33results about How to "Improve dialogue efficiency" patented technology

The invention relates to an on-line question and answer method based on an intelligent robot, which is characterized by comprising the following steps: firstly, a dialogue window body is built between a user and the intelligent robot, and automatic greeting words are sent to a customer service side from a server side; afterwards, the user sends words to the server side, the server side identifies the semantics of the words of the user, combines a matched answer in a question and answer knowledge base, and sends an answer to the customer service side where the user is; after the question of the user is solved, the dialogue ends; finally, the user appraises the intelligent robot and fills in an opinion, and all dialogues during interaction between the user and the server side are recorded into logs. Therefore, the user can enjoy service in an all-weather and no-wait way within 24 hours, and questions encountered in the operation process can be solved by self-service question and answer. Meanwhile, the typewriting time of a customer service staff is saved, the dialogue efficiency is improved, the input of repetitive questions is avoided, the uniform management of common questions and answers is realized, and the intelligent on-line business consultation, business guidance, product sales and other service can be realized.

The invention relates to a conversation robot system. The system comprises a multimedia processing module, a multi-channel access module, a scene decision module, a daily conversation subsystem, a question-answer subsystem, an active conversation system, a task conversation subsystem and a state management module. According to the system, besides the abilities of daily conversation, knowledge questioning and answering and task conversation, a conversation robot further can actively trigger a series of specific conversation skills according to specific scenes, chat with a user, collect and verify answers of the user and execute corresponding service commands.

A computer server may comprise a processor-readable storage medium comprising a set of instructions for matching and broadcasting people-to-search information; and a processor in communication with the storage medium. When executing the set of instructions, the processor is programmed to: provide an online communication platform to a plurality of client devices including a first client device and a second client device; receive people-to-search information from the second device; associate the people-to-search information with the second client device; receive information of a suspected person from the first client device; compare the suspected person information with the people-to-search information; and when the suspected person information matches the people-to-search information, send the suspected person information to the second client device associated with the photo information of the people-to-search.

The invention discloses an information recommendation method and device, a storage medium and a server. The information recommendation method includes the steps: obtaining a determined keyword set which includes at least one keyword; according to a constructed knowledge map, determining at least one candidate sub-node corresponding to each keyword, and a plurality of candidate main nodes corresponding to each candidate sub-node; determining a recommendation degree of each candidate sub-node according to the keyword set and the candidate main nodes; determining a target node from the candidatemain nodes and the candidate sub-node according to the recommendation degree; and generating recommendation information according to the target node and providing the recommendation information to theuser, thus being able to be adapted to various types of conversation scenarios, having a wide application range, greatly reducing the number of conversations and improving the dialogue efficiency.

The invention discloses a voice message processing method and device. The voice message processing method comprises the following steps: acquiring voice information of a voice source and target position information; determining a target translation strategy according to the target position information; translating the voice information according to the target translation strategy to obtain translation information; and outputting the translation information. With the adoption of the voice information processing method, a user can perform translation without repeatedly selecting translation ways; the operation is simple, and the conversation efficiency is high.

The invention belongs to the technical field of artificial intelligence, and provides a short summary generation method, a database construction method and a human-machine conversation method. The short summary generation method based on entity classifications comprises the following steps: obtaining entities, and classifying the entities; selecting proper summary contents according to the classifications of the entities; selecting summary algorithms corresponding to the summary contents, and extracting summary information from encyclopedic pages of the entities; splicing the summary information to obtain a short summary. According to the short summary generation method based on the entity classifications, in a human-machine conversation process, the simplicity and the generalization for answering encyclopedic knowledge-related questions can be improved.

The invention discloses an intelligent conversation method and device, computer equipment and a storage medium, and relates to the field of artificial intelligence. The method can comprise the steps of acquiring conversation information input by a user; and if it is determined that intentions are not expressed in the conversation information and a current state is in a scene corresponding to one intention, executing the following predetermined processing processes of if it is determined that entities are expressed in the conversation information, performing processing for any expressed entity,if it is determined that the entities with the same type as the entity are not collected in the current scene, collecting the entities, otherwise, replacing the collected entities with the same typewith the entity, and clearing the entities mismatched with the entity in other collected entities; and if it is determined that the entities needed to be collected in the current scene are all collected, generating reply verbal tricks according to the intention and the collected entities. By using the scheme, the accuracy and fluency of conversations can be improved.

The invention relates to a conversation robot control method and device, computer equipment and a storage medium. The method comprises the following steps: receiving an input first voice recognition text; determining intention information of the first voice recognition text according to a trained language model, wherein the trained language model is obtained by training conversation sample data; and when the intention information of the first voice recognition text is a pause response, controlling the conversation robot to enter a pause response state. By adopting the method, the conversationefficiency of the conversation robot can be improved.

The invention discloses an intention interruption intelligent voice robot scheme without affecting dialogue management. The intention interruption intelligent voice robot scheme comprises an outboundcontrol center, an NLU module, a DM module, an NLG module and a TTS module, an ASR module with a mark bit, and an interruption filtering module. According to the ASR module, a mark bit is added basedon an original voice-to-character function, and a current ASR result is distinguished through the mark bit; and the interruption filtering module is used for carrying out intention filtering on an ASRresult through a corresponding method when the playStatus of the ASR module with the mark bit is 1, and carrying out interruption intention detection on a user. According to the invention, a problemthat traditional volume detection (VAD) cannot separate noise and client voice is solved by modifying the ASR module and adding the interruption filtering module; accurate and efficient interruption is conducted according to the intention of a client, and the conversation efficiency of the intelligent voice call-out robot and the client experience are improved; and meanwhile, different interruption intentions are configured according to customer categories, so that diversification and individuation are realized, the integration and maintainability of the system are greatly improved, and the maintenance cost of the system is reduced.

A speech dialogue system generates a response sentence in a way to improve the efficiency of the dialogue with the user, based on a result of estimation on an attribute of a proper name in an utterance of a user. The system includes a database attribute estimation unit to estimate the attribute of the input proper name by utilizing a database, and a web attribute estimation unit to estimate an attribute of an input proper name by utilizing information on the web. A reliability integration unit calculates integrated reliability of estimation for each of possible attributes obtained from the estimation by the units, by integrating first reliability of the estimation. A response generation unit generates a response sentence to an input utterance based on the integrated reliabilities of the possible attributes.

The embodiment of the invention discloses a man-machine conversation method and related equipment, and the method comprises the steps: inputting a user question into a similarity model for processing,outputting N1 first similarities, enabling the N1 first similarities to be in one-to-one correspondence with N1 standard questions included in the similarity model, and enabling N1 to be an integer greater than 1; determining an answer strategy feature set corresponding to the N1 standard questions according to the user questions, the N1 standard questions and the N1 first similarities; inputtingthe answer strategy feature set into an answer strategy model for processing, and outputting an answer strategy result; selecting n2 standard questions from the N1 standard questions according to theanswer strategy result, displaying the N2 standard questions wherein N2 is a positive integer. By adopting the embodiment of the invention, efficiency of man-machine conversation can be improved.

The embodiment of the invention provides a dialogue data processing method for multiple rounds of dialogues and a related device, and relates to the field of human-computer interaction. The dialogue data processing method comprises the following steps: in response to user operation, obtaining dialogue data of this round and analyzing a control information set, wherein the analysis control information set represents the dialogue state of the historical dialogue, and the dialogue state is used for controlling the analysis process of the dialogue data; analyzing the dialogue data of this round according to the analysis control information set; and updating the analysis control information set according to the obtained current analysis data and the analysis control information set, and generating current answer data to reply to the user. The dialogue state is used for controlling the analysis process of the dialogue data in this round; and in this round of dialogue, the data of this roundof dialogue can be analyzed and skipped to the next round of dialogue according to the analysis control information set, and the skipping of multiple rounds of dialogues can be completed without designing a large number of fixed process templates in advance, so that the configuration of a multi-round dialogue technology can be simplified, and the dialogue efficiency is improved.

The embodiments of the invention disclose a people finding system based on a mobile terminal and a method thereof. The people finding system includes: a first client which is intended for receiving relevant information and pictures transmitted by a public account docking module, acquiring pictures of a suspect based on the relevant information and pictures, and transmitting the pictures of the suspect to the public account docking module; a second client which is intended for transmitting relevant information and pictures of a to-be-found people to the public account docking module; the public account docking module which is intended for receiving the relevant information and pictures of the to-be-found people transmitted by the second client and receiving the pictures of the suspect transmitted by the first client; a server which is intended for matching the picture of the suspect with the relevant information and pictures of the to-be-found people, and if the matching succeeds, transmitting the relevant information and pictures of the suspect to the second client. According to the embodiments of the invention, the system can increase conversation efficiency, save the time for people finding and increase efficiency of people finding, and increase success rate of people finding.

The invention discloses a method for dynamically acquiring information by a robot. The problems of poor information timeliness and low security are solved. According to the scheme, the called user information can be dynamically collected and utilized, and the information is fed back into conversations, improving robot dialog efficiency, and ensuring the timeliness and safety of the information. Atext translated by the recognition module is analyzed and extracted through the data processor according to a specified mode, or the latest information is obtained by inquiring the user information system at the necessary moment, so that the whole dialogue is more intelligent and efficient, and meanwhile, the timeliness and safety of the user information are guaranteed. The system comprises a client module, an input module, an identification module, a node setting module, a database, a judgment module, a data processor, a condition judgment module, a network communication module, a data querymodule, a remote data service module, a language conversion module, an output module, a feedback module and a repeated setting module, and the client module is connected with the input module.

The invention provides a conversation processing method and device, electronic equipment and a storage medium, and the method comprises the steps: carrying out the first-level intention recognition ofa first round of corpus after the first round of corpus is received, and obtaining a first first-level intention and the confidence degree thereof; based on the confidence degree corresponding to thefirst first-level intention, screening out a second first-level intention of which the confidence degree is greater than a confidence degree threshold value of the questioning intention from the first first-level intention; if the confidence degree of the target first-level intention with the maximum confidence degree in the second first-level intention is smaller than an intention confirmation confidence degree threshold value, asking the target first-level intention back to a user; and if the first reply corpus of the user for the target first-level intention is a confirmation general judgment intention, triggering a target conversation task matched with the target first-level intention. According to the method and the device, the user does not need to switch corpus verbal techniques tore-trigger the intention by asking for the undefined intention triggered by the user, so that the conversation efficiency is improved, the man-machine interaction efficiency is high, and the conversation experience of the user is improved.

The invention relates to the technical field of natural language processing, in particular to an end-to-end target guide type conversation method based on deep learning, which comprises the followingsteps: S1, acquiring a previous round of user conversation and a current conversation history, and initializing a sequence to a sequence model according to the previous round of user conversation andthe current conversation history; S2, determining a row number of a current knowledge base entity; S3, determining a column number of the current knowledge base entity; S4, obtaining an optimal matching entity through an attention mechanism; and S5, iteratively executing the steps S2-S4 until the next round of conversation is finally output. According to the invention, the technical defects of lowend-to-end conversation efficiency and influence on personalized experience of users in the prior art are overcome, and good user service is provided through the accuracy of natural language conversation.

The invention is suitable for the technical field of artificial intelligence, and provides a conversation state determination method, terminal equipment and a storage medium, and the method comprises the steps: responding to a received current input statement inputted by a user in a current round of conversation, and obtaining a system response statement for a previous input statement; selecting candidate domain information matched with the current input statement and the system response statement from a pre-constructed candidate domain information set to serve as current domain information, and selecting candidate slot position information matched with the current input statement and the system response statement from a pre-constructed candidate slot position information set to serve as current slot position information; and inputting the current domain information, the historical domain state information, the current slot position information and the historical slot position state information into a pre-trained dialogue state model to obtain dialogue state information of the current round of dialogue. In addition, the invention also relates to a blockchain technology.

The embodiment of the invention discloses an intention recognition method and device and a storage medium. The method comprises the steps of obtaining a current voice and a previous answer of a user;wherein the previous answer represents an answer having a preset association degree with the current dialect in at least one answer corresponding to at least one voice before the current voice; processing the current voice and the previous answer based on a graph neural network model to obtain features corresponding to the current voice and the previous answer; performing preset times of processing on the features corresponding to the current speaking and the previous answer based on an attention model, and extracting features of preset dimensions in the features corresponding to the current speaking and the previous answer; and determining an intention corresponding to the current speaking of the user based on a feature of a preset dimension in the features corresponding to the current speaking and the previous answer.

The invention relates to a conversation robot system. The system comprises a multimedia processing module, a multi-channel access module, a scene decision module, a daily conversation subsystem, a question-answer subsystem, an active conversation system, a task conversation subsystem and a state management module. According to the system, besides the abilities of daily conversation, knowledge questioning and answering and task conversation, a conversation robot further can actively trigger a series of specific conversation skills according to specific scenes, chat with a user, collect and verify answers of the user and execute corresponding service commands.

The invention relates to an active interactive dialogue robot system and a method for realizing that a dialogue robot makes a dialogue with a user actively. The system comprises a conversation management module, a question generating module, an answer acquiring module, an answer checking module and a command executing module, wherein the conversation management module is used for managing all the questions of the dialogue robot in an active conversation process and managing the conversation state in the active conversation process; the question generating module is used for sending a question to the user according to the conversation state in the conversation management module; and the command executing module is used for executing corresponding service commands after the answer of the user passes check by the dialogue robot. According to the dialogue robot, a large amount of labor cost is saved; and in addition, different national languages or different dialects or voices of stars are adopted for the dialogue robot to communicate with the user, and therefore the interactivity of the whole process is very high, and the user can easily accept and cooperate with the interaction.

The invention belongs to the technical field of artificial intelligence, and provides a conversational method and device based on an aggregation card, computer equipment and a computer readable storage medium. The method comprises the steps of obtaining voice input by a user, performing voice recognition on the voice to obtain a dialogue text corresponding to the voice, performing natural languageprocessing on the dialogue text to obtain an entity and an intention contained in the dialogue text, obtaining target display content corresponding to the intention according to the entity and the intention, and according to the target display content, calling the preset display control corresponding to the target display content, displaying the target display content in the form of a card through the preset display control and the target display content in the form of a card, and also displaying the target display content in the form of a structure so that the preset display control can be automatically assembled according to the target display content. Dynamic aggregation of cards is realized, and convenience of dialogue text display is improved.

A computer keyboard alphabetic layout method involves the prior functional key bottons, cursor control ones, numeral ones and alphabetic ones in their original areas. It features all the alphabitic key bottons are arranged in the order of English alphabet list. This layout improves typist's operation with higher efficience.

The invention discloses a dialogue content analysis method based on regularization and text truncation, which can bring the following advantages: the dialogue content analysis method is put into a working mobile phone to work in a plug-in form, so that multi-machine interchange and universality can be realized; through manual dialogue, dialogue content analysis is carried out mostly through manual repeated listening for voice quality inspection and call audio analysis, so that the labor cost can be reduced, and unnecessary repeated work is avoided; through man-machine conversation, the problems of misjudgment and the like caused by pure manual conversation are avoided, and errors possibly caused by conversation content analysis are reduced; through man-machine conversation, the conversation efficiency is improved, and the problem that manual conversation content analysis cannot be fed back to the system quickly and intervenes in the conversation is solved, and interference on the conversation is avoided; and interaction experience of dialogue content analysis, dialogue analysis speed and accuracy are greatly improved.

The present invention relates to a multi-round intelligent question answering method, system, controller and medium. The method includes: acquiring question sentence information of the current round and previous round; Round's intents and entities, and previous rounds' intents, entities, and answer template types; select the optimal answer based on the obtained current round's intents and entities, and previous rounds' intents, entities, and answer template types. The present invention realizes multiple rounds of question and answer based on intention recognition and entity extraction, improves the accuracy of multiple rounds of question and answer, and reduces maintenance cost.

The invention discloses a multi-domain dialogue generation method, device and equipment based on meta-learning and a medium, which are applied to network marketing and comprise the following steps: acquiring a training dialogue sample and a test dialogue sample of at least one target domain; processing the training dialogue sample of each target field to obtain corresponding training loss; updating the initialized model based on the training loss to obtain an intermediate model; processing the test dialogue sample by adopting the intermediate model to obtain corresponding test loss, and updating the intermediate model based on the test loss to obtain a multi-field dialogue model; crawling published information related to the product keywords, wherein the published information comprises information to be replied; inputting the published information into a multi-domain dialogue model generated based on meta-learning, and obtaining a target dialogue corresponding to the to-be-replied information; and replying the target dialogue to a text reply area corresponding to the to-be-replied information so as to realize automatic marketing in multiple target fields and intelligently generate the target dialogue.