CN107644641A - Session operational scenarios recognition methods, terminal and computer-readable recording medium - Google Patents
Session operational scenarios recognition methods, terminal and computer-readable recording medium Download PDFInfo
- Publication number
- CN107644641A CN107644641A CN201710636464.5A CN201710636464A CN107644641A CN 107644641 A CN107644641 A CN 107644641A CN 201710636464 A CN201710636464 A CN 201710636464A CN 107644641 A CN107644641 A CN 107644641A
- Authority
- CN
- China
- Prior art keywords
- scene
- alternate scenes
- information
- user session
- operational scenarios
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of session operational scenarios recognition methods, comprise the following steps:Receive the user session information of input;Based on user session information, preset alternate scenes are screened using preset scene Recognition rule, obtain first kind alternate scenes corresponding to user session information;And based on user session information, preset alternate scenes are screened using scene discrimination model, obtain the second class alternate scenes corresponding to user session information;Based on first kind alternate scenes and the second class alternate scenes, user session information is carried out to strengthen study processing, obtains optimal session operational scenarios corresponding with user session information;Judge whether optimal session operational scenarios are identical with current session scene, if differing, using optimal session operational scenarios as current session scene.The invention also discloses a kind of session operational scenarios identification terminal and computer-readable recording medium.The present invention realizes to be accurately identified in session operational scenarios change procedure to scene.
Description
Technical field
The present invention relates to session operational scenarios identification technology field, more particularly to a kind of session operational scenarios recognition methods, terminal and
Computer-readable recording medium.
Background technology
Automatic question answering refers to answer the problem of user proposes automatically using computer to meet appointing for user knowledge demand
Business, it is a kind of advanced form of information service.In recent years, as the rapid development of artificial intelligence, automatic question answering have become standby
The extensive research direction of concerned and development prospect, automatic question answering are considered as whether verifier has natural language understanding energy
One of main task of power, the research of automatic question answering are advantageous to promote the development of artificial intelligence related discipline.
But current automatically request-answering system is perfect not enough, many particular problems and difficulty are still faced.Current intelligence
Robot is mostly single-wheel conversational system, does not consider the data such as context of dialogue information, user's history dialog information, provides very
More answers are discontinuous, loftier, had a strong impact on Consumer's Experience.For such case, we have proposed a kind of more scenes to know
Method for distinguishing, in dialog procedure, it can actively adapt to user according to data such as the current input of user, dialog history information
The scene of dialogue so that dialogue more remarkable fluency.
The content of the invention
It is a primary object of the present invention to provide a kind of session operational scenarios recognition methods, identification terminal and computer-readable deposit
Storage media, it is intended to solve during intelligence machine person to person carries out interactive voice, it is difficult to accurately identify the technology of scene changes
Problem.
To achieve the above object, the present invention provides a kind of session operational scenarios recognition methods, the session operational scenarios recognition methods bag
Include:
Receive the user session information of input;
Based on the user session information, preset alternate scenes are screened using preset scene Recognition rule, obtained
To first kind alternate scenes corresponding to the user session information;And based on the user session information, differentiated using scene
Model screens to the preset alternate scenes, obtains the second class alternate scenes corresponding to the user session information;
Based on the first kind alternate scenes and the second class alternate scenes, the user session information is strengthened
Study is handled, and obtains optimal session operational scenarios corresponding with the user session information;
Judge whether the optimal session operational scenarios and current session scene are identical, if differing, by the optimal dialogue
Scene is as current session scene.
Preferably, it is described to be based on the user session information, alternate scenes are carried out using preset scene Recognition rule
Screening Treatment, the step of obtaining first kind alternate scenes corresponding to the user session information, include:
Extract the keyword in the user session information;
Based on the keyword, scene related to the keyword in the preset alternate scenes is screened, and by described in
Related scene is as the first kind alternate scenes corresponding to the user session information.
Preferably, it is described to be based on user session information, the preset alternate scenes are sieved using scene discrimination model
Choosing is handled, and the step of obtaining the second class alternate scenes corresponding to the user session information includes:
Extract the characteristic information in the user session information;
Based on the characteristic information, the preset alternate scenes are calculated by the scene discrimination model and believed with the feature
The matching degree of breath, and using the higher part scene of matching degree as the second class candidate field corresponding to the user session information
Scape.
Preferably, it is described to be based on the first kind alternate scenes and the second class alternate scenes, to the user session
Information include the step of strengthening study processing, obtain optimal session operational scenarios corresponding with the user session information:
It is action, the first kind alternate scenes and the second class alternate scenes as shape using the user session information
State, the user session information is carried out to strengthen study processing, for waiting from the first kind alternate scenes and second class
Optimal session operational scenarios are filtered out in the scape of selected scenes.
To achieve the above object, the present invention also provides a kind of identification terminal, and the identification terminal includes:
It is stored with the memory of session operational scenarios recognizer;
Processor, it is configured to perform the session operational scenarios recognizer to perform operations described below:
Receive the user session information of input;
Based on the user session information, preset alternate scenes are screened using preset scene Recognition rule, obtained
To first kind alternate scenes corresponding to the user session information;And based on the user session information, differentiated using scene
Model screens to the preset alternate scenes, obtains the second class alternate scenes corresponding to the user session information;
Based on the first kind alternate scenes and the second class alternate scenes, the user session information is strengthened
Study is handled, and obtains optimal session operational scenarios corresponding with the user session information;
Judge whether the optimal session operational scenarios and current session scene are identical, if differing, by the optimal dialogue
Scene is as current session scene.
Alternatively, the user session information is based on described in performing, using preset scene Recognition rule to preset candidate
Scene is screened, and is obtained the operation of first kind alternate scenes corresponding to the user session information and is included:
Extract the keyword in the user session information;
Based on the keyword, scene related to the keyword in the preset alternate scenes is screened, and by described in
Related scene is as the first kind alternate scenes corresponding to the user session information.
Alternatively, the user session information is based on described in performing, using scene discrimination model to the preset candidate field
Scape is screened, and is obtained the operation of the second class alternate scenes corresponding to the user session information and is included:
Extract the characteristic information in the user session information;
Based on the characteristic information, the preset alternate scenes are calculated by the scene discrimination model and believed with the feature
The matching degree of breath, and using the higher part scene of matching degree as the second class candidate field corresponding to the user session information
Scape.
Preferably, based on the first kind alternate scenes and the second class alternate scenes described in performing, to the user
Dialog information carries out strengthening study processing, and obtaining the operation of optimal session operational scenarios corresponding with the user session information includes:
It is action, the first kind alternate scenes and the second class alternate scenes as shape using the user session information
State, the user session information is carried out to strengthen study processing, for waiting from the first kind alternate scenes and second class
Optimal session operational scenarios are filtered out in the scape of selected scenes.
To achieve the above object, the present invention also provides a kind of computer-readable recording medium, the computer-readable storage
Session operational scenarios recognizer is stored with medium, is realized when the session operational scenarios recognizer is executed by processor any as described above
Described in session operational scenarios recognition methods the step of.
In the present invention, specifically in scene change procedure, when session operational scenarios change, first using preset
Scene Recognition rule and scene discrimination model carry out first time screening to preset alternate scenes, and obtain corresponding alternate scenes,
Then programmed screening is carried out to alternate scenes using strengthening learning strategy, and then obtains optimal session operational scenarios, last basis
The optimal suitable dialogue result of session operational scenarios selection replys user, it is achieved thereby that the accurate knowledge to current session scene changes
Not, so lifted user's man-machine interaction usage experience.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of the embodiment of intelligent robot session operational scenarios recognition methods one of the present invention;
Fig. 2 is the refinement schematic flow sheet of step S20 in Fig. 1;
Fig. 3 is the refinement schematic flow sheet of step S30 in Fig. 1;
Fig. 4 is the refinement schematic flow sheet of step S40 in Fig. 1.
The realization, functional characteristics and advantage of the object of the invention will be described further referring to the drawings in conjunction with the embodiments.
Embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
Reference picture 1, Fig. 1 are the schematic flow sheet of the embodiment of intelligent robot session operational scenarios recognition methods one of the present invention.This
In embodiment, session operational scenarios recognition methods comprises the following steps:
Step S10, receive the user session information of input;
In the present embodiment, as user and intelligent robot dialogue, intelligent robot passes through voice interactive system and receives use
The dialog information of family input.Wherein, the mode of intelligent robot reception user session information includes:Sound is entered into row information to turn
Change, utilize the sound recognition system in system.
Sound is entered into row information conversion sound is exactly converted into electric wave, then sound is converted into intelligent robot to know
Other information, robot carry out respective feedback by receive information.
Using tone, tone color during sound recognition system by identifying user etc., pass through information corresponding to tone and tone color
Make corresponding feedback.
Step S20, based on the user session information, preset alternate scenes are carried out using preset scene Recognition rule
Screening, obtains first kind alternate scenes corresponding to the user session information;
In the present embodiment, when receiving the dialog information of user's input, by using scene Recognition rule, to preset field
Scape information is screened, and obtains corresponding alternate scenes.Scene Recognition rule is specifically based on the practical experience of people and set
Session rules, and setting of the present embodiment for preset scene Recognition rule is unlimited.
For example the rule by obtaining in dialog information carries out scene Recognition.For example, when user's input dialogue information is
When " who are you, and how old is this year ", the rule of acquisition is exactly " identity " and " age ", and now scene Recognition rule may be set to
The basic condition introduction of people, therefore corresponding answer can be " hello, and I cries little Bai, two years old this year ".For another example when user is defeated
When to enter dialog information be " I will inquire about air ticket ", the dialogue keyword of acquisition is exactly " inquiry air ticket ", now scene Recognition rule
The offer for being set to inquiry air ticket mode is opened up, therefore corresponding answer can be " when the departure time is ".
For another example, scene Recognition is carried out by problem and the corresponding of answer.For example, when user's input dialogue information is " you
Who is more handsome with me " when, now using scene Recognition rule directly corresponding to answer can be " certainly I " or " I more
It is handsome ".For another example when user's input dialogue information is " good night ", now can directly answer " good night ".
Step S30, based on the user session information, the preset alternate scenes are sieved using scene discrimination model
Choosing, obtains the second class alternate scenes corresponding to the user session information;
In the present embodiment, scene discrimination model is a kind of algorithm of machine learning, particular by many relevant informations
All extract, with reference to historical information, look for historical information and the information that is extracted with the presence or absence of similar, can if similar
Similar scene is classified as one kind.The algorithm of discrimination model has a variety of, and the algorithm of discrimination model is used in the present embodiment
Type be unlimited.In the present embodiment, used discrimination model calculation can be:Calculate characteristic information and candidate field
The size of the correlation of scape information, and then determine alternate scenes.
For example, when user session information is " I wants body-building ", can now obtain outside user profile, can also be according to this information
Get user to want to take exercise, talk with the problem of referring on exercise or body-building before more being obtained from dialog history
Relevant information, when obtaining related characteristic information including user view understanding, dialog history information, user preference information, just
According to characteristic information, scene is screened by discrimination model, answer now can " go to ××× health club to go
" or " go moat border area run ", without being " I should not body-building ".
For another example when user session information is " what be for dinner ", associated scenario that may be present just has in the scene
" dinner goes to western-style restaurant to have beefsteak ", " dinner goes to have Hunan cuisine ", " dinner do-it-yourself fried rice with eggs " or " I does not have supper " etc..With
On four kinds of scenes to have a similitude be exactly to have supper, but the 4th and first three substantially just have a pair having supper
As different problem, first three when user and the 4th becomes robot.So when user session says that " dinner is eaten assorted
" when, it is clear that the 4th is not meet scene now.
It is pointed out that above-mentioned steps S20 and S30 execution sequencing are unlimited, for example step S20 is first carried out, and
Perform step S30 again afterwards, or first carry out step S30, then enter simultaneously to execution step S20, or step S20 and S30 again
OK.
Step S40, based on the first kind alternate scenes and the second class alternate scenes, to the user session information
Carry out strengthening study processing, obtain optimal session operational scenarios corresponding with the user session information;
In the present embodiment, enhancing study is the process of an autonomous learning, and by constantly learning, selecting one can reach
The optimal action of target.Enhancing study can be realized in several ways, due to having individual value during study is strengthened
Function (state value function), Q functions (action value function) or strategy etc., and value function, Q functions or strategy are according to not
Same task carries out different definition, so being unlimited for mode of its realization in the present embodiment.
In the present embodiment, the second class time is obtained by the first kind alternate scenes obtained by step S20 and by step S30
Alternate scenes set of the selected scenes scape as enhancing study.When obtaining user session information action, using this alternate scenes as state
Set, by strengthening screening of the study realization to alternate scenes.
For example, when user session information is " what be for dinner ", by scene Recognition rule and scene discrimination model
For the first time first kind alternate scenes and the second class alternate scenes are obtained after screening.Now alternate scenes can include:" dinner
Go to western-style restaurant to have beefsteak ", " dinner goes to have Hunan cuisine ", " dinner do-it-yourself fried rice with eggs ", " dinner oneself is cooked ", " after 6 points
Have supper again ", " I does not have supper " etc..
When user getting " what be for dinner " the problem of, intelligent robot has a clear understanding of the target of user, by being worth letter
Several calculating, the corresponding value of each scene, such as, the value corresponding to " dinner goes to western-style restaurant to have beefsteak " is 54, " dinner
Go to have Hunan cuisine " corresponding to value be 63, the value corresponding to " dinner oneself is cooked " is 72, " 6 points after have supper again " institute
Corresponding value is 81, the value corresponding to " dinner do-it-yourself fried rice with eggs " is 90, and now different values represents close with target
Degree (full marks 100, that is, reach target), so now optimal answer can be " dinner do-it-yourself fried rice with eggs ".
In the present embodiment, by strengthening the calculating of the value function in learning, a marking row is carried out to scene using numerical value
Sequence, choose marking one scene of highest and exported as optimal scene.
Step S50, judge whether the optimal session operational scenarios and current session scene are identical, if differing, by described in
Optimal session operational scenarios are as current session scene;
In the present embodiment, when filtering out optimal session operational scenarios by strengthening study, by optimal session operational scenarios and currently
Session operational scenarios are contrasted, if optimal session operational scenarios do not change with current degreeization scene, current session scene is constant, if right
Than finding that optimal session operational scenarios and current session scene are different, then using optimal session operational scenarios as newest current session scene
Output.For example, before user and the scene of robot dialogue is topic scene on weather, and work as user again with robot
During dialogue, according to dialog information screen when on the topic scene had a meal, now can be by session operational scenarios from pass above
In the topic scene transitions of weather be on the topic scene had a meal.
In the present embodiment, it is session operational scenarios the most suitable in optimal session operational scenarios and current session operational scenarios are defeated
Go out, and corresponding operation is made according to session operational scenarios the most suitable.
Embodiment two:
Reference picture 2, Fig. 2 are the refinement schematic flow sheet of step S20 in Fig. 1.Based on above-described embodiment one, in the present embodiment
In, above-mentioned steps S20 further comprises:
Step S201, extract the keyword in the user session information;
Step S202, based on the keyword, scene related to the keyword in the preset alternate scenes is screened,
And using the related scene as the first kind alternate scenes corresponding to the user session information.
In the present embodiment, when receiving the dialog information of user's input, keyword therein is extracted according to dialog information,
Then the higher scene of degree related to the keyword in preset alternate scenes is screened according to the keyword extracted, and
Using the higher part scene of degree of correlation as first kind alternate scenes.
For example, the dialog information of user's input is " I wants to eat Sichuan cuisine ", and when receiving this input information, Extracting Information
In keyword " having a meal " and " Sichuan cuisine ", now go to be matched with preset alternate scenes according to keyword, obtain degree of correlation
Higher part scene.When obtaining keyword " eating " and " food ", the scene that can now match may just be related to attached
The scene information in nearly Sichuan cuisine shop, and then obtain more on the scene of " having a meal " and " Sichuan cuisine " as first kind alternate scenes.
For another example the dialog information of user's input is " I wants body-building ", when receiving this information, can be apparent from
Acquisition keyword therein " body-building ", the scene information now obtained may be exactly:" what project of body-building has ", " body-building
Place where ", a series of scenes related to " body-building " such as " how body-building " just appear in alternate scenes, such as intelligence
Can the dialogue of robot output be " you be intended to run also is intended to gymnasium ", user can select that " I wants to go to race again
Step ".It now just have chosen the scene of dialogue.
In the present embodiment, by recognition rule obtain first kind alternate scenes information, the keyword of the dialog information of acquisition,
It is the session operational scenarios information set and the session operational scenarios matched are also to refer to the session operational scenarios of keyword, will not be with user
Change and change.
Embodiment three:
Reference picture 3, Fig. 3 are the refinement schematic flow sheet of step S30 in Fig. 1.Based on above-described embodiment one, in the present embodiment
In, above-mentioned steps S30 further comprises:
Step S301, extract the characteristic information in the user session information;
Step S302, based on the characteristic information, by the scene discrimination model calculate the preset alternate scenes with
The matching degree of the characteristic information, and using the higher part scene of matching degree as corresponding to the user session information
Two class alternate scenes.
In the present embodiment, when receiving the dialog information of user's input, the characteristic information in dialog information is extracted, is passed through
Preset alternate scenes and this feature information are carried out to the calculating of matching degree, the higher part scene of matching degree is obtained and is used as the
Two class alternate scenes, wherein the characteristic information extracted includes user view, this dialog information and the last period dialog information
Relevance and the Preference of user on problems etc. in preset alternate scenes.
For example, when the dialog information of user's input is " what noon eats ", while user profile is obtained, according to this
Information and user asked before this it is similar the problem of, characteristic information now may just have the frequent consumer behavior of user, user
Often where eat, taste of user etc., by contrasting these characteristic informations so as to select place of more suitably having a meal, now
It may will reply " going to ××× dining room to eat Sichuan cuisine " or " oneself be in and cook " etc..
For another example user session information is " I wants body-building ", now discrimination model in addition to it can obtain user profile, always according to
This acquisition of information needs to take exercise to user, and now also not merely simply obtains the information in dialog information, more can be according to right
The historical information of words, before being obtained from dialog history dialogue referred to and gathering on the problem of exercise or body-building, and from collection
The selection preference of user, the physical training style of concern and place for often going etc. are extracted in conjunction, when the certain correlated characteristic of acquisition
When, just according to characteristic information, scene is screened by discrimination model, may will reply and " go to ××× health club
Go ahead " or " going to moat border area to run ".
Pass through embodiment two and embodiment three, it is found that set because preset scene Recognition rule is one
Unalterable rule, so lack certain flexibility in identification process, during in face of a problem, it can reply eternal
Only one or several answers fixed;And user view can accurately be identified by preset discrimination model, enter
And more particularly suitable answer is provided, but the method need to accumulate a certain amount of dialog information data.Then pass through this two methods pair
Preset alternate scenes carry out first time screening, select more suitable part scene, reduce sieve for secondary accurate screening
Select scope.
Example IV:
Reference picture 4, Fig. 4 are the refinement schematic flow sheet of step S40 in Fig. 1.Based on above-described embodiment one, in the present embodiment
In, above-mentioned steps S40 further comprises:
Step S401, receives first kind alternate scenes and the second class alternate scenes, and using the alternate scenes received as
Strengthen the state set learned;
Step S402, the action using dialog information as enhancing study, filtered out by strengthening study in state set
Optimum state, and then determine optimal session operational scenarios.
In the present embodiment, after first kind alternate scenes and the second class alternate scenes are received, by alternate scenes
Each scene is as a state, using user session information as an action, by strengthening value function or Q functions in learning
Each state is calculated with reference to current action, the value being calculated is ranked up, the maximum optimal shape of conduct of selected value
State, scene corresponding to the optimum state are optimal session operational scenarios.
By scene obtained by rule and scene discrimination model for example, when user session information is " I will go body-building "
To first kind alternate scenes and the second class alternate scenes scene set can have " going to ××× health club to go ahead ",
" go moat border area run ", " going for a fitness ", " going to do a body-building card ", " I should not body-building " etc.
Deng.Intelligent robot has a clear understanding of the target of user, by the calculating of value function, the corresponding value of each scene, such as,
Value corresponding to " me should not body-building " is 54, the value corresponding to " going to moat border area to run " is 63, " goes ××× body-building all
Happy portion goes ahead " corresponding to value be 72, the value corresponding to " going to do a body-building card " is 81, " going for a fitness "
Corresponding value is 90, so now optimal answer can be " going for a fitness ".
The present invention also protects a kind of session operational scenarios identification terminal.
In the present invention, in the embodiment of terminal one of the present invention, session operational scenarios identification terminal includes:
It is stored with the memory of session operational scenarios recognizer;Processor, it is configured to perform the session operational scenarios recognizer
To perform operations described below:
Receive the user session information of input;
Based on the user session information, preset alternate scenes are screened using preset scene Recognition rule, obtained
To first kind alternate scenes corresponding to the user session information;And based on the user session information, differentiated using scene
Model screens to the preset alternate scenes, obtains the second class alternate scenes corresponding to the user session information;
Based on the first kind alternate scenes and the second class alternate scenes, the user session information is strengthened
Study is handled, and obtains optimal session operational scenarios corresponding with the user session information;
Judge whether the optimal session operational scenarios and current session scene are identical, if differing, by the optimal dialogue
Scene is as current session scene.
In the present embodiment, as user and intelligent robot dialogue, intelligent robot passes through voice interactive system and receives use
The dialog information of family input.Wherein, the mode of intelligent robot reception user session information includes:Sound is entered into row information to turn
Change, utilize the sound recognition system in system.
Sound is entered into row information conversion sound is exactly converted into electric wave, then sound is converted into intelligent robot to know
Other information, robot carry out respective feedback by receive information.
Using tone, tone color during sound recognition system by identifying user etc., pass through information corresponding to tone and tone color
Make corresponding feedback.
In the present embodiment, when receiving the dialog information of user's input, by using scene Recognition rule, to preset field
Scape information is screened, and obtains corresponding alternate scenes.Scene Recognition rule is specifically based on the practical experience of people and set
Session rules, and setting of the present embodiment for preset scene Recognition rule is unlimited.
For example carry out scene Recognition by obtaining the keyword in dialog information.For example, when user's input dialogue information is
When " who are you, and how old is this year ", the rule of acquisition is exactly " identity " and " age ", and now scene Recognition rule may be set to
The basic condition introduction of people, therefore corresponding answer can be " hello, and I cries little Bai, two years old this year ".For another example when user is defeated
When to enter dialog information be " I will inquire about air ticket ", the dialogue keyword of acquisition is exactly " inquiry air ticket ", now scene Recognition rule
The offer for being set to inquiry air ticket mode is opened up, therefore corresponding answer can be " during the departure time when ".
For another example, scene Recognition is carried out by problem and the corresponding of answer.For example, when user's input dialogue information is " you
Who is more handsome with me " when, now using scene Recognition rule directly corresponding to answer can be " certainly I " or " I more
It is handsome ".For another example when user's input dialogue information is " good night ", now can directly answer " good night ".
In the present embodiment, scene discrimination model is a kind of algorithm of machine learning, particular by many relevant informations
All extract, with reference to historical information, look for historical information and the information that is extracted with the presence or absence of similar, can if similar
Similar scene is classified as one kind.The algorithm of discrimination model has a variety of, and the algorithm of discrimination model is used in the present embodiment
Type be unlimited.In the present embodiment, used discrimination model calculation can be:Calculate characteristic information and candidate field
The size of the correlation of scape information, and then determine alternate scenes.
For example, when user session information is " I wants body-building ", can now obtain outside user profile, can also be according to this information
Get user to want to take exercise, talk with the problem of referring on exercise or body-building before more being obtained from dialog history
Relevant information, when obtaining related characteristic information including user view understanding, dialog history information, user preference information, just
According to characteristic information, scene is screened by discrimination model, answer now can " go to ××× health club to go
" or " go moat border area run ", without being " I should not body-building ".
For another example when user session information is " what be for dinner ", associated scenario that may be present just has in the scene
" dinner goes to western-style restaurant to have beefsteak ", " dinner goes to have Hunan cuisine ", " dinner do-it-yourself fried rice with eggs " or " I does not have supper " etc..With
On four kinds of scenes to have a similitude be exactly to have supper, but the 4th and first three substantially just have a pair having supper
As different problem, first three when user and the 4th becomes robot.So when user session says that " dinner is eaten assorted
" when, it is clear that the 4th is not meet scene now.
In the present embodiment, enhancing study is the process of an autonomous learning, and by constantly learning, selecting one can reach
The optimal action of target.Enhancing study can be realized in several ways, due to having individual value during study is strengthened
Function (state value function), Q functions (action value function) or strategy etc., and value function, Q functions or strategy are according to not
Same task carries out different definition, so being unlimited for mode of its realization in the present embodiment.
In the present embodiment, the second class time is obtained by the first kind alternate scenes obtained by step S20 and by step S30
Alternate scenes set of the selected scenes scape as enhancing study.When obtaining user session information action, using this alternate scenes as state
Set, by strengthening screening of the study realization to alternate scenes.
For example, when user session information is " what be for dinner ", by scene Recognition rule and scene discrimination model
For the first time first kind alternate scenes and the second class alternate scenes are obtained after screening.Now alternate scenes can include:" dinner
Go to western-style restaurant to have beefsteak ", " dinner goes to have Hunan cuisine ", " dinner do-it-yourself fried rice with eggs ", " dinner oneself is cooked ", " after 6 points
Have supper again ", " I does not have supper " etc..
When user getting " what be for dinner " the problem of, intelligent robot has a clear understanding of the target of user, by being worth letter
Several calculating, the corresponding value of each scene, such as, the value corresponding to " dinner goes to western-style restaurant to have beefsteak " is 54, " dinner
Go to have Hunan cuisine " corresponding to value be 63, the value corresponding to " dinner oneself is cooked " is 72, " 6 points after have supper again " institute
Corresponding value is 81, the value corresponding to " dinner do-it-yourself fried rice with eggs " is 90, and now different values represents close with target
Degree (full marks 100, that is, reach target), so now optimal answer can be " dinner do-it-yourself fried rice with eggs ".
In the present embodiment, by strengthening the calculating of the value function in learning, a marking row is carried out to scene using numerical value
Sequence, choose marking one scene of highest and exported as optimal scene.
In the present embodiment, when filtering out optimal session operational scenarios by strengthening study, by optimal session operational scenarios and currently
Session operational scenarios are contrasted, if optimal session operational scenarios do not change with current degreeization scene, current session scene is constant, if right
Than finding that optimal session operational scenarios and current session scene are different, then using optimal session operational scenarios as newest current session scene
Output.For example, before user and the scene of robot dialogue is topic scene on weather, and work as user again with robot
During dialogue, according to dialog information screen when on the topic scene had a meal, now can be by session operational scenarios from pass above
In the topic scene transitions of weather be on the topic scene had a meal.
In the present embodiment, it is session operational scenarios the most suitable in optimal session operational scenarios and current session operational scenarios are defeated
Go out, and corresponding operation is made according to session operational scenarios the most suitable.
Further alternative, in the embodiment of terminal one of the present invention, computing device is based on the user session information, adopts
Preset alternate scenes are screened with preset scene Recognition rule, the first kind corresponding to the user session information is obtained and waits
The operation of selected scenes scape includes:
Extract the keyword in the user session information;
Based on the keyword, scene related to the keyword in the preset alternate scenes is screened, and by described in
Related scene is as the first kind alternate scenes corresponding to the user session information.
In the present embodiment, when receiving the dialog information of user's input, keyword therein is extracted according to dialog information,
Then the higher scene of degree related to the keyword in preset alternate scenes is screened according to the keyword extracted, and
Using the higher part scene of degree of correlation as first kind alternate scenes.
For example, the dialog information of user's input is " I wants to eat Sichuan cuisine ", and when receiving this input information, Extracting Information
In keyword " having a meal " and " Sichuan cuisine ", now go to be matched with preset alternate scenes according to keyword, obtain degree of correlation
Higher part scene.When obtaining keyword " eating " and " food ", the scene that can now match may just be related to attached
The scene information in nearly Sichuan cuisine shop, and then obtain more on the scene of " having a meal " and " Sichuan cuisine " as first kind alternate scenes.
For another example the dialog information of user's input is " I wants body-building ", when receiving this information, can be apparent from
Acquisition keyword therein " body-building ", the scene information now obtained may be exactly:" what project of body-building has ", " body-building
Place where ", a series of scenes related to " body-building " such as " how body-building " just appear in alternate scenes, such as intelligence
Can the dialogue of robot output be " you be intended to run also is intended to gymnasium ", user can select that " I wants to go to race again
Step ".It now just have chosen the scene of dialogue.
In the present embodiment, by recognition rule obtain first kind alternate scenes information, the keyword of the dialog information of acquisition,
It is the session operational scenarios information set and the session operational scenarios matched are also to refer to the session operational scenarios of keyword, will not be with user
Change and change.
Further alternative, in the embodiment of terminal one of the present invention, computing device is based on the user session information, adopts
The preset alternate scenes are screened with scene discrimination model, obtain the second class candidate corresponding to the user session information
The operation of scene includes:
Extract the characteristic information in the user session information;
Based on the characteristic information, the preset alternate scenes are calculated by the scene discrimination model and believed with the feature
The matching degree of breath, and using the higher part scene of matching degree as the second class candidate field corresponding to the user session information
Scape.
In the present embodiment, when receiving the dialog information of user's input, the characteristic information in dialog information is extracted, is passed through
Preset alternate scenes and this feature information are carried out to the calculating of matching degree, the higher part scene of matching degree is obtained and is used as the
Two class alternate scenes, wherein the characteristic information extracted includes user view, this dialog information and the last period dialog information
Relevance and the Preference of user on problems etc. in preset alternate scenes.
For example, when the dialog information of user's input is " what noon eats ", while user profile is obtained, according to this
Information and user asked before this it is similar the problem of, characteristic information now may just have the frequent consumer behavior of user, user
Often where eat, taste of user etc., by contrasting these characteristic informations so as to select place of more suitably having a meal, now
It may will reply " going to ××× dining room to eat Sichuan cuisine " or " oneself be in and cook " etc..
For another example user session information is " I wants body-building ", now discrimination model in addition to it can obtain user profile, always according to
This acquisition of information needs to take exercise to user, and now also not merely simply obtains the information in dialog information, more can be according to right
The historical information of words, before being obtained from dialog history dialogue referred to and gathering on the problem of exercise or body-building, and from collection
The selection preference of user, the physical training style of concern and place for often going etc. are extracted in conjunction, when the certain correlated characteristic of acquisition
When, just according to characteristic information, scene is screened by discrimination model, may will reply and " go to ××× health club
Go ahead " or " going to moat border area to run ".
Further alternative, in the embodiment of terminal one of the present invention, computing device is based on the first kind alternate scenes
With the second class alternate scenes, the user session information is carried out to strengthen study processing, obtains believing with the user session
The operation of optimal session operational scenarios includes corresponding to breath:
Receive first kind alternate scenes and the second class alternate scenes, and the shape that the alternate scenes received are learned as enhancing
State set;
Action using dialog information as enhancing study, learn to filter out the optimum state in state set by strengthening,
And then determine optimal session operational scenarios.
In the present embodiment, after first kind alternate scenes and the second class alternate scenes are received, by alternate scenes
Each scene is as a state, using user session information as an action, by strengthening value function or Q functions in learning
Each state is calculated with reference to current action, the value being calculated is ranked up, the maximum optimal shape of conduct of selected value
State, scene corresponding to the optimum state are optimal session operational scenarios.
By scene obtained by rule and scene discrimination model for example, when user session information is " I will go body-building "
To first kind alternate scenes and the second class alternate scenes scene set can have " going to ××× health club to go ahead ",
" go moat border area run ", " going for a fitness ", " going to do a body-building card ", " I should not body-building " etc.
Deng.Intelligent robot has a clear understanding of the target of user, by the calculating of value function, the corresponding value of each scene, such as,
Value corresponding to " me should not body-building " is 54, the value corresponding to " going to moat border area to run " is 63, " goes ××× body-building all
Happy portion goes ahead " corresponding to value be 72, the value corresponding to " going to do a body-building card " is 81, " going for a fitness "
Corresponding value is 90, so now optimal answer can be " going for a fitness ".
The embodiment of the present invention also proposes a kind of computer-readable recording medium.
Session operational scenarios recognizer, the session operational scenarios recognizer are stored with computer-readable recording medium of the present invention
The step of session operational scenarios recognition methods in embodiment as described above is realized when being executed by processor.
The session operational scenarios recognition methods that the present embodiment proposes, during specific scene changes, when session operational scenarios occur
When change, first time sieve is carried out to preset alternate scenes using preset scene Recognition rule and scene discrimination model first
Choosing, and obtains corresponding alternate scenes, then carries out programmed screening to alternate scenes using strengthening learning strategy, and then obtain
Optimal session operational scenarios, user is finally replied according to the optimal suitable dialogue result of session operational scenarios selection, it is achieved thereby that right
Current session scene changes accurately identify, and then lift the usage experience of user's man-machine interaction.
It should be noted that herein, term " comprising ", "comprising" or its any other variant are intended to non-row
His property includes, so that process, method, article or system including a series of elements not only include those key elements, and
And also include the other element being not expressly set out, or also include for this process, method, article or system institute inherently
Key element.In the absence of more restrictions, the key element limited by sentence "including a ...", it is not excluded that including this
Other identical element also be present in the process of key element, method, article or system.
The embodiments of the present invention are for illustration only, do not represent the quality of embodiment.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can add the mode of required general hardware platform to realize by software, naturally it is also possible to by hardware, but in many cases
The former is more preferably embodiment.Based on such understanding, technical scheme is substantially done to prior art in other words
Going out the part of contribution can be embodied in the form of software product, and the computer software product is stored in one as described above
In storage medium (such as ROM/RAM, magnetic disc, CD), including some instructions to cause a station terminal equipment (can be mobile phone,
Computer, server, air conditioner, or network equipment etc.) perform method described in each embodiment of the present invention.
The preferred embodiments of the present invention are these are only, are not intended to limit the scope of the invention, it is every to utilize this hair
The equivalent structure or equivalent flow conversion that bright specification and accompanying drawing content are made, or directly or indirectly it is used in other related skills
Art field, is included within the scope of the present invention.
Claims (9)
1. a kind of session operational scenarios recognition methods, it is characterised in that the session operational scenarios recognition methods comprises the following steps:
Receive the user session information of input;
Based on the user session information, preset alternate scenes are screened using preset scene Recognition rule, obtain institute
State first kind alternate scenes corresponding to user session information;And based on the user session information, using scene discrimination model
The preset alternate scenes are screened, obtain the second class alternate scenes corresponding to the user session information;
Based on the first kind alternate scenes and the second class alternate scenes, enhancing study is carried out to the user session information
Processing, obtains optimal session operational scenarios corresponding with the user session information;
Judge whether the optimal session operational scenarios and current session scene are identical, if differing, by the optimal session operational scenarios
As current session scene.
2. session operational scenarios recognition methods as claimed in claim 1, it is characterised in that it is described to be based on the user session information,
Screening Treatment is carried out to alternate scenes using preset scene Recognition rule, obtains the first kind corresponding to the user session information
The step of alternate scenes, includes:
Extract the keyword in the user session information;
Based on the keyword, scene related to the keyword in the preset alternate scenes is screened, and by the correlation
Scene as the first kind alternate scenes corresponding to the user session information.
3. session operational scenarios recognition methods as claimed in claim 1, it is characterised in that it is described to be based on user session information, use
Scene discrimination model carries out Screening Treatment to the preset alternate scenes, obtains the second class corresponding to the user session information and waits
The step of selected scenes scape, includes:
Extract the characteristic information in the user session information;
Based on the characteristic information, the preset alternate scenes and the characteristic information are calculated by the scene discrimination model
Matching degree, and using the higher part scene of matching degree as the second class alternate scenes corresponding to the user session information.
4. session operational scenarios recognition methods as claimed in claim 1, it is characterised in that described to be based on the first kind alternate scenes
With the second class alternate scenes, the user session information is carried out to strengthen study processing, obtains believing with the user session
Include corresponding to breath the step of optimal session operational scenarios:
Using the user session information be action, the first kind alternate scenes and the second class alternate scenes as state, it is right
The user session information carries out strengthening study processing, for from the first kind alternate scenes and the second class alternate scenes
In filter out optimal session operational scenarios.
5. a kind of session operational scenarios identification terminal, it is characterised in that the session operational scenarios identification terminal includes:
It is stored with the memory of session operational scenarios recognizer;
Processor, it is configured to perform the session operational scenarios recognizer to perform operations described below:
Receive the user session information of input;
Based on the user session information, preset alternate scenes are screened using preset scene Recognition rule, obtain institute
State first kind alternate scenes corresponding to user session information;And based on the user session information, using scene discrimination model
The preset alternate scenes are screened, obtain the second class alternate scenes corresponding to the user session information;
Based on the first kind alternate scenes and the second class alternate scenes, enhancing study is carried out to the user session information
Processing, obtains optimal session operational scenarios corresponding with the user session information;
Judge whether the optimal session operational scenarios and current session scene are identical, if differing, by the optimal session operational scenarios
As current session scene.
6. session operational scenarios identification terminal as claimed in claim 5, it is characterised in that perform described based on user session letter
Preset alternate scenes are screened by breath using preset scene Recognition rule, obtain corresponding to the user session information the
The operation of a kind of alternate scenes includes:
Extract the keyword in the user session information;
Based on the keyword, scene related to the keyword in the preset alternate scenes is screened, and by the correlation
Scene as the first kind alternate scenes corresponding to the user session information.
7. session operational scenarios identification terminal as claimed in claim 5, it is characterised in that perform described based on user session letter
The preset alternate scenes are screened, obtained second corresponding to the user session information by breath using scene discrimination model
The operation of class alternate scenes includes:
Extract the characteristic information in the user session information;
Based on the characteristic information, the preset alternate scenes and the characteristic information are calculated by the scene discrimination model
Matching degree, and using the higher part scene of matching degree as the second class alternate scenes corresponding to the user session information.
8. session operational scenarios identification terminal as claimed in claim 5, it is characterised in that be based on the first kind candidate described in performing
Scene and the second class alternate scenes, the user session information is carried out to strengthen study processing, obtained and the user couple
The operation of optimal session operational scenarios includes corresponding to words information:
Using the user session information be action, the first kind alternate scenes and the second class alternate scenes as state, it is right
The user session information carries out strengthening study processing, for from the first kind alternate scenes and the second class alternate scenes
In filter out optimal session operational scenarios.
9. a kind of computer-readable recording medium, it is characterised in that dialogue field is stored with the computer-readable recording medium
Scape recognizer, realized when the session operational scenarios recognizer is executed by processor as any one of Claims 1-4
The step of session operational scenarios recognition methods.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710636464.5A CN107644641B (en) | 2017-07-28 | 2017-07-28 | Dialog scene recognition method, terminal and computer-readable storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710636464.5A CN107644641B (en) | 2017-07-28 | 2017-07-28 | Dialog scene recognition method, terminal and computer-readable storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107644641A true CN107644641A (en) | 2018-01-30 |
CN107644641B CN107644641B (en) | 2021-04-13 |
Family
ID=61110969
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710636464.5A Active CN107644641B (en) | 2017-07-28 | 2017-07-28 | Dialog scene recognition method, terminal and computer-readable storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107644641B (en) |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108804603A (en) * | 2018-05-29 | 2018-11-13 | 北京灵智优诺科技有限公司 | Man-machine written dialogue method and system, server, medium |
CN109086860A (en) * | 2018-05-28 | 2018-12-25 | 北京光年无限科技有限公司 | A kind of exchange method and system based on visual human |
CN110858226A (en) * | 2018-08-07 | 2020-03-03 | 北京京东尚科信息技术有限公司 | Conversation management method and device |
CN110880324A (en) * | 2019-10-31 | 2020-03-13 | 北京大米科技有限公司 | Voice data processing method and device, storage medium and electronic equipment |
CN111290953A (en) * | 2020-01-22 | 2020-06-16 | 华为技术有限公司 | Method and device for analyzing test logs |
CN111475206A (en) * | 2019-01-04 | 2020-07-31 | 优奈柯恩(北京)科技有限公司 | Method and apparatus for waking up wearable device |
CN111813900A (en) * | 2019-04-10 | 2020-10-23 | 北京猎户星空科技有限公司 | Multi-turn conversation processing method and device, electronic equipment and storage medium |
CN111881254A (en) * | 2020-06-10 | 2020-11-03 | 百度在线网络技术(北京)有限公司 | Method and device for generating dialogs, electronic equipment and storage medium |
CN112487170A (en) * | 2020-12-14 | 2021-03-12 | 南京三眼精灵信息技术有限公司 | Scene configuration-oriented human-computer interaction dialogue robot system |
WO2021129439A1 (en) * | 2019-12-28 | 2021-07-01 | 科大讯飞股份有限公司 | Voice recognition method and related product |
CN113488036A (en) * | 2020-06-10 | 2021-10-08 | 海信集团有限公司 | Multi-round voice interaction method, terminal and server |
CN113822058A (en) * | 2021-09-18 | 2021-12-21 | 上海明略人工智能(集团)有限公司 | Dialog information extraction method, system and computer readable storage medium |
Citations (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1417679A (en) * | 2001-10-21 | 2003-05-14 | 微软公司 | Application abstraction aimed at dialogue |
JP2004233676A (en) * | 2003-01-30 | 2004-08-19 | Honda Motor Co Ltd | Interaction controller |
CN1781140A (en) * | 2003-03-20 | 2006-05-31 | 索尼株式会社 | Audio conversation device, method, and robot device |
CN1881206A (en) * | 2005-06-15 | 2006-12-20 | 富士通株式会社 | Dialog system |
CN101551998A (en) * | 2009-05-12 | 2009-10-07 | 上海锦芯电子科技有限公司 | A group of voice interaction devices and method of voice interaction with human |
US20090265022A1 (en) * | 2008-04-21 | 2009-10-22 | Microsoft Corporation | Playback of multimedia during multi-way communications |
CN102395013A (en) * | 2011-11-07 | 2012-03-28 | 康佳集团股份有限公司 | Voice control method and system for intelligent television |
CN102708454A (en) * | 2012-05-14 | 2012-10-03 | 北京奇虎科技有限公司 | Method and device for providing solution of terminal fault |
CN103413549A (en) * | 2013-07-31 | 2013-11-27 | 深圳创维-Rgb电子有限公司 | Voice interaction method and system and interaction terminal |
CN103456301A (en) * | 2012-05-28 | 2013-12-18 | 中兴通讯股份有限公司 | Ambient sound based scene recognition method and device and mobile terminal |
CN103974366A (en) * | 2014-04-28 | 2014-08-06 | 南京邮电大学 | Wireless body area network routing method based on reinforcement learning |
CN104464733A (en) * | 2014-10-28 | 2015-03-25 | 百度在线网络技术(北京)有限公司 | Multi-scene managing method and device of voice conversation |
CN104506906A (en) * | 2014-11-12 | 2015-04-08 | 科大讯飞股份有限公司 | Voice interaction assisting method and system based on television scene elements and voice assistant |
WO2015125560A1 (en) * | 2014-02-19 | 2015-08-27 | シャープ株式会社 | Information processing device, voice interaction device, and control program |
CN105575386A (en) * | 2015-12-18 | 2016-05-11 | 百度在线网络技术(北京)有限公司 | Method and device for voice recognition |
CN105975511A (en) * | 2016-04-27 | 2016-09-28 | 乐视控股(北京)有限公司 | Intelligent dialogue method and apparatus |
CN106020488A (en) * | 2016-06-03 | 2016-10-12 | 北京光年无限科技有限公司 | Man-machine interaction method and device for conversation system |
CN106356070A (en) * | 2016-08-29 | 2017-01-25 | 广州市百果园网络科技有限公司 | Audio signal processing method and device |
US20170070615A1 (en) * | 2014-02-28 | 2017-03-09 | Dolby Laboratories Licensing Corporation | Perceptually Continuous Mixing in a Teleconference |
CN106528522A (en) * | 2016-08-26 | 2017-03-22 | 南京威卡尔软件有限公司 | Scenarized semantic comprehension and dialogue generation method and system |
CN106847271A (en) * | 2016-12-12 | 2017-06-13 | 北京光年无限科技有限公司 | A kind of data processing method and device for talking with interactive system |
EP3360138B1 (en) * | 2015-10-07 | 2020-02-12 | Robert Bosch GmbH | System and method for audio scene understanding of physical object sound sources |
-
2017
- 2017-07-28 CN CN201710636464.5A patent/CN107644641B/en active Active
Patent Citations (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1417679A (en) * | 2001-10-21 | 2003-05-14 | 微软公司 | Application abstraction aimed at dialogue |
JP2004233676A (en) * | 2003-01-30 | 2004-08-19 | Honda Motor Co Ltd | Interaction controller |
CN1781140A (en) * | 2003-03-20 | 2006-05-31 | 索尼株式会社 | Audio conversation device, method, and robot device |
CN1881206A (en) * | 2005-06-15 | 2006-12-20 | 富士通株式会社 | Dialog system |
US20090265022A1 (en) * | 2008-04-21 | 2009-10-22 | Microsoft Corporation | Playback of multimedia during multi-way communications |
CN101551998A (en) * | 2009-05-12 | 2009-10-07 | 上海锦芯电子科技有限公司 | A group of voice interaction devices and method of voice interaction with human |
CN102395013A (en) * | 2011-11-07 | 2012-03-28 | 康佳集团股份有限公司 | Voice control method and system for intelligent television |
CN102708454A (en) * | 2012-05-14 | 2012-10-03 | 北京奇虎科技有限公司 | Method and device for providing solution of terminal fault |
CN103456301A (en) * | 2012-05-28 | 2013-12-18 | 中兴通讯股份有限公司 | Ambient sound based scene recognition method and device and mobile terminal |
CN103413549A (en) * | 2013-07-31 | 2013-11-27 | 深圳创维-Rgb电子有限公司 | Voice interaction method and system and interaction terminal |
WO2015125560A1 (en) * | 2014-02-19 | 2015-08-27 | シャープ株式会社 | Information processing device, voice interaction device, and control program |
US20170070615A1 (en) * | 2014-02-28 | 2017-03-09 | Dolby Laboratories Licensing Corporation | Perceptually Continuous Mixing in a Teleconference |
CN103974366A (en) * | 2014-04-28 | 2014-08-06 | 南京邮电大学 | Wireless body area network routing method based on reinforcement learning |
CN104464733A (en) * | 2014-10-28 | 2015-03-25 | 百度在线网络技术(北京)有限公司 | Multi-scene managing method and device of voice conversation |
CN104506906A (en) * | 2014-11-12 | 2015-04-08 | 科大讯飞股份有限公司 | Voice interaction assisting method and system based on television scene elements and voice assistant |
EP3360138B1 (en) * | 2015-10-07 | 2020-02-12 | Robert Bosch GmbH | System and method for audio scene understanding of physical object sound sources |
CN105575386A (en) * | 2015-12-18 | 2016-05-11 | 百度在线网络技术(北京)有限公司 | Method and device for voice recognition |
CN105975511A (en) * | 2016-04-27 | 2016-09-28 | 乐视控股(北京)有限公司 | Intelligent dialogue method and apparatus |
CN106020488A (en) * | 2016-06-03 | 2016-10-12 | 北京光年无限科技有限公司 | Man-machine interaction method and device for conversation system |
CN106528522A (en) * | 2016-08-26 | 2017-03-22 | 南京威卡尔软件有限公司 | Scenarized semantic comprehension and dialogue generation method and system |
CN106356070A (en) * | 2016-08-29 | 2017-01-25 | 广州市百果园网络科技有限公司 | Audio signal processing method and device |
CN106847271A (en) * | 2016-12-12 | 2017-06-13 | 北京光年无限科技有限公司 | A kind of data processing method and device for talking with interactive system |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109086860A (en) * | 2018-05-28 | 2018-12-25 | 北京光年无限科技有限公司 | A kind of exchange method and system based on visual human |
CN109086860B (en) * | 2018-05-28 | 2022-03-15 | 北京光年无限科技有限公司 | Interaction method and system based on virtual human |
CN108804603A (en) * | 2018-05-29 | 2018-11-13 | 北京灵智优诺科技有限公司 | Man-machine written dialogue method and system, server, medium |
CN108804603B (en) * | 2018-05-29 | 2021-07-23 | 北京灵智优诺科技有限公司 | Man-machine written dialogue method and system, server and medium |
CN110858226A (en) * | 2018-08-07 | 2020-03-03 | 北京京东尚科信息技术有限公司 | Conversation management method and device |
CN111475206A (en) * | 2019-01-04 | 2020-07-31 | 优奈柯恩(北京)科技有限公司 | Method and apparatus for waking up wearable device |
CN111813900A (en) * | 2019-04-10 | 2020-10-23 | 北京猎户星空科技有限公司 | Multi-turn conversation processing method and device, electronic equipment and storage medium |
CN111813900B (en) * | 2019-04-10 | 2023-12-08 | 北京猎户星空科技有限公司 | Multi-round dialogue processing method and device, electronic equipment and storage medium |
CN110880324A (en) * | 2019-10-31 | 2020-03-13 | 北京大米科技有限公司 | Voice data processing method and device, storage medium and electronic equipment |
WO2021129439A1 (en) * | 2019-12-28 | 2021-07-01 | 科大讯飞股份有限公司 | Voice recognition method and related product |
CN111290953A (en) * | 2020-01-22 | 2020-06-16 | 华为技术有限公司 | Method and device for analyzing test logs |
CN111290953B (en) * | 2020-01-22 | 2021-09-14 | 华为技术有限公司 | Method and device for analyzing test logs |
CN113488036A (en) * | 2020-06-10 | 2021-10-08 | 海信集团有限公司 | Multi-round voice interaction method, terminal and server |
CN111881254A (en) * | 2020-06-10 | 2020-11-03 | 百度在线网络技术(北京)有限公司 | Method and device for generating dialogs, electronic equipment and storage medium |
CN112487170A (en) * | 2020-12-14 | 2021-03-12 | 南京三眼精灵信息技术有限公司 | Scene configuration-oriented human-computer interaction dialogue robot system |
CN112487170B (en) * | 2020-12-14 | 2023-12-15 | 南京三眼精灵信息技术有限公司 | Man-machine interaction dialogue robot system facing scene configuration |
CN113822058A (en) * | 2021-09-18 | 2021-12-21 | 上海明略人工智能(集团)有限公司 | Dialog information extraction method, system and computer readable storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN107644641B (en) | 2021-04-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107644641A (en) | Session operational scenarios recognition methods, terminal and computer-readable recording medium | |
CN105487663B (en) | A kind of intension recognizing method and system towards intelligent robot | |
CN106096576B (en) | A kind of intelligent Service method of robot | |
CN106095833A (en) | Human computer conversation's content processing method | |
CN106649524B (en) | A kind of deep learning intelligent response system of the modified based on computer cloud data | |
CN109189901A (en) | Automatically a kind of method of the new classification of discovery and corresponding corpus in intelligent customer service system | |
CN107862018A (en) | Food cooking method recommendation method and range hood | |
CN109032591A (en) | A kind of crowdsourcing software developer recommended method neural network based | |
CN110091335A (en) | Learn control method, system, equipment and the storage medium with robot | |
CN106649849A (en) | Text information base building method and device and searching method, device and system | |
CN109857844A (en) | Intension recognizing method, device, electronic equipment based on dialog text of ordering | |
CN107368590B (en) | Method, storage medium and application server for recommending questions and answers for user | |
US20120226642A1 (en) | Method and apparatus for considering multi-user preference based on multi-user-criteria group | |
CN108320786A (en) | A kind of Chinese meal vegetable recommendation method based on deep neural network | |
CN113239185B (en) | Method and device for making teaching courseware and computer readable storage medium | |
CN109992781A (en) | Processing, device, storage medium and the processor of text feature | |
CN110222254A (en) | Course recommended method, device, equipment and computer readable storage medium | |
CN111078856A (en) | Group chat conversation processing method and device and electronic equipment | |
Parde et al. | Grounding the meaning of words through vision and interactive gameplay | |
CN110619371A (en) | Food cooking time length recommendation method, storage medium and electric chafing dish | |
CN109918409A (en) | Equipment portrait construction method and device, storage medium and equipment | |
CN108197635A (en) | Cooking mode display method and device and range hood | |
CN107135418A (en) | A kind of control method and device of video playback | |
CN108052858A (en) | Control method of range hood and range hood | |
CN106205288A (en) | A kind of implementation method training robot |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |