CN112035632A - Preferred distribution method and system suitable for multi-conversation robot collaboration task - Google Patents
Preferred distribution method and system suitable for multi-conversation robot collaboration task Download PDFInfo
- Publication number
- CN112035632A CN112035632A CN202010849121.9A CN202010849121A CN112035632A CN 112035632 A CN112035632 A CN 112035632A CN 202010849121 A CN202010849121 A CN 202010849121A CN 112035632 A CN112035632 A CN 112035632A
- Authority
- CN
- China
- Prior art keywords
- conversation
- skill
- service platform
- distribution
- preference
- 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
- 238000009826 distribution Methods 0.000 title claims abstract description 115
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000013507 mapping Methods 0.000 claims description 12
- 230000002452 interceptive effect Effects 0.000 claims description 10
- 238000005457 optimization Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000002159 abnormal effect Effects 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 2
- 238000013473 artificial intelligence Methods 0.000 abstract description 6
- 238000007726 management method Methods 0.000 description 32
- 230000002776 aggregation Effects 0.000 description 9
- 238000004220 aggregation Methods 0.000 description 9
- 230000004044 response Effects 0.000 description 4
- 230000004927 fusion Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000013145 classification model Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000013179 statistical model Methods 0.000 description 2
- 230000001960 triggered effect Effects 0.000 description 2
- 238000009827 uniform distribution Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000012550 audit Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3343—Query execution using phonetics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Computational Linguistics (AREA)
- Entrepreneurship & Innovation (AREA)
- Strategic Management (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Economics (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Health & Medical Sciences (AREA)
- Acoustics & Sound (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Marketing (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to the technical field of preferred distribution of multi-conversation robots on an artificial intelligence platform, in particular to a preferred distribution method and a preferred distribution system suitable for multi-conversation robot cooperation tasks.
Description
Technical Field
The invention relates to the technical field of preferential distribution of multi-conversation robots on an artificial intelligence platform, in particular to a preferential distribution method and a preferential distribution system suitable for multi-conversation robot cooperation tasks.
Background
Artificial Intelligence (AI) is a new technical science for researching and developing theories, methods, techniques and application systems for simulating, extending and expanding human Intelligence. The main contents of the artificial intelligence subject research comprise: knowledge representation, automatic push and search methods, machine learning and knowledge acquisition, knowledge processing systems, natural language understanding, computer vision, intelligent robots, automatic programming, and the like.
An artificial intelligence platform is a platform system capable of realizing artificial intelligence, and mainly comprises an AIUI open platform, a Baidu UNIT open platform, an AlibaAliGenie open platform, a Cibiz DUI open platform and the like which are communicated with flying.
The existing open platforms have respective advantages and disadvantages, so that the satisfaction degree of a user can be greatly improved if the dominant service resources of each family are aggregated, at the moment, when the user initiates a request through the aggregation platform, the aggregation platform can distribute any open platform for the user through a fixed distribution strategy, but the proper open platform cannot be distributed according to the requirement of the user, so that the problem that the user experience and public praise cannot be improved under the condition that the access load pressure of the aggregation platform rises is caused.
Therefore, a preferred distribution method and a preferred distribution system suitable for the multi-conversation robot collaboration task are produced.
Disclosure of Invention
The invention provides a preferred distribution method and a preferred distribution system suitable for a multi-conversation robot collaboration task, and mainly solves the problems that each existing conversation service platform has respective advantages and disadvantages, but when a plurality of conversation service platforms are aggregated, a proper conversation service platform is difficult to provide according to the requirements of users, so that the user experience and public praise cannot be improved when the access load pressure of the aggregation platform rises.
The invention provides a preferred distribution method suitable for a multi-conversation robot collaboration task, which is characterized by comprising the following steps of:
s1, inputting all conversation service platforms and skill fields in advance, and defining the mapping relation between any conversation service platform and any skill field;
s2, receiving a dialogue request input by a front end, and acquiring the specific request content of the dialogue request;
s3, matching the skill field corresponding to the specific request content, and outputting a skill field identification result;
s4, according to the skill field recognition result, inquiring the preference value of the current user to different conversation service platforms in the current skill field;
s5, inquiring original scores of all the conversation service platforms in the current skill field, and superposing the preference values and the original scores according to a first preset calculation formula to obtain a final score of any one conversation service platform;
and S6, distributing the specific request content to the conversation service platform with the highest final score in the current skill field, and returning corresponding interactive information after receiving the specific request content by the conversation service platform.
Preferably, the step S3 specifically includes:
s31, matching the skill field corresponding to the specific request content, if the matching is successful, outputting the skill field recognition result, and if the matching is failed, executing the next step;
and S32, the specific request content is converted into text, text classification results are identified through skills, whether the text classification results correspond to any skill field is further judged, if yes, the sentence pattern template of the specific request content is extracted and updated to the skill field identification template, and the skill field identification results are output.
Preferably, in the step S4, the preference value is specifically obtained by acquiring a praise frequency, a dialog interruption frequency, and a dialog abnormal frequency of the current user for the specific dialog service platform, and calculating according to a second preset calculation formula.
Preferably, the step S5 specifically includes:
s51, inquiring original scores of all the conversation service platforms in the current skill field according to a preset unified preferred distribution strategy;
s52, obtaining the final score of any one of the conversation service platforms after the preference value and the original score are superposed according to the first preset formula, and forming a first scoring result;
s53, judging whether the first scoring result is consistent with the unified preferred distribution strategy, if so, directly executing the step S6, otherwise, executing the next step;
and S54, counting the dialog service platforms with the highest final score in the skill field, judging whether the adjustment times of any one dialog service platform reach a first preset value, if so, starting online verification, updating the unified preferential distribution strategy, and executing the next step, otherwise, not making a reaction.
Preferably, the step S6 specifically includes:
s61, screening to obtain the dialogue service platform with the highest final score, and converting the specific request content into a format required by the current dialogue service platform;
and S62, the dialogue service platform receives the specific request content and returns the corresponding interactive information.
Preferably, after the step S6, there is provided the step of:
and S7, acquiring and updating the preference value of the user to the current conversation service platform.
The invention also provides a preferred distribution system suitable for the multi-conversation robot collaboration task, which mainly comprises a platform management module, an input module, a user preference management module, a preferred distribution management module and an output module;
the platform management module is used for inputting all the conversation service platforms and the skill field; the system is also used for defining the mapping relation between any one conversation service platform and the skill field;
the input module is used for receiving a conversation request; the system is also used for acquiring the specific request content of the conversation request;
the user preference management module is used for managing preference values of users to each conversation service platform;
the preferred distribution management module is used for storing the original score of the conversation service platform; also for calculating a final score for the conversation service platform; and is further configured to determine a distribution object of the specific requested content;
and the output module is used for outputting the interactive information to the user.
Preferably, the input module comprises a conversation request receiving module and a conversation skill identification module;
the conversation request receiving module is used for receiving a conversation request; the system is also used for acquiring the specific request content of the conversation request;
the conversation skill identification module is used for textualizing the specific request content; the system is also used for storing the concrete request content after the text; text classification results of the specific request content further used for skill recognition texting; and the method is also used for self-learning optimization when the text classification result cannot be matched with any skill field.
Preferably, the user preference management module comprises a user preference learning sub-module and a personal distribution policy preference management module;
the user preference learning submodule is used for storing skill use preference and satisfaction of a user and sending related data to the personal distribution strategy preference management module;
and the personal distribution strategy preference management module is used for generating and managing personal skill distribution strategy preferences of the user.
Preferably, the preferential distribution management module is configured to preset a uniform preferential distribution policy; the system is also used for storing the original score of the conversation service platform; the system is also used for judging an optimal distribution object according to the calculated final score; the system is also used for sending a user request to the optimal distribution object; and the method is also used for self-learning optimization when the optimal distribution object is judged not to meet the unified preferential distribution strategy.
From the above, the following beneficial effects can be obtained by applying the technical scheme provided by the invention:
firstly, the preferred distribution method and the preferred distribution system provided by the invention mainly calculate the optimal distribution object through the preference value of each dialogue service platform by the user, so that the preference of different users can be set, and the user experience is ensured;
secondly, the preferred distribution method and the preferred distribution system provided by the invention only initiate a request to any conversation service platform aiming at the conversation request of the user, thereby greatly reducing the request times and improving the response speed of the conversation service platform.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of a preferred distribution method in embodiment 1 of the present invention;
FIG. 2 is a schematic diagram of the actual operation of the preferred distribution method in embodiment 1 of the present invention;
fig. 3 is a system block diagram of the preferential distribution system in embodiment 2 of the present invention.
Detailed Description
The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
The existing conversation service platforms have respective advantages and disadvantages, but when a plurality of conversation service platforms are aggregated, a proper conversation service platform is difficult to provide according to the requirements of users, so that the problem that the user experience and praise cannot be improved when the access load pressure of the aggregation platform rises is caused.
Example 1
As shown in fig. 1 and fig. 2, in order to solve the above problem, the present embodiment proposes a preferential distribution method suitable for a multi-session robot collaboration task, which mainly includes the following steps:
s1, inputting conversation service platforms and skill fields in advance, and defining the mapping relation between any conversation service platform and any skill field;
s2, receiving a dialogue request input by the front end, and acquiring the specific request content of the dialogue request;
s3, matching the skill field corresponding to the specific request content, and outputting a skill field identification result;
s4, according to the skill field recognition result, inquiring the preference value of the current user to different conversation service platforms in the current skill field;
s5, inquiring original scores of all the conversation service platforms in the current skill field, and superposing the preference values and the original scores according to a first preset calculation formula to obtain a final score of any conversation service platform;
and S6, distributing the specific request to a conversation service platform with the highest final score in the current skill field, and returning corresponding interactive information after receiving the specific request by the conversation service platform.
Preferably, but not limited to, the different dialog service platforms in step S1 may have the same skills, that is, the different dialog service platforms should belong to at least one skill field, for example, music skills, and may be performed by the following procedures:
preferably, but not limited to, the dialog request input at the front end in step S2 needs to be preprocessed by text command, i.e. stop words and wireless auxiliary words, to obtain the specific request content of the dialog request, and if the text command is "please play a liudebua" ice rain "bar | for me |)! The dialogue request receiving module defines a stop word bank and an invalid auxiliary word bank, and after the operation of stopping words and invalid auxiliary words, the original text request is converted into the output of 'playing liu de hua ice rain', and the following procedures can be referred.
More specifically, step S3 specifically includes:
s31, matching the skill field corresponding to the specific request content, if the matching is successful, outputting a skill field recognition result, and executing the step S4, if the matching is failed, executing the next step;
and S32, the specific request content is converted into text, the text classification result is identified through skills, whether the text classification result corresponds to any skill field is further judged, if yes, the sentence pattern template of the specific request content is extracted and updated to the corresponding skill field identification module, and the skill field identification result is output.
Preferably, but not limited to, in the embodiment, when the skill field corresponding to the specific request content is determined, the specific request content is determined to correspond to the current skill field by determining that the sentence pattern template of the specific request content coincides with any template in the current skill field identification templates, so that accurate identification of the aggregation platform to the skill field in the embodiment can be ensured.
The specific identification process of step S3 in this embodiment is: firstly, reading the skill intentions of the specific request contents in the previous round, then carrying out current skill identification on the request texts and the intentions in the previous round, and if the matching is successful, directly outputting the skill identification result; if the request is unsuccessful, the specific request content is sent to a text classification module for skill identification, the text classification result is identified as the current affiliated skill and output, if the skill identified by the text classification is consistent with the previous round of conversation skill, the request text and the corresponding skill are automatically updated and audited, and after manual audit and sentence pattern template extraction, a template is added, so that the continuous optimization of the accuracy of the recognizer is completed. The text classification module can be realized based on the existing statistical or deep learning model, and supposing that a skill classification model is trained based on 100 skills corpus at present, the multi-round dialogue intention matcher mainly matches whether the current request belongs to the multi-round dialogue request according to the stored mapping of the previous-round skill-current request sentence pattern template, the mapping format is as follows, and if the multi-round dialogue intention matcher is not matched with the template of the request, the multi-round dialogue intention matcher is delivered to the text classification module to be identified as the music skill.
More specifically, in step S4, the preference value is obtained by specifically acquiring the praise frequency, the dialog interruption frequency, and the dialog abnormal frequency of the current user for the specific dialog service platform, and calculating according to a second preset calculation formula.
Preferably, but not limited to, the obtained praise frequency, the session interruption frequency and the session abnormal frequency are mainly scores of the user for the session service platform before the user is on the aggregation platform, the second preset calculation formula is,
the preference score of music skills distributed to a certain conversation service platform is user's praise frequency 1-conversation interruption frequency 2-conversation abnormal frequency 3.
More specifically, step S5 specifically includes:
s51, inquiring the original scores of all the conversation service platforms in the current skill field according to a preset unified preferred distribution strategy;
s52, obtaining the final score of any conversation service platform after superposing the preference value and the original score according to a first preset formula, and forming a first scoring result;
s53, judging whether the first scoring result is consistent with the unified preferred distribution strategy, if so, directly executing the step S6, otherwise, executing the next step;
and S54, counting the dialog service platforms with the highest final score in the skill field, judging whether the adjustment times of any dialog service platform reach a first preset value, if so, starting online auditing, updating a unified preferential distribution strategy, and executing the next step, otherwise, not making a response.
Preferably, but not limited to, the first preset calculation formula:
the final score of the music skill distributed to the conversation service platform is the original score of the music skill distributed to the conversation service platform 4+ the preference score of the music skill distributed to the conversation service platform 5.
Preferably, but not limited to, the unified preferential distribution strategy is a ranking list of each session service platform obtained according to the original score, and the ranking list of the ranking list is a distribution object recommended to the user according to the unified preferential distribution strategy, so that when a first scoring result obtained by referring to the preference value of the user is different from the unified preferential distribution strategy in this embodiment, the optimal distribution object of the first scoring result should be preferentially considered.
In the present embodiment, the specific operation procedure of step S54 is as follows:
if the optimal distribution object screened out through the final scoring is inconsistent with the optimal value in the previous uniform distribution strategy, the technical field and the new optimal distribution object are stored, the adjustment condition of the distribution strategy by an online user is counted, if the adjustment frequency of a certain technical distribution object is greater than a first preset value, an online auditing function is automatically triggered, the dynamic update of the preferred distribution strategy is completed through manual auditing, and the preferred distribution strategy configuration file format is as follows:
more specifically, step S6 specifically includes:
s61, screening out the dialogue service platform with the highest final score, and converting the specific request content into the format required by the current dialogue service platform;
and S62, the dialogue service platform receives the specific request content and returns corresponding interactive information.
In the embodiment, an optimal distribution object (a conversation service platform) is screened out according to the final score of each conversation service platform, and then specific request content is packaged into a format required by the conversation service platform to be distributed and is distributed to the optimal conversation service platform for processing; after the conversation service platform returns the result, the semantic result of the conversation service management platform is called to be converted into a uniform semantic response format, meanwhile, the interactive information to be issued is packaged into a final result and is fed back to the front end, and the specific output format is as follows:
more specifically, after step S6, there is provided the step of:
and S7, acquiring and updating the preference value of the user for the current conversation service platform.
In the present embodiment, the preference value updated in step S7 is mainly used to calculate the first scoring result in step S52, so that the optimal distribution policy in the present embodiment can be adjusted according to the preferences of different users.
Example 2
As shown in fig. 3, in order to solve the foregoing problem, the present embodiment proposes a preferential distribution system suitable for multi-session robot collaboration tasks, which mainly includes a platform management module 10, an input module 20, a user preference management module 30, a preferential distribution management module 40, and an output module 50. The platform management module 10 is used for inputting all conversation service platforms and skill fields; the system is also used for defining the mapping relation between any conversation service platform and the skill field; an input module 20 for receiving a dialogue request; the system is also used for acquiring the specific request content of the conversation request; the user preference management module 30 is used for managing preference values of the user to each conversation service platform; the preferred distribution management module 40 is used for storing the original scores of the conversation service platforms; also used for calculating the final score of the dialogue service platform; also for determining a distribution object of the specific requested content; and an output module 50 for outputting the interactive information to the user.
Preferably, but not limited to, the preferred classification system in this embodiment includes a front end and a back end, and the platform management module 10 is disposed on the back end.
Preferably, but not limited to, the platform management module 10 mainly includes a skill/support dialog service platform mapping submodule, a voice request format conversion submodule, and a voice request result adaptation and fusion submodule, where the skill/support dialog service platform mapping submodule is used for taking charge of information management of each dialog service platform, and mapping of semantic skills and each dialog service platform (corresponding to step S1 in embodiment 1); a voice request format conversion sub-module, configured to take charge of specific format conversion of semantic requests of each conversational service platform (corresponding to step S61 in embodiment 1); and a voice request result adaptive fusion sub-module, configured to take charge of unified adaptive fusion of semantic response results of each session service platform (corresponding to step S62 in embodiment 1).
More specifically, the input module 20 includes a conversation request receiving module and a conversation skill recognition module; the conversation request receiving module is used for receiving a conversation request; the system is also used for acquiring the specific content of the conversation request; the conversation skill identification module is used for textualizing specific request content; the system is also used for storing the concrete request content after the text; text classification results for the specific request content that is also used for skill recognition texting; and the method is also used for self-learning optimization when the text classification result cannot be matched with any skill field.
Preferably, but not limited to, the preferred classification system in this embodiment includes a front end and a back end, wherein the session request receiving module is disposed at the front end, and the session skill recognition module is disposed at the back end.
Preferably, but not limited to, the dialogue skill recognition module can be used for meeting multiple turns of dialogue intention matching, can also be used for recognizing corresponding skills of current specific request contents by combining multiple turns of dialogue intention matching results, and can also be used for self-learning optimization in charge of multiple turns of dialogue intention recognition and skill recognition accuracy.
In this embodiment, the multiple rounds of dialogs can be understood that the specific request content is a sentence template with composite keywords, that is, when the aggregation system receives the specific request content, the keywords are extracted first, the skill fields of the keywords are identified one by one, and then intersection points of the skill fields are found out to determine the exact skill field of the specific request content. Therefore, the technical means of multi-turn dialog is applied in this embodiment.
In this embodiment, the input module 20 reads the skill intention of the previous round, then sends the request text and the previous round of intention to the multi-round dialog intention matching recognizer for performing the current skill recognition, and if the matching is successful, directly outputs the skill recognition result; if the answer is not successful, the request text is sent to a text classification module for skill identification, the text classification result is identified as the current affiliated skill and output, if the skill identified by the text classification is consistent with the previous round of conversation skill, the request text and the corresponding skill are automatically submitted to a multi-round conversation intention matcher for updating and checking, after manual checking and sentence pattern template extraction, a template is additionally added to the multi-round conversation intention matching template, and the continuous optimization of the accuracy of the multi-round conversation intention matching identifier is completed. The text classification module can be implemented based on the existing statistical or deep learning model, and assuming that a skill classification model is trained based on 100-skill corpus, the multi-turn dialog intention matcher mainly matches whether the current request belongs to the multi-turn dialog request according to the stored mapping of the previous-turn skill-current request sentence pattern template.
More specifically, the user preference management module 30 includes a user preference learning sub-module and a personal distribution policy preference management module. The user preference learning submodule is used for storing skill use preference and satisfaction of a user and sending related data to the personal distribution strategy preference management module; and the personal distribution strategy preference management module is used for generating and managing personal skill distribution strategy preferences of the user.
Preferably, but not limited to, the preferential distribution system of the embodiment includes a front end and a back end, the user preference learning sub-module is disposed at the front end, and the personal distribution policy preference management module is disposed at the back end.
Preferably, but not limited to, after receiving the specific request content/the textual specific request content, the user preference management module 30 queries the personal distribution policy preference configuration, obtains a distribution satisfaction score of the current user in the current skill field, and the score is obtained by the user preference learning module according to the history data of the manual approval frequency/the interruption frequency of the conversation task/the abnormal frequency of the conversation task of the user learned by the front end in the previous using process, and scoring according to the user preference learning scoring rule.
Preferably, but not limited to, after the front-end user receives the interaction information, the user preference learning module may count the praise action, the interruption condition of the session flow, and whether the session is abnormal, etc. of the front-end user, and timely return the counted information to the user preference management module 30 for updating the distribution preference policy score.
More specifically, the preferential distribution management module 40 is configured to preset a uniform preferential distribution policy; the system is also used for storing the original scores of the conversation service platforms; the system is also used for judging an optimal distribution object according to the calculated final score; the system is also used for sending a user request to the optimal distribution object; and the method is also used for self-learning optimization when the optimal distribution object is judged not to meet the unified preferential distribution strategy.
In this embodiment, after receiving the request text, the skill information to be distributed, and the user skill distribution preference score, the preferential distribution management module 40 firstly queries a uniform preferential distribution policy configuration according to the skill information to be distributed, obtains an original score of each distribution object (session service platform) corresponding to the skill, and then adds the user skill distribution preference score to the original score of the distribution object according to a certain weight to obtain a new distribution object score; then, screening out an optimal distribution object (a conversation service platform) according to the final score of each conversation service platform, calling a semantic request format conversion module of a conversation service platform management module 10 to package the request parameters into a format required by the conversation service platform to be distributed, and distributing the format to the optimal conversation service platform for processing; if the optimal distribution object screened out through the final score is inconsistent with the optimal value in the previous uniform distribution strategy, the skill and the new optimal distribution object are stored in a self-learning module, the self-learning module can count the adjustment condition of the distribution strategy by an online user, if the adjustment frequency of a certain skill distribution object is greater than a preset threshold value, the online auditing function of the self-learning module can be automatically triggered, and the dynamic updating of the preferential distribution strategy is completed through manual auditing.
In summary, according to the preferred distribution method and system applicable to the multi-session robot collaboration task provided by the embodiment, the most suitable session service platform is provided for different users in a targeted manner mainly by referring to the preference conditions of the users to different session service platforms, so that the preference management of the aggregation platform to the users is improved, and the experience degree of the users is further improved.
The above-described embodiments do not limit the scope of the present invention. Any modification, equivalent replacement, and improvement made within the spirit and principle of the above-described embodiments should be included in the protection scope of the technical solution.
Claims (10)
1. A preferential distribution method suitable for a multi-conversation robot collaboration task is characterized by comprising the following steps:
s1, inputting all conversation service platforms and skill fields in advance, and defining the mapping relation between any conversation service platform and any skill field;
s2, receiving a dialogue request input by a front end, and acquiring the specific request content of the dialogue request;
s3, matching the skill field corresponding to the specific request content, and outputting a skill field identification result;
s4, according to the skill field recognition result, inquiring the preference value of the current user to different conversation service platforms in the current skill field;
s5, inquiring original scores of all the conversation service platforms in the current skill field, and superposing the preference values and the original scores according to a first preset calculation formula to obtain a final score of any one conversation service platform;
and S6, distributing the specific request content to the conversation service platform with the highest final score in the current skill field, and returning corresponding interactive information after receiving the specific request content by the conversation service platform.
2. The preferential distribution method applicable to multi-dialogue robot collaboration tasks as claimed in claim 1, wherein the step S3 specifically comprises:
s31, matching the skill field corresponding to the specific request content, if the matching is successful, outputting the skill field recognition result, and executing the step S4, if the matching is failed, executing the next step;
and S32, the specific request content is converted into text, text classification results are identified through skills, whether the text classification results correspond to any skill field is further judged, if yes, the sentence pattern template of the specific request content is extracted and updated to the skill field identification template, and the skill field identification results are output.
3. The preferential distribution method suitable for the multi-conversation robot collaboration task as claimed in claim 2, wherein:
in the step S4, the preference value is specifically obtained by obtaining the praise frequency, the dialog interruption frequency, and the dialog abnormal frequency of the current user for the specific dialog service platform, and calculating according to a second preset calculation formula.
4. The preferential distribution method suitable for multi-dialogue robot collaboration tasks as claimed in claim 3, wherein the step S5 specifically comprises:
s51, inquiring original scores of all the conversation service platforms in the current skill field according to a preset unified preferred distribution strategy;
s52, obtaining the final score of any one of the conversation service platforms after the preference value and the original score are superposed according to the first preset formula, and forming a first scoring result;
s53, judging whether the first scoring result is consistent with the unified preferred distribution strategy, if so, directly executing the step S6, otherwise, executing the next step;
and S54, counting the dialog service platforms with the highest final score in the skill field, judging whether the adjustment times of any one dialog service platform reach a first preset value, if so, starting online verification, updating the unified preferential distribution strategy, and executing the next step, otherwise, not making a reaction.
5. The preferential distribution method suitable for multi-dialogue robot collaboration tasks as claimed in claim 4, wherein the step S6 specifically comprises:
s61, screening to obtain the dialogue service platform with the highest final score, and converting the specific request content into a format required by the current dialogue service platform;
and S62, the dialogue service platform receives the specific request content and returns the corresponding interactive information.
6. A preferred distribution method suitable for multi-dialogue robot collaboration tasks according to any claim 1-5, wherein after the step S6, the method comprises the following steps:
and S7, acquiring and updating the preference value of the user to the current conversation service platform.
7. A preferential distribution system suitable for multi-conversation robot collaboration tasks is characterized in that: the system comprises a platform management module, an input module, a user preference management module, a preferred distribution management module and an output module;
the platform management module is used for inputting all the conversation service platforms and the skill field; the system is also used for defining the mapping relation between any one conversation service platform and the skill field;
the input module is used for receiving a conversation request; the system is also used for acquiring the specific request content of the conversation request;
the user preference management module is used for managing preference values of users to each conversation service platform;
the preferred distribution management module is used for storing the original score of the conversation service platform; also for calculating a final score for the conversation service platform; and is further configured to determine a distribution object of the specific requested content;
and the output module is used for outputting the interactive information to the user.
8. A preferential distribution system for multi-session robotic collaboration tasks as claimed in claim 7 wherein: the input module comprises a conversation request receiving module and a conversation skill identification module;
the conversation request receiving module is used for receiving a conversation request; the system is also used for acquiring the specific request content of the conversation request;
the conversation skill identification module is used for textualizing the specific request content; the system is also used for storing the concrete request content after the text; text classification results of the specific request content further used for skill recognition texting; and the method is also used for self-learning optimization when the text classification result cannot be matched with any skill field.
9. A preferential distribution system for multi-session robotic collaboration tasks as claimed in claim 8 wherein: the user preference management module comprises a user preference learning sub-module and an individual distribution strategy preference management module;
the user preference learning submodule is used for storing skill use preference and satisfaction of a user and sending related data to the personal distribution strategy preference management module;
and the personal distribution strategy preference management module is used for generating and managing personal skill distribution strategy preferences of the user.
10. A preferential distribution system for multi-session robotic collaboration tasks as claimed in claim 9 wherein:
the preferred distribution management module is used for presetting a uniform preferred distribution strategy; the system is also used for storing the original score of the conversation service platform; the system is also used for judging an optimal distribution object according to the calculated final score; the system is also used for sending a user request to the optimal distribution object; and the method is also used for self-learning optimization when the optimal distribution object is judged not to meet the unified preferential distribution strategy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010849121.9A CN112035632B (en) | 2020-08-21 | 2020-08-21 | Preferential distribution method and system suitable for multi-dialogue robot cooperation tasks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010849121.9A CN112035632B (en) | 2020-08-21 | 2020-08-21 | Preferential distribution method and system suitable for multi-dialogue robot cooperation tasks |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112035632A true CN112035632A (en) | 2020-12-04 |
| CN112035632B CN112035632B (en) | 2024-08-02 |
Family
ID=73580383
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010849121.9A Active CN112035632B (en) | 2020-08-21 | 2020-08-21 | Preferential distribution method and system suitable for multi-dialogue robot cooperation tasks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112035632B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113626568A (en) * | 2021-07-30 | 2021-11-09 | 平安普惠企业管理有限公司 | Man-machine conversation control method and device for robot, computer equipment and medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090167553A1 (en) * | 2007-12-30 | 2009-07-02 | Jin Hong | Open Mobile Online Reservation and Ordering Systems |
| CN109408799A (en) * | 2018-08-14 | 2019-03-01 | 优视科技（中国）有限公司 | Semantic decision-making technique and system |
| US20190158666A1 (en) * | 2017-11-17 | 2019-05-23 | Thrio, Inc. | Ai-based compliance and preference system |
| CN109977208A (en) * | 2019-03-22 | 2019-07-05 | 北京中科汇联科技股份有限公司 | It is a kind of to merge FAQ and task and the actively conversational system of guidance |
| CN110209791A (en) * | 2019-06-12 | 2019-09-06 | 百融云创科技股份有限公司 | It is a kind of to take turns dialogue intelligent speech interactive system and device more |
| US20190347363A1 (en) * | 2018-05-09 | 2019-11-14 | International Business Machines Corporation | Using a dialog system for integrating multiple domain learning and problem solving |
| CN111078846A (en) * | 2019-11-25 | 2020-04-28 | 青牛智胜(深圳)科技有限公司 | Multi-turn dialog system construction method and system based on business scene |
| CN111161717A (en) * | 2019-12-26 | 2020-05-15 | 苏州思必驰信息科技有限公司 | Skill scheduling method and system for voice conversation platform |
| CN111191000A (en) * | 2019-12-20 | 2020-05-22 | 北京淇瑀信息科技有限公司 | Dialog management method, device and system of intelligent voice robot |
| CN111209363A (en) * | 2019-12-25 | 2020-05-29 | 华为技术有限公司 | Corpus data processing method, apparatus, server and storage medium |
-
2020
- 2020-08-21 CN CN202010849121.9A patent/CN112035632B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090167553A1 (en) * | 2007-12-30 | 2009-07-02 | Jin Hong | Open Mobile Online Reservation and Ordering Systems |
| US20190158666A1 (en) * | 2017-11-17 | 2019-05-23 | Thrio, Inc. | Ai-based compliance and preference system |
| US20190347363A1 (en) * | 2018-05-09 | 2019-11-14 | International Business Machines Corporation | Using a dialog system for integrating multiple domain learning and problem solving |
| CN109408799A (en) * | 2018-08-14 | 2019-03-01 | 优视科技（中国）有限公司 | Semantic decision-making technique and system |
| CN109977208A (en) * | 2019-03-22 | 2019-07-05 | 北京中科汇联科技股份有限公司 | It is a kind of to merge FAQ and task and the actively conversational system of guidance |
| CN110209791A (en) * | 2019-06-12 | 2019-09-06 | 百融云创科技股份有限公司 | It is a kind of to take turns dialogue intelligent speech interactive system and device more |
| CN111078846A (en) * | 2019-11-25 | 2020-04-28 | 青牛智胜(深圳)科技有限公司 | Multi-turn dialog system construction method and system based on business scene |
| CN111191000A (en) * | 2019-12-20 | 2020-05-22 | 北京淇瑀信息科技有限公司 | Dialog management method, device and system of intelligent voice robot |
| CN111209363A (en) * | 2019-12-25 | 2020-05-29 | 华为技术有限公司 | Corpus data processing method, apparatus, server and storage medium |
| CN111161717A (en) * | 2019-12-26 | 2020-05-15 | 苏州思必驰信息科技有限公司 | Skill scheduling method and system for voice conversation platform |
Non-Patent Citations (2)
| Title |
|---|
| 李宏基: "物流服务平台中推荐系统的框架设计", 科技传播, no. 19, 31 October 2010 (2010-10-31), pages 235 * |
| 薛伟莲等: "一种基于对话的电子商务推荐系统", 辽宁师范大学学报(自然科学版), vol. 34, no. 02, 30 June 2011 (2011-06-30), pages 175 - 178 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113626568A (en) * | 2021-07-30 | 2021-11-09 | 平安普惠企业管理有限公司 | Man-machine conversation control method and device for robot, computer equipment and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112035632B (en) | 2024-08-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10319381B2 (en) | Iteratively updating parameters for dialog states | |
| CN107146610B (en) | Method and device for determining user intention | |
| US20180004729A1 (en) | State machine based context-sensitive system for managing multi-round dialog | |
| CN104538031B (en) | Intelligent sound service development cloud platform and method | |
| CN107025297A (en) | A kind of chat robots and its automatic chatting method | |
| US20200051568A1 (en) | Tailoring an interactive dialog application based on creator provided content | |
| US20040249638A1 (en) | Method and apparatus for performing plan-based dialog | |
| WO2008128423A1 (en) | An intelligent dialog system and a method for realization thereof | |
| WO2008119270A1 (en) | A chatting robot system and a method, a device for chatting automatically | |
| CN111161726B (en) | Intelligent voice interaction method, device, medium and system | |
| CN111611382A (en) | Dialect model training method, dialog information generation method, device and system | |
| US11995523B2 (en) | Systems and methods for determining training parameters for dialog generation | |
| CN111178081A (en) | Semantic recognition method, server, electronic device and computer storage medium | |
| CN111128175B (en) | Spoken language dialogue management method and system | |
| CN114238607A (en) | Deep interactive AI intelligent job-searching consultant method, system and storage medium | |
| US11893310B2 (en) | System command processing | |
| DE112022000504T5 (en) | Interactive content delivery | |
| CN117909483B (en) | Personalized dialogue method, system, equipment and medium | |
| CN112035632A (en) | Preferred distribution method and system suitable for multi-conversation robot collaboration task | |
| CN113132214A (en) | Conversation method, device, server and storage medium | |
| CN113111157A (en) | Question-answer processing method, device, computer equipment and storage medium | |
| CN114490994B (en) | Conversation management method and device | |
| CN113689851B (en) | Scheduling professional language understanding system and method | |
| CN107943299A (en) | Emotion rendering method and device, computer equipment and computer-readable recording medium | |
| CN114238606A (en) | Intelligent voice customer service robot conversation management method and device |
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 |