corresponding with Second Problem in knowledge base, is fed back by first user
Satisfaction, and the user feedback being stored in the first user type of the first answer correspondence of Second Problem in knowledge base is satisfied
Degree, is weighted average computation, and the result of calculating is as the first user class of the first answer correspondence of the Second Problem after updating
The satisfaction feedback of the user of type；And first signal generating unit 1703 for when first problem mates completely with Second Problem and ought
First user type corresponding with Second Problem in knowledge base not in the presence of, using first user satisfaction feedback as update after
The first of Second Problem answers the satisfaction feedback of the type user corresponding to corresponding first user type.
It addition, see Fig. 6, in one embodiment, a kind of man-machine interactive system selects module 130 to farther include:
First selects unit 1301, second to select unit the 1302, the 3rd to select unit 1303 and the 4th to select unit 1304.
Wherein, first select unit 1301 for existing when first user type corresponding with Second Problem in knowledge base
Time, select the highest answer of user feedback satisfaction of first user type as the from least one of Second Problem is answered
One answers；Second select unit 1302 for when first user type corresponding with Second Problem in knowledge base not in the presence of, from
At least one of Second Problem selects the use that user type immediate with the attribute information of first user type is corresponding in answering
The answer that the feedback satisfaction at family is the highest is answered as first；3rd select unit 1303 for when in knowledge base with Second Problem pair
In the presence of the first user type answered, by anti-for the user of each first user type answering correspondence of Second Problem in knowledge base
Feedback satisfaction is normalized and obtains probability, answers the answer selecting to obtain, as first according to this probability to this each
Answer；4th select unit 1304 for when first user type corresponding with Second Problem in knowledge base not in the presence of, will be with
In knowledge base, the immediate user type of attribute information of each first user type answering correspondence of Second Problem is corresponding
User feedback satisfaction is normalized and obtains probability, answers, to this each, the answer selecting to obtain according to this probability, makees
It it is the first answer.Thus, improve accuracy and the multiformity of the answer of knowledge based storehouse acquisition first problem.
Above-mentioned man-machine interactive system, first passes through the first acquisition module 110 and obtains problem and the user type of user, its
In, the problem of the user of acquisition is as first problem, and the user type of acquisition is as first user type；Then by coupling mould
Block 120 retrieves the problem maximum with first problem matching degree as Second Problem in all problems from knowledge base, then leads to
Cross and select module 130 satisfied according to the type user feedback answering corresponding first user type of Second Problem in knowledge base
Degree, selects the answer of Second Problem in knowledge base, and obtain answers as the answer to first problem, i.e. first time
Answer, answer transmission to user finally by sending module 140 by first.The present embodiment achieves when machine is talked with user,
Relevant to customer problem and that user feedback is good answer is searched out from off-line or online corpus or chat data storehouse, this
One mode takes full advantage of the dissimilar user feedback to the answer of conventional same or like problem, it is achieved that machine is to user
The prediction of hobby, has preferable experience property and the suitability.
Embodiment described above only have expressed the several embodiments of the present invention, and it describes more concrete and detailed, but also
Therefore the restriction to the scope of the claims of the present invention can not be interpreted as.It should be pointed out that, for those of ordinary skill in the art
For, without departing from the inventive concept of the premise, it is also possible to make some deformation and improvement, these broadly fall into the guarantor of the present invention
Protect scope.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.
Claims (12)
1. a man-machine interaction method, it is characterised in that comprise the following steps:
Obtaining problem and the user type of user, wherein, the problem of the user of acquisition is as first problem, the user type of acquisition
As first user type；
All problems from knowledge base retrieves the problem maximum with described first problem matching degree as Second Problem；
According to the user feedback satisfaction answering corresponding described first user type of Second Problem described in knowledge base, to knowing
Knowing the answer of Second Problem described in storehouse to select, obtain answers as the answer to described first problem, i.e. first time
Answer；
Answer described first and send to described user.
Method the most according to claim 1, it is characterised in that also include: be pre-created described knowledge base；
Wherein, described knowledge base includes: at least one problem；
At least one of each described problem is answered；
At least one user type corresponding with each described answer；And
The user feedback satisfaction of every kind of described user type.
Method the most according to claim 1, it is characterised in that also include:
Obtain user's satisfaction to described first answer, as first user satisfaction feedback；
Described knowledge base is updated according to described first user satisfaction feedback；
By by no less than the described first problem of predetermined number, described first answer, described first user type and described the
One user feedback satisfaction is added into described knowledge base, forms big data knowledge storehouse.
Method the most according to claim 3, it is characterised in that know according to the renewal of described first user satisfaction feedback
Know storehouse to specifically include:
When described first problem and described Second Problem Incomplete matching, by described first problem, described first answer, described
First user type, described first user satisfaction feedback are added into described knowledge base；
When described first problem mate completely with described Second Problem and when in knowledge base with described Second Problem corresponding described in
In the presence of first user type, by described first user satisfaction feedback, and it is stored in described in described knowledge base second and asks
Described the first of topic answers the described user feedback satisfaction of corresponding described first user type, is weighted average computation,
The result calculated answers the user of corresponding described first user type as described first of the described Second Problem after updating
Satisfaction feedback；Or
When described first problem mate completely with described Second Problem and when in knowledge base with described Second Problem corresponding described in
First user type not in the presence of, using described first user satisfaction feedback as described the of the described Second Problem after updating
The one user feedback satisfaction answering corresponding described first user type.
Method the most according to claim 1, it is characterised in that the described answer according to Second Problem described in knowledge base
The user feedback satisfaction of corresponding described first user type, the answer to Second Problem described in knowledge base selects,
Obtain answers as the answer to described first problem, and the i.e. first step answered specifically includes:
In the presence of described first user type corresponding with described Second Problem in knowledge base, from described Second Problem at least
One answer in select the answer that the described user feedback satisfaction of described first user type is the highest to answer as described first；
Or
When described first user type corresponding with described Second Problem in knowledge base not in the presence of, from described Second Problem to
A few answer select the described user corresponding with the immediate user type of attribute information of described first user type anti-
The answer that feedback satisfaction is the highest is answered as described first.
Method the most according to claim 1, it is characterised in that the described answer according to Second Problem described in knowledge base
The user feedback satisfaction of corresponding described first user type, the answer to Second Problem described in knowledge base selects,
Obtain answers as the answer to described first problem, and the i.e. first step answered specifically includes:
In the presence of described first user type corresponding with described Second Problem in described knowledge base, by institute in described knowledge base
The described user feedback satisfaction of each the described first user type answering correspondence stating Second Problem is normalized and obtains
Probability, answers, to this each, the answer selecting to obtain according to this probability, answers as described first；Or
When described first user type corresponding with described Second Problem in described knowledge base not in the presence of, will be with described knowledge base
Described in each of Second Problem to answer the immediate user type of attribute information of corresponding described first user type corresponding
Described user feedback satisfaction be normalized and obtain probability, answer this each according to this probability and select obtain to return
Answer, answer as described first.
7. a man-machine interactive system, it is characterised in that including:
First acquisition module, for obtaining problem and the user type of user, wherein, the problem of the user of acquisition is asked as first
Topic, the user type of acquisition is as first user type；
Matching module, retrieves the problem maximum with described first problem matching degree in all problems from knowledge base and makees
For Second Problem；
Select module, anti-for the user answering corresponding described first user type according to Second Problem described in knowledge base
Feedback satisfaction, the answer to Second Problem described in knowledge base selects, and obtain answers as to described first problem
Answering, i.e. first answers；
Sending module, sends to described user for answering described first.
System the most according to claim 7, it is characterised in that also include: creation module, is used for being pre-created described knowledge
Storehouse；
Wherein, described knowledge base includes: at least one problem；
At least one of each described problem is answered；
At least one user type corresponding with each described answer；And
The user feedback satisfaction of every kind of described user type.
System the most according to claim 7, it is characterised in that also include:
Second acquisition module, for obtaining user's satisfaction to described first answer, as first user satisfaction feedback；
More new module, for updating described knowledge base according to described first user satisfaction feedback；
Add module, for by by no less than the described first problem of predetermined number, described first answer, described first user
Type and described first user satisfaction feedback are added into described knowledge base, form big data knowledge storehouse.
System the most according to claim 9, it is characterised in that described more new module includes:
Adding device, for during when described first problem and described Second Problem Incomplete matching, by described first problem, described
First answer, described first user type, described first user satisfaction feedback are added into described knowledge base；
Computing unit, for mating completely and when asking with described second in knowledge base with described Second Problem when described first problem
In the presence of the described first user type that topic is corresponding, by described first user satisfaction feedback, and it is stored in described knowledge base
Described in the described first described user feedback satisfaction answering corresponding described first user type of Second Problem, add
Weight average calculates, and the result of calculating answers corresponding described first user as described first of the described Second Problem after updating
The user feedback satisfaction of type；Or
First signal generating unit, for mating completely with described Second Problem and work as in knowledge base when described first problem with described the
Described first user type corresponding to two problems not in the presence of, using described first user satisfaction feedback as described in after updating
Described the first of Second Problem answers the user feedback satisfaction of corresponding described first user type.
11. systems according to claim 7, it is characterised in that described selection module includes:
First selects unit, is used in the presence of described first user type corresponding with described Second Problem in knowledge base, from
At least one of described Second Problem selects the highest the returning of described user feedback satisfaction of described first user type in answering
Answer and answer as described first；Or
Second select unit, for when described first user type corresponding with described Second Problem in knowledge base not in the presence of,
The immediate user class of attribute information with described first user type is selected from least one answer of described Second Problem
The answer that described user feedback satisfaction that type is corresponding is the highest is answered as described first.
12. systems according to claim 7, it is characterised in that described selection module includes:
3rd selects unit, is used in the presence of described first user type corresponding with described Second Problem in knowledge base, will
The described user feedback satisfaction of the described first user type of each answer correspondence of Second Problem described in described knowledge base
It is normalized and obtains probability, answer, to this each, the answer selecting to obtain according to this probability, answer as described first；
Or
4th select unit, for when described first user type corresponding with described Second Problem in knowledge base not in the presence of,
By closest for the attribute information of the described first user type corresponding with each answer of Second Problem described in described knowledge base
Described user feedback satisfaction corresponding to user type be normalized and obtain probability, according to this probability answer this each into
Row selects the answer obtained, and answers as described first.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610374251.5A CN106126516A (en) | 2016-05-31 | 2016-05-31 | Man-machine interaction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610374251.5A CN106126516A (en) | 2016-05-31 | 2016-05-31 | Man-machine interaction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106126516A true CN106126516A (en) | 2016-11-16 |
Family
ID=57270073
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610374251.5A Pending CN106126516A (en) | 2016-05-31 | 2016-05-31 | Man-machine interaction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106126516A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107967308A (en) * | 2017-11-16 | 2018-04-27 | 百度在线网络技术（北京）有限公司 | A kind of processing method of intelligent interaction, device, equipment and computer-readable storage medium |
| CN108109616A (en) * | 2016-11-25 | 2018-06-01 | 松下知识产权经营株式会社 | Information processing method, information processing unit and program |
| CN111914073A (en) * | 2020-07-15 | 2020-11-10 | 中国联合网络通信集团有限公司 | Customer service response method, device, equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6282534B1 (en) * | 1998-03-13 | 2001-08-28 | Intel Corporation | Reverse content indexing |
| KR20020058533A (en) * | 2000-12-30 | 2002-07-12 | 오길록 | Construction of Knowledge Base for Question/Answering on Internet |
| CN101178718A (en) * | 2007-05-17 | 2008-05-14 | 腾讯科技（深圳）有限公司 | Knowledge sharing system, problem searching method and problem publish method |
| CN104699708A (en) * | 2013-12-09 | 2015-06-10 | 中国移动通信集团北京有限公司 | Self-learning method and device for customer service robot |
| CN105159996A (en) * | 2015-09-07 | 2015-12-16 | 百度在线网络技术（北京）有限公司 | Deep question-and-answer service providing method and device based on artificial intelligence |
| CN105824970A (en) * | 2016-04-12 | 2016-08-03 | 华南师范大学 | Robot interaction method and system based on big data knowledge base and user feedback |
-
2016
- 2016-05-31 CN CN201610374251.5A patent/CN106126516A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6282534B1 (en) * | 1998-03-13 | 2001-08-28 | Intel Corporation | Reverse content indexing |
| KR20020058533A (en) * | 2000-12-30 | 2002-07-12 | 오길록 | Construction of Knowledge Base for Question/Answering on Internet |
| CN101178718A (en) * | 2007-05-17 | 2008-05-14 | 腾讯科技（深圳）有限公司 | Knowledge sharing system, problem searching method and problem publish method |
| CN104699708A (en) * | 2013-12-09 | 2015-06-10 | 中国移动通信集团北京有限公司 | Self-learning method and device for customer service robot |
| CN105159996A (en) * | 2015-09-07 | 2015-12-16 | 百度在线网络技术（北京）有限公司 | Deep question-and-answer service providing method and device based on artificial intelligence |
| CN105824970A (en) * | 2016-04-12 | 2016-08-03 | 华南师范大学 | Robot interaction method and system based on big data knowledge base and user feedback |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108109616A (en) * | 2016-11-25 | 2018-06-01 | 松下知识产权经营株式会社 | Information processing method, information processing unit and program |
| CN107967308A (en) * | 2017-11-16 | 2018-04-27 | 百度在线网络技术（北京）有限公司 | A kind of processing method of intelligent interaction, device, equipment and computer-readable storage medium |
| US11308948B2 (en) | 2017-11-16 | 2022-04-19 | Baidu Online Network Technology (Beijing) Co., Ltd. | Intelligent interaction processing method and apparatus, device and computer storage medium |
| CN111914073A (en) * | 2020-07-15 | 2020-11-10 | 中国联合网络通信集团有限公司 | Customer service response method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105824970B (en) | Robot interactive method and system based on big data knowledge base and user feedback | |
| CN105938484B (en) | Robot interaction method and system based on user feedback knowledge base | |
| CN104753985B (en) | Session list display methods and device | |
| CN110812843B (en) | Interactive method and device based on virtual image and computer storage medium | |
| US20180075763A1 (en) | System and method of generating recommendations to alleviate loneliness | |
| CN106126516A (en) | Man-machine interaction method and system | |
| JP5775242B1 (en) | Information processing apparatus, information processing method, and information processing program | |
| EP3857468A1 (en) | Recommendation method and system and method and system for improving a machine learning system | |
| CN112114918A (en) | Intelligent device, server, intelligent system and related interface display method | |
| KR20190083127A (en) | System and method for trainning convolution neural network model using image in terminal cluster | |
| US20230119860A1 (en) | Matching system, matching method, and matching program | |
| CN110491486A (en) | A kind of medical services reservation scheduling method | |
| CN106204062A (en) | The customer information list updating method of customer service system and device | |
| CN106126541B (en) | Personalized solution recommendation method and system based on big data knowledge base | |
| CN112911073B (en) | Intelligent knowledge graph construction method and device for outbound flow dialogue content | |
| EP1280080A1 (en) | Method and apparatus for modelling user behaviour | |
| KR20000058902A (en) | Internet-based adaptive testing system and real-time item parameter estimating system | |
| JP5775243B1 (en) | Information processing apparatus, information processing method, and information processing program | |
| JP6943190B2 (en) | Manual search system | |
| EP3644242A1 (en) | System and method for optimizing a service station layout | |
| KR102379153B1 (en) | Roommaid smart matching service providing system, method and application | |
| CN113225435A (en) | Method, device and readable storage medium for estimating call connection probability | |
| CN111145015A (en) | Reservation-free withdrawal amount determining method and device | |
| CN113268580B (en) | Session subject migration path mining method and device, computer equipment and medium | |
| CN112040075A (en) | Agent voice auxiliary system based on voice recognition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20161116 |
|
| WD01 | Invention patent application deemed withdrawn after publication |