CN112291430B

CN112291430B - Intelligent response method and device based on identity confirmation

Info

Publication number: CN112291430B
Application number: CN202011147707.7A
Authority: CN
Inventors: 李国华; 张伟萌
Original assignee: Beijing Moran Cognitive Technology Co Ltd
Current assignee: Beijing Moran Cognitive Technology Co Ltd
Priority date: 2020-10-23
Filing date: 2020-10-23
Publication date: 2021-07-30
Anticipated expiration: 2040-10-23
Also published as: CN112291430A

Abstract

The invention discloses an intelligent response method based on identity confirmation, which comprises the steps of receiving call information, and determining a calling telephone number according to the call information; acquiring attribute information of the calling telephone number according to the calling telephone number; judging whether the attribute information of the calling telephone number belongs to a first attribute classification; if the attribute information of the calling telephone number does not belong to the first attribute classification, starting a first timer, and if the first timer is overtime and the user does not answer, executing S305; s305, executing call takeover, and determining a first answering task logic according to the attribute information of the calling phone number; receiving first voice information of a calling party, recording the voice information, and converting the voice information into first dialogue data; and determining candidate information of the relevant slot position in the first answering task logic according to the first answering data. By the method, a better intelligent response function can be realized, and user experience is improved.

Description

Intelligent response method and device based on identity confirmation

Technical Field

The embodiment of the invention relates to the technical field of information processing, in particular to an intelligent response method and device based on identity confirmation.

Background

With the development and popularization of computer technology, intelligent technologies such as human-computer interaction provide convenient and fast services in various aspects of people's life. Human-Computer Interaction (HCI) technology refers to a technology for realizing Human-Computer Interaction in an efficient manner through Computer input and output devices. At present, the function and the training mode of the voice assistant are relatively limited, so that the physical examination of a user is poor. How to optimize the automatic learning ability and the automatic processing user task ability of the voice assistant becomes a problem to be solved urgently.

Disclosure of Invention

Aiming at the problems in the prior art, the invention provides an intelligent response method based on identity confirmation, which is characterized by comprising the following steps:

step 301, receiving call information, and determining a calling telephone number according to the call information;

step 302, acquiring attribute information of the calling phone number according to the calling phone number;

step 303, judging whether the attribute information of the calling telephone number belongs to a first attribute classification; if the attribute information of the calling telephone number belongs to the first attribute classification, executing step 305, and if the attribute information of the calling telephone number does not belong to the first attribute classification, executing step 304;

step 304, starting a first timer, and if the first timer is overtime and the user does not answer, executing step 305;

step 305, executing call takeover, and determining a first answering task logic according to the attribute information of the calling telephone number;

step 306, receiving first voice information of a calling party, recording the voice information, and converting the voice information into first dialogue data;

step 307, determining candidate information of a relevant slot in the first listening task logic according to the first dialogue data.

The invention also provides an intelligent answering device based on identity confirmation, which is characterized by comprising:

the access module receives call information and determines a calling telephone number according to the call information;

the attribute information determining module is used for acquiring the attribute information of the calling telephone number according to the calling telephone number;

the attribute classification determining module is used for judging whether the attribute information of the calling telephone number belongs to a first attribute classification; if the attribute information of the calling telephone number belongs to the first attribute classification, sending a notification message to a call takeover module, and if the attribute information of the calling telephone number does not belong to the first attribute classification, sending the notification message to a timing monitoring module;

the timing monitoring module starts a first timer, and if the first timer is overtime and a user does not answer, a notification message is sent to the call takeover module;

the call takeover module executes call takeover and determines a first answering task logic according to the attribute information of the calling telephone number;

the control module is used for receiving first voice information of a calling party, recording the voice information and converting the voice information into first dialogue data;

and the control module determines candidate information of a relevant slot position in the first answering task logic according to the first dialogue data.

By the method and the device, the intelligent response function of the automatic assistant can be realized, the optimized intelligent response based on identity confirmation is realized, meanwhile, the real-time visual conversation display is provided, the convenient conversation intervention and the training optimization of the automatic assistant are realized, and the user experience is improved.

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 introduced below, and it is obvious that the drawings in the following description are 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 method of training an automated assistant in one embodiment of the invention.

FIG. 2 is a schematic of association logic and task logic in an automated assistant in one embodiment of the invention.

Fig. 3 is an intelligent response method based on identity confirmation in an embodiment of the present invention.

FIG. 4 is a dialog processing interface in one embodiment of the invention.

FIG. 5 is a method for automatic learning of a voice assistant in one embodiment of the invention.

FIG. 6 is a method for a voice assistant to place an intelligent call in accordance with an embodiment of the present invention.

Fig. 7 is an AR-based smart phone method according to an embodiment of the present invention.

FIG. 8 is a method for automatic learning by a voice assistant in one embodiment of the invention.

FIG. 9 is an automated assistant's training device in one embodiment of the invention.

Fig. 10 is an intelligent answering device based on identity confirmation in an embodiment of the invention.

FIG. 11 is an apparatus for automated learning of a voice assistant, in accordance with an embodiment of the present invention.

FIG. 12 is an intelligent calling device of a voice assistant in one embodiment of the invention.

Fig. 13 is an AR-based smart communicator in an embodiment of the present invention.

Fig. 14 is an automatic learning apparatus of a voice assistant in an embodiment of the invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings. The embodiments and specific features of the embodiments of the present invention are detailed descriptions of technical solutions of the embodiments of the present invention, and are not limited to technical solutions of the present invention, and the technical features of the embodiments and the embodiments of the present invention may be combined with each other without conflict.

The method can be applied to any device or equipment with voice interaction capacity, such as computers, mobile phones, tablet computers, car machines, vehicle-mounted terminals, intelligent sound boxes, set top boxes, intelligent household appliances and the like.

Example one

Referring to fig. 1, an embodiment of the present invention provides a training method for an automated assistant, which is characterized in that:

step 101, collecting calling telephone number information during incoming call, classifying and storing the calling numbers, and setting a calling number attribute matching relation table aiming at the calling numbers;

step 102, collecting call voice information, and storing the voice information in a classified manner according to the calling telephone number;

103, recognizing the voice information, distinguishing a conversation body, and marking conversation data in the voice information according to a telephone number corresponding to the conversation body;

step 103 further comprises storing the dialog data in the form of tagged text in an incoming call data repository, wherein the telephone number of the calling party is used as a primary index, and the call ending time of the call incoming from the calling party is used as a secondary index;

104, clustering the dialogue data according to the number attribute corresponding to the calling telephone number in the number attribute matching relation table;

105, analyzing the dialogue data in each cluster, extracting key data in the dialogue, and creating a first initial answering task logic based on the cluster; associating the first initial listening task logic with the corresponding calling number and number attribute information.

Wherein the automated assistant may be a voice assistant.

For example, for the incoming phone number 137 × 1234, first, the relevant information of the address book may be matched, for example, the phone number is identified as property in the address book, and then the number attribute item in the number attribute matching relation table is filled according to the identification result, for example, the supplementary number attribute is property.

If the incoming phone number 957 is not matched in the address list, the number is inquired as an express number through networking, and the number attribute item in the number attribute matching relation table is updated according to the identification result, if the number attribute corresponding to the number is supplemented as an express.

The calling number attribute matching relationship table may be as follows:

TABLE 1

Calling number	Short number	Prefix number	Number attribute	Association class
					137****1234	125831137****1234	Property industry
957****			Express delivery	10
					158****0001		Family member

For special numbers, both the number carrying the prefix and/or the short number, and the original number, e.g. the number carrying the prefix, are stored, and the original number 137 × 1234 is determined by separating the prefixes, e.g. 125831137 × 1234; for the short number, determining an original number by inquiring the corresponding relation of the short number; if there is no short number and prefix number, the above item in the number attribute matching relation table may be left blank, or the item may be omitted.

When the same number has two kinds of attribute information, the priority can be configured for the number attribute, such as the high priority is family class, the next highest priority is work class, the low priority is service class, etc.

The association type item in the calling number attribute matching relation table can be whether a subsequent association operation exists for the calling number or not; information description or identifier designation can be adopted, for example, if the number is none or 01, the number represents that no subsequent association processing exists in history statistics after the number is called; further, the initial associated category items may all be null or 00, supplementing the item based on subsequent data updates. And when the association item is that connection exists or is marked as 10, indicating that subsequent association processing exists in the history statistics after the calling number calls this time.

When the association item is marked as 01, the voice assistant may temporarily not start the association logic to monitor the session data of the call, and when the association item is marked as 10 or an initial value 00, the voice assistant may synchronously start the association logic to monitor the session data of the call.

Specifically, the user may authorize the starting of the automatic assistant training function, authorize the call authority of the user terminal application, such as starting the automatic assistant training function, authorize the answering or dialing of a call to the call module, and record a call.

The user can set that only the related recording or the processed dialogue data is stored locally, or a private cloud data storage area is adopted for data storage.

Specifically, when the user answers the incoming call 957, it is determined that the call is an express call, and a recording session is recorded by starting a recording function: if the calling party is X express, 2 express are informed by the telephone, and whether XX express points are needed to be released or delivery is needed. The user is at home with the desire to home in the delivery. And determining voice information of two parties of the conversation by recording the conversation and based on voiceprint analysis.

Clustering the dialogue data by counting historical incoming voice data, such as dialogue data of express class calls; conversation data for a bank-like call;

if no related voice data exists, a first cluster can be determined according to the number attribute corresponding to the calling phone number in the current dialogue data, and when the voice data is analyzed subsequently, whether the number attribute corresponding to the calling phone number is the same as the former number attribute or not is determined, if so, the calling phone number is classified into the first cluster, and if not, the calling phone number is determined into a second cluster.

The dialogue data in each cluster is analyzed to extract key data in the dialogue, for example, for the previous example, key information: express, at home, deliver goods;

for session data of express class calls, possible answers by the user are: the express delivery cabinet is placed at no home, or the door front object area is placed at no home. For the above situation, the extracted key information includes: express, away from home, express cabinet/delivery;

and determining specific slot position attributes in the task logic, such as a calling clustering slot position attribute, a user position slot position attribute and a feedback information slot position attribute, aiming at the extracted key information.

And establishing a logic relation of the key information, setting a candidate information item of a specific slot position attribute based on the logic relation, and establishing a first initial answering task logic based on the cluster.

Referring to FIG. 2, which further illustrates the associative logic, task logic relationship diagram in an automated assistant, training of which also includes automatic learning and training to build interrelations between task logic through the associative logic.

Preferably, based on the opening authority, acquiring subsequent operations of the user after the call of the call incoming by the calling party is ended, for example, setting a first time threshold, taking the call ending time as the starting time, taking the user opening a second application as the terminating time, and recording the operations of the second application for the user if the difference between the terminating time and the starting time is less than or equal to the first predetermined time threshold;

or collecting an operation log of a user, determining a plurality of operations of the user after the call of the calling incoming call is finished, analyzing a time difference value between the operation starting time and the call finishing time, obtaining the user operation of which the time difference value is less than or equal to a first preset time threshold, and determining the operation process of the user through the operation log of the user operation of which the time difference value is less than or equal to the first preset time threshold;

the recording mode of the operation of the user is configured in advance, such as the operation process of the user is authorized to be recorded, and the operation content data is recorded, for example, when the user opens a communication application to make a call, recording the recording information of the call made by the user based on the pre-authorization of the user, or for example, when the user opens an instant communication application, the operation content data comprises a chat record;

and acquiring specific operation content data of the user through the user operation log. For example, the user makes an outgoing call.

Further, the following steps are adopted for processing the outgoing call of the user:

collecting called telephone number information during calling, acquiring called numbers of the calling, classifying and storing the called numbers, and setting a called number attribute matching relation table for the called numbers;

collecting call voice information, and classifying and storing the voice information according to the called telephone number;

recognizing the voice information, distinguishing a conversation main body, and marking conversation data in the voice information according to a telephone number corresponding to the conversation main body;

storing the marked text-form dialogue data in a calling data resource library, wherein the called telephone number is used as a primary index, and the calling start time is used as a secondary index;

clustering the conversation data according to the number attribute corresponding to the called telephone number in the called number attribute matching relation table;

the conversation data in each cluster is analyzed, key data in the conversation is extracted, and a first initial call task logic based on the clusters is created.

Wherein the processing for the outgoing call of the user may be independent of the processing for the incoming call.

Wherein, the called number attribute matching relation table can be as follows

TABLE 2

Called number	Short number	Prefix number	Number attribute	Scene class	Association class
						10086	Operator	Schedule board
010****			Bank	Schedule board
						179****1101	Family member	Location/schedule

Wherein the scene class may include one or more categories, such as a location scene class, a schedule scene class, etc.;

for example, when a user dials an express call, the position information of the user when the user dials a call is acquired, or when the user dials a merchant call, the destination position information of the user navigation is acquired by using a navigation setting route, or when the user dials a carrier call, the time and item information of a general schedule or a user preset schedule is acquired.

Specifically, the called numbers are classified and stored, and a called number attribute matching relation table is set for the called numbers; as shown in the above table, when the user settles the call at the end of the month of the schedule prompt, the call is dialed 10086, and the cost or service query is performed, so that the called number attribute matching relationship table content is filled as: called number: 10086, number attribute: operator, scene class: and (6) scheduling.

based on the recognized voice information, the dialog body is distinguished, and both data records are marked by using the user number and the called number 10086.

Recording the call start time, using the called number 10086 as a primary index, the call start time as a secondary index, storing the mark in a call data resource library, and associating the scene data with the dialogue data, for example, setting the scene data as a tertiary index;

clustering the dialogue data according to the number attribute operator corresponding to 10086 in the called number attribute matching relation table; for example, obtaining the dialogue data with the number attribute of the operator in the call data resource library, and uniformly classifying the dialogue data into the category of the operator;

analyzing the dialogue data in the operator cluster, extracting key data and scene data in the dialogue data, such as balance, value-added service, automatic query/manual query, key press, unsubscription and the like,

and determining specific slot attributes in the task logic according to the extracted key information, such as calling clustering slot attributes, prompting slot attributes, menu slot attributes, scene slot attributes and the like.

And establishing a logic relation of the key information, setting a candidate information item of a specific slot position attribute based on the logic relation, and establishing a first initial call task logic based on the cluster.

Further, the processing for the outgoing call of the user may also be associated with the incoming call processing.

Preferably, a difference between a call ending time when the user answers the calling incoming call and a call starting time when the user dials the called call is determined, it is determined whether or not the difference is less than or equal to a first time threshold, if the time is less than or equal to the first time threshold, acquiring first conversation data with the telephone number of the calling party and the conversation end time as indexes, and second dialogue data using the called telephone number and the call start time as indexes, performing statistical analysis on the first dialogue data and the second dialogue data, extracting similar and associated contents, if the first dialogue data and the second dialogue data are associated data, determining a calling number corresponding to the first dialogue data and a number attribute corresponding to the number according to the first dialogue data, determining a first initial answering task logic associated with the calling number according to the calling number and the number attribute corresponding to the calling number; meanwhile, a called number corresponding to the second dialogue data and a number attribute corresponding to the called number are determined, and a first initial call task logic relevant to the called number is determined according to the called number and the number attribute corresponding to the called number; and training the association logic based on the first initial answering task logic and the first initial calling task logic, thereby enriching the relation among the training logic based on the triggering condition, the triggering event and the triggering result. Thereby providing a more accurate intelligent operation flow.

Specifically, the user may make a next call based on the received call, such as receiving the incoming call 957, determining that the call is an express call, and recording the conversation by starting the recording function: if the calling party is X express, 2 express are informed by the telephone, and whether XX express points are needed to be released or delivery is needed. The user answers that the express delivery cabinet is not at home and requires to place the express delivery cabinet; the user then dials the family phone 179 x 1101, informing the family that a subsequent courier is needed at the courier cabinet. At the moment, address information of the call is acquired, conversation data of the call is analyzed, key data in the conversation are extracted, and a first initial call task logic based on the cluster is created.

And training the association logic of the first initial answering task logic and the first initial calling task logic, wherein the association logic can comprise a trigger condition attribute slot position, a trigger scene attribute slot position and a trigger result attribute slot position. Based on information content filled in a specific slot position in the first initial answering task logic, determining candidate information of a triggering condition attribute slot position and candidate information of a triggering scene attribute slot position, for example, determining that the triggering condition attribute slot position candidate information is an express cabinet according to a feedback information slot position attribute, determining that the triggering scene attribute slot position candidate information is a non-family location according to a user position slot position attribute, and executing the first initial calling task logic;

the above examples are merely illustrative and are not limiting embodiments, for example, by learning conversation data for incoming and outgoing calls, and also determining that the execution process of the association logic suitable for the current user is that the user is not at home (out of home), deposits a courier cabinet, and initiates a call; the user is not at home (local), deposits the express delivery cabinet, does not initiate the call in the follow-up. Namely, the slot position setting and execution mode in the association logic can be perfected through training, so that the slot position setting and execution mode is close to the use requirement of a user.

In addition, the method can also train the application task logic, for example, the execution process of the association logic suitable for the current user is determined to be that the user is not at home (other places), an express cabinet is stored, a call is initiated, and when the called party does not answer, the appropriate application task logic is determined to be called in the association logic, so that the short message is automatically edited and sent to the called party.

Further, a plurality of task logics or associated logics are optimally trained based on the updated data, or a training interface can be opened for a plurality of task logics or associated logics after training is completed, and the operations of adding, modifying and deleting the associated logics and the task logics by the user are received through the training interface, so that the training of the automatic assistant is further optimized.

By the method of the first embodiment of the invention, the automatic learning capability and the automatic processing user task capability of the voice assistant can be optimized, and the user experience is improved.

Example two

Referring to fig. 3, on the basis of the first embodiment, the second embodiment provides an intelligent response method based on identity confirmation, wherein the method includes but is not limited to:

Wherein, according to the calling phone number, acquiring the attribute information of the calling phone number specifically includes: calling a contact list, inquiring whether the calling number is in the contact list, and if so, determining the attribute information of the calling phone number through the contact list; if not, inquiring a calling number attribute matching relation table, determining attribute information of the calling phone number, and if the calling phone number is not contained in the calling number attribute matching relation table, sending a number inquiry request to a cloud server to inquire the attribute information of the calling number; and if the cloud server cannot determine the attribute information of the calling number, setting the attribute information of the calling number as an unauthenticated telephone number.

Further, the calling number may correspond to a plurality of number attributes, and when the calling number corresponds to a plurality of number attributes, the priority level of the number attributes is set, and the attribute classification is determined by the number attribute with the highest priority level.

For example, a calling number is marked as express delivery on a public platform, when a family is called by the calling number, the calling number is marked as a family for another current user, based on the priority level of the acquired number attribute, the priority level of the family is higher than that of the express delivery, the number attribute of the calling number is determined to be the family, and the attribute classification of the family is determined based on the family.

The first attribute is classified into a call takeover attribute preset by a user;

wherein the call takeover attribute may include one or more number attributes, such as express, bank, carrier, and/or unauthenticated.

Specifically, for example, the user receives the incoming call 158, the number is not marked in the contact list, the attribute information of the calling phone number is inquired, the previous record is found, the attribute information corresponding to the number is acquired as express delivery, it is determined that the first attribute class set by the user contains the express delivery, call takeover is executed, the voice assistant takes over the phone call, in this case, ringing can be played, and the voice assistant simulates answering.

And determining the first answering task logic of the express delivery according to the attribute information of the calling telephone number as the express delivery.

Preferably, referring to fig. 4, in the process of performing call takeover, first instruction information of a user is received, and a dialog processing interface is displayed based on the first instruction information;

presenting, at the conversation processing interface, the first conversation data and second conversation data generated based on the first conversation data.

The method comprises the steps of presetting a dialogue processing interface to start a hot key or item prompt, entering the dialogue processing interface, for example, displaying notification information based on clicking of the hot key or item prompt by a user, entering the dialogue processing interface based on clicking of the notification information by the user, or prompting that the user is in a call takeover state currently by adopting a breathing lamp flashing frequency, and directly jumping to enter the dialogue processing interface by unlocking.

Specifically, after the voice assistant answers, the voice recording function is started, the voice data storage area based on the calling number is determined, and the first voice data of the calling is received: for example, you have a courier, at home; and storing the voice data in real time, and generating a character record through voice recognition. Based on the slot position setting of the relevant conversation rule of the first answering task logic, extracting key data in the first voice data, filling a corresponding slot position, obtaining a next conversation operation, and based on the slot position setting of the next conversation operation, such as a response conversation operation, extracting suitable candidate information corresponding to the slot position. For example, if it is determined that the user position is not in the set home position area, the candidate information is selected: if the express cabinet is not at home, the express cabinet fills the corresponding slot position, so that a second conversation is generated, and if the express cabinet is not at home, the express cabinet is placed in a community express cabinet.

And continuously recording the conversation data in the conversation, and synchronously recording the text version information of the second conversation in a preset storage area and displaying the text version information on a conversation processing interface when the second conversation is generated.

Further, the user triggers an entry into a dialog handling interface, where the currently taken over call is presented in dialog form,

and receiving second instruction information of the user, wherein the second instruction information is instruction information of the user intention intervention conversation.

For example, the user inserts the third dialog data by clicking an intervention button of the dialog handling interface, or the user inserts the third dialog data in an input field of the dialog handling interface.

For example, the user clicks an interrupt/exit key below the session processing interface, if the user selects the interrupt key, the call processing is suspended, the user can record voice information as third session data, the voice assistant records the third session data, and starts a second timer after the user finishes recording the voice information, and after the second timer expires and no new voice data is input, the call processing is continuously taken over. And if the user selects the exit key, exiting to take over the call processing and entering into a user answering mode.

Or the user can input text or voice information through an input area of the dialogue processing interface, for example, a content input area is arranged below the dialogue display area, when the user edits the dialogue data and selects to send the dialogue data, the dialogue data edited by the user is determined to be third dialogue data, the third dialogue data is displayed on the dialogue processing interface, and the first answering task logic performs subsequent processing based on the extracted key data of the third dialogue data.

Further, whether trigger data corresponding to the association logic exists in the execution process of the first answering task logic is judged, and if the trigger data exists, the association logic determines whether to start and execute other task logic based on the trigger data after the execution of the first answering task logic is finished.

Specifically, after the user hears the call, it is determined that the association class for the calling number is a first association logic, after the call is ended, the first association logic is called, candidate information of one or more judgment condition slots in the first association logic is determined based on the conversation data and the user association data, logic judgment is performed based on the candidate information, whether a second task logic is started or not is determined, and when the second task logic needs to be started, the second task logic is triggered to be executed.

Further, after executing the first answering task logic or the second task logic, performing optimization training on the plurality of task logics or the associated logics according to newly acquired data information during the running period of the first answering task logic, the associated logics and the second task logic, or opening a training interface aiming at the trained plurality of task logics or associated logics, and receiving the addition, modification and deletion operations of the user on the associated logics and the content of the task logics through the training interface, thereby further optimizing the training of the automated assistant.

By the method of the second embodiment of the invention, the intelligent response function of the automatic assistant can be realized, the optimized intelligent response based on identity confirmation can be realized, meanwhile, the real-time visual conversation display is provided, the convenient conversation intervention and the training optimization of the automatic assistant are realized, and the user experience is improved.

EXAMPLE III

Referring to fig. 5, on the basis of the foregoing embodiment, a third embodiment of the present invention further provides an automatic learning method of a voice assistant;

step 501, collecting called telephone number information during calling, acquiring a called number of the calling, and determining the number attribute of the called number;

step 502, storing the call voice information after call completion, and storing the call voice information in a classified manner according to the called telephone number;

step 503, recognizing the voice information, distinguishing a conversation body, and marking the conversation data in the voice information according to a telephone number corresponding to the conversation body;

step 504, inquiring the application log, and determining whether an application operation record and/or an application notification record exists in a first preset time range before the call starting time; if so, go to step 505;

step 505, acquiring detailed operation data and/or notification records of the application in a second preset time range; analyzing the detailed operation data and/or the notification record of the application in a second preset time range and the dialogue data, and judging whether the detailed operation data and/or the notification record of the application in the second preset time range and the dialogue data have an association relation or not; if so, go to step 506;

step 506, when the detailed operation data of the application in the second predetermined time range has an association relation with the dialogue data, extracting the detailed operation data of the application in the second predetermined time range and key data in the dialogue data, and creating a first initial association logic.

The method comprises the steps of collecting called telephone number information during calling, acquiring called number of the calling, and determining number attribute of the called number specifically comprises

Collecting called telephone number information during calling to obtain a called number of the calling;

searching a called number attribute relationship matching relation table according to the called number, and extracting the number attribute of the called number if the called number attribute relationship matching relation table has the number attribute corresponding to the called number;

if the called number attribute relationship matching relationship table does not have the number attribute corresponding to the called number, searching a contact list, determining whether the contact list has the number attribute corresponding to the called number or not,

if the called number exists, extracting the number attribute corresponding to the called number, updating a called number attribute matching relation table based on the called number and the number attribute of the called number, if the called number does not exist, inquiring the registration information of the called number through a cloud end, determining the number attribute corresponding to the called number, and updating the called number attribute matching relation table based on the called number and the number attribute of the called number.

Specifically, for example, a user calls a called phone, the number attribute of the called phone number is determined to be a family for the called phone number, and when the call is determined to be initiated, voice data of the call is recorded, for example, express delivery exists, you need to go to a cell express delivery cabinet to take the call when the call is taken out, and the goods taking code is XX.

The recognizing the voice information, distinguishing a conversation body, and marking the conversation data in the voice information according to the telephone number corresponding to the conversation body further comprises:

and storing the marked text-form conversation data in a calling data resource library, wherein the called telephone number is used as a primary index, and the calling start time is used as a secondary index.

Specifically, by storing voice data and generating a text record through voice recognition, a subsequent user may invoke viewing of the text record, for example, the text record may be visually presented in the form of a dialog interface.

Further, the call end time is recorded and set as a secondary index.

Preferably, according to the number attribute corresponding to the called telephone number, determining a cluster corresponding to the number attribute, and determining a corresponding call task logic based on the cluster; and extracting key data in the conversation according to the analysis of the conversation data, and performing optimization training on the slot position setting and the slot position candidate information of the call task logic.

Further preferably, when there is no corresponding call task logic, extracting key data in the conversation according to the analysis of the conversation data, and creating a first initial call task logic based on the cluster.

Specifically, if a call task logic for the family class already exists, the call task logic for the family class is acquired, and based on the conversation data of the current conversation: if express delivery exists, you need to go to a community express cabinet to take the express delivery after work, and the goods taking code is XX. Extracting key data in the dialogue data: express delivery, a community express delivery cabinet and a goods taking code XX; matching the key data with relevant slot position information in the call task logic of the family class to enrich candidate information of relevant slot positions; and if unmatched key data exist, determining whether slot position information needs to be supplemented, if so, generating a corresponding slot position, and taking the key word as candidate information of the slot position.

Further, when the detailed operation data of the application in a second predetermined time range has an association relation with the dialogue data, extracting key data in the detailed operation data of the application in the second predetermined time range and the dialogue data, acquiring user state information corresponding to the dialogue data, and creating a first initial association logic based on the key data and the user state information.

Specifically, the user may customize or default a predetermined time period, for example, 1 minute, 30 seconds, etc., extract the call start time, determine a time range based on the call start time and the predetermined time period, filter log information according to the time range, and determine detailed operation data and/or notification records of one or more applications within the time range;

for example, 30 seconds ago, schedule prompt, ticket booking or booking registration is needed; or opening an express APP or a short message, clicking to acquire a pickup code, or browsing and reading the APP by the user.

And determining whether the operation or prompt aiming at the application A is the starting condition of the current conversation by judging the relevance of the current conversation data and the detailed operation data and/or the notification records of the one or more applications.

If the causal connection exists, extracting key data in the data related to the application A, such as express delivery, pickup code, and simultaneously based on the key data in the dialogue data: the express delivery, the goods taking code and the user state information corresponding to the dialogue data create a first initial association logic, and the triggering condition attribute slot of the association logic may include: associating the application slot position and triggering an event slot position; the trigger scene attribute slot position may include a trigger state slot position; for example: configuring candidate information of the associated application slot position to comprise an application A, and triggering the candidate information of the event slot position to comprise express; the candidate information for the trigger status slot includes a user location attribute.

And a relation network is obtained by connecting all different kinds of information together through continuously enriching the knowledge graph. The logical relationship between different slots in the voice assistant and the causal combination relationship between related information can also utilize the knowledge graph, thereby realizing the capability of autonomously analyzing problems.

For example, based on the express delivery appearing in the application a, the execution of the association logic is triggered, and according to the execution rule of the association logic, when the application a appears the express delivery and the user is not at home, the corresponding call task execution logic can be automatically called.

Further, a plurality of task logics or associated logics are optimally trained based on updated data, or a training interface can be opened for a plurality of task logics or associated logics after training is completed, and the operations of adding, modifying and deleting the associated logics and the task logics by a user are received through the training interface, so that the training of the voice assistant is further optimized.

By the method, the automatic learning capability and the automatic processing user task capability of the voice assistant can be optimized, and the user experience is improved.

Example four

Referring to fig. 6, on the basis of the foregoing embodiments, a fourth embodiment of the present invention further provides an intelligent calling method of a voice assistant;

601, determining a current application scene, and judging whether a triggering condition of a call task is met according to the application scene; if the triggering condition of the call task is satisfied, go to step 602;

step 602, obtaining key data information of the current application scenario, and providing the key data information to a first association logic;

step 603, determining a first call task logic to be enabled through the first association logic according to the key data information; pushing the key data information to the first call task logic through the first association logic;

step 604, controlling the first call task logic to initiate a call for a target user; generating first dialogue data according to the received key data information; and after the call is established, playing the voice information corresponding to the first call data to the target user.

The determining the current application scenario further comprises: the method comprises the steps of collecting information of a current application program of a user terminal, and determining a current application scene according to the information of the current application program of the user terminal.

Preferably, the information of the current application program of the user terminal includes a notification message that the user receives the application program or an application view message that the user opens.

For example, the calendar application prompts the user to make an appointment after 5 minutes, or the user opens a short message and views the message content to express the delivered XXX.

The acquiring information of a current application program of the user terminal and acquiring the current application scene specifically comprises determining a candidate task category according to the information of the current application program of the user terminal and a processing rule; judging whether the candidate task category belongs to a first calling task cluster; and if the cluster belongs to the first call task cluster, meeting the triggering condition of the call task.

For example, in the previous example, if the calendar application prompts the user to make an appointment after 5 minutes, it is determined that the user may need to perform subsequent 114 dialing, and thus it is determined that the current application scenario is the task category to be handled by calling;

the current application scenario is determined either only according to user settings or based on processing rules generated by historical data statistics, for example, the user needs to inquire the call charge through 10086 the day before the end of the month, or the user needs to call the husband to take five hours to school to pick up the child for a given time every wednesday.

And if the current time is the preset prompting time period of the wednesday, the current application scene is the task category to be processed by calling.

If the current application scene is the task category of the message to be sent, the current application scene is determined based on the historical data statistics of the user that the user generally adopts the short message sending query mode.

Specifically, when the current application scenario is a task category to be handled by calling, it is determined that the candidate task category belongs to a calling task cluster, that is, subsequent call processing needs to be performed.

At this time, key data information of the current application scene is acquired, for example: husband, five-click, school, and child pickup. And the first association logic determines a first call task logic needing to be called according to the key data information and provides the key data information to the first call task logic.

Further specifically, the first call task logic may determine the called number through the key data information "husband" and initiate a call to the child's father.

After the call is connected, the first call task logic determines first dialogue data according to the key data information and candidate information of the relevant slot position: for example, the senior citizen, XX is followed from 5 pm to X school today.

In addition, further, after the conversation starts, the recording function is started, and the conversation data of the called number is stored in the conversation data resource library. The called telephone number is used as a primary index, and the call start time is used as a secondary index.

Preferably, second voice data replied by the called party is received, the second dialogue data is generated through voice recognition, and key data information in the second dialogue data is extracted according to the slot position setting of the relevant dialogue rule of the first call task logic to generate the next dialogue data. For example, the second session data is extracted, the corresponding slot is filled, and the next session data is obtained. For example, in the second dialog data, child dad replies with: today, you need overtime, and you have 5 free spots to pick up. At this time, according to the dialogue data, extracting corresponding key data: overtime, you, if, connect children.

And the first call task logic determines candidate key information of the next conversation according to the second conversation data and the slot positions of the multiple rounds of conversations of the first call task logic, generates third conversation data and plays voice information corresponding to the third conversation data to the called party.

The literal version information for the third dialogue data is synchronously recorded in the dialogue data resource library.

Further, a user can enter a conversation processing interface through triggering, and the conversation processing interface displays conversation data of the two parties in the calling process.

The user may insert the third dialog data by clicking an intervention button of the dialog handling interface or the user may insert the third dialog data in an input field of the dialog handling interface.

For example, the user clicks an interrupt/exit button below the dialog processing interface, if the user selects the interrupt button, the call processing is suspended, the user may enter voice information as inserted dialog data, or the user may enter text or voice information through an input region of the dialog processing interface, for example, a content input region is set below the dialog display region, when the user edits the dialog data and selects to send, the user-edited dialog data is determined as inserted dialog data, the inserted dialog data is also recorded, and corresponding voice information is provided to the called user. If the user selects the exit key, the call processing is exited and the user call mode is entered. And in the call process, the recording and the storage of the conversation data are kept.

Further, a plurality of task logics or associated logics are optimally trained based on newly acquired data information, or a training interface can be opened for a plurality of task logics or associated logics after training is completed, and the operations of adding, modifying and deleting the associated logics and the task logics by a user are received through the training interface, so that the training of the voice assistant is further optimized.

By the method, the intelligent calling function of the voice assistant can be realized, the optimized intelligent calling based on the application scene is realized, meanwhile, the real-time visual conversation display is provided, the convenient conversation intervention and the training optimization of the automatic assistant are realized, and the user experience is improved.

EXAMPLE five

Referring to fig. 7, on the basis of the foregoing embodiment, a fifth embodiment of the present invention further provides an AR-based intelligent call method

Step 701, acquiring scene information of a position area where a user is located in real time through an Augmented Reality (AR) device;

step 702, determining the augmented reality AR device type, and determining one or more candidate objects based on the augmented reality AR device type;

step 703, acquiring attribute information of the candidate target object, and starting a first association logic based on the attribute information of the candidate target object;

step 704, the first association logic determines a first call task logic to be enabled according to the attribute information of the candidate target object;

step 705, determining a phone number corresponding to the candidate target according to the attribute information of the candidate target, and initiating a call based on the phone number by the first call task logic.

Wherein the step 702 further comprises:

determining the one or more objects within a central region of the scene information if the Augmented Reality (AR) device is AR glasses worn by a user;

judging whether the stay time of the target object in the central area of the scene is greater than or equal to a first preset time threshold or not;

if the target object is greater than or equal to a first preset time threshold, determining the target object as a candidate target object;

and determining attribute information of the candidate target object, wherein the attribute information of the candidate target object comprises the category of the candidate target object and the associated telephone number of the candidate target object.

Specifically, for example, the user wears AR glasses, and road information viewed or surrounding buildings are displayed through the AR glasses.

Determining a plurality of buildings which are currently in the sight range, and determining that the XX restaurant is a candidate target object when the XX restaurant is found in the central area of the sight of the user within a certain time; the voice assistant further obtains relevant information of XX restaurant, such as booking a phone call, consulting a phone call, business hours, etc.

Wherein, the step 702 may further include:

determining the one or more target objects in the scene information if the augmented reality AR device is a vehicle-mounted AR device;

judging the current vehicle speed of a user, and further acquiring the attribute information of the target object if the current vehicle speed of the user is less than or equal to a second threshold value, wherein the attribute information of the target object comprises the category of the target object;

judging whether the category of the target object belongs to a first business category, if so, determining the target object as a candidate target object; and determining the associated telephone number of the candidate target object.

Further, the first commercial category is a gourmet category.

Step 703, obtaining the attribute information of the candidate target object, and starting the first association logic based on the attribute information of the candidate target object further includes

And triggering and starting a first association logic based on that the attribute information of the candidate target object is a food class and the candidate target object and the user position are in the same predetermined area range.

In step 704, the first correlation logic determines, according to the attribute information of the candidate target object, that the first call task logic to be enabled further includes

The first association logic determines a candidate calling task logic type according to the attribute information of the candidate target object and the user attribute information;

and determining a first call task logic to be started according to the candidate call task logic type.

For example, clustering can be performed for different call tasks, such as business handling, take-away booking, conversation exchange, etc.;

and determining the type of the call task logic which needs to start the takeaway subscription class currently based on the association logic, and determining the first call task logic to be started based on the association between the data.

Further, the method further includes step 706, the first call task logic generates first dialogue data according to the attribute information of the candidate target object and the user attribute information; and after the call is established, playing the voice information corresponding to the first call data to the called user.

The user may insert the dialog data by clicking an intervention button of the dialog handling interface or the user may insert the dialog data in an input field of the dialog handling interface.

Furthermore, the attribute information of the user can comprise user portrait information, the number of people having a meal and the time of having a meal; the user representation information may include user dining preferences. The number of people having a meal and the meal time can be set according to a user, or the number of people in the vehicle can be determined through vehicle-mounted camera shooting, or information related to the number of people having a meal and the meal time can be extracted based on user conversation information in a preset time range, or the information related to the number of people having a meal and the meal time can be determined based on historical data.

Further, the step 702 includes:

determining a business area where the user is located according to the one or more target objects;

based on the commercial district, inquiring one or more candidate objects which accord with the user attribute information;

Wherein, the step 702 may further include:

if the augmented reality AR device is a vehicle-mounted AR device, judging whether a user sets a navigation destination, and if not, determining the one or more target objects in the scene information;

determining a first preset area based on the driving speed of a vehicle of a user and a current business area, and inquiring one or more candidate objects which accord with the attribute information of the user in the first preset area;

Preferably, if the user sets a navigation destination, determining a business area where the navigation destination is located according to the navigation destination;

Further, when the determined candidate object is plural, the priority level of the candidate object is determined based on the user attribute information,

and the first call task logic initiates calls in sequence according to the priority level.

In addition, the user can view the dialogue data based on the dialogue processing interface at any time.

By the method, the intelligent calling function of the voice assistant can be realized, the optimized scene determination based on enhanced display and the intelligent seat booking based on food scenes are realized, meanwhile, the real-time visual conversation display is provided, the convenient conversation intervention and the training optimization of the voice assistant are realized, and the user experience is improved.

EXAMPLE six

With reference to fig. 8, on the basis of the foregoing embodiment, a sixth embodiment of the present invention further provides an automatic learning method for a voice assistant

Step 801, collecting calling telephone number information during incoming call, classifying and storing the calling numbers, and setting/updating a calling number attribute matching relation table aiming at the calling numbers;

step 802, determining attribute information of the calling phone number according to the calling phone number;

step 803, judge whether there is a non-authenticated telephone number, if there is a non-authenticated telephone number, obtain the conversation data corresponding to voice message according to the said non-authenticated telephone number label;

step 804, determining whether key data information related to number attributes exists according to the dialogue data corresponding to the voice information marked by the unauthenticated phone number; if so, go to step 805;

step 805, obtaining number attributes of unauthenticated phone numbers, and updating the number attributes to a calling number attribute matching relation table;

and 806, sharing the number attribute of the unauthenticated phone number to the intelligent interaction platform.

Specifically, for example, when the subscriber receives a telephone call of 010 × 1234, the number is not recorded in the calling number attribute matching relationship table; the identification information for the number is also not present via a networked query. The number is determined to be an unauthenticated telephone number.

Step 801 specifically includes judging whether a calling number attribute matching relationship table exists, and if not, creating the calling number attribute matching relationship table.

Step 802 further includes determining whether a number attribute corresponding to the calling number exists in a calling number attribute matching relationship table; if the attribute information of the calling phone number exists, determining the attribute information of the calling phone number according to a calling number attribute matching relation table, if the calling phone number is not contained in the calling number attribute matching relation table, calling a contact list, inquiring whether the calling phone number is in the contact list, if so, determining the attribute information of the calling phone number through the contact list, and if not, sending a number inquiry request to a cloud server to inquire the attribute information of the calling phone number; and if the cloud server does not register or mark the attribute information of the calling number, determining that the attribute information of the calling number is an unauthenticated telephone number.

Further, the step 804 further includes, if not present, performing step 807;

step 807, judging whether the unauthenticated number is a virtual number, and if not, acquiring the calling grade of the unauthenticated telephone number; judging whether the calling grade of the unauthenticated telephone number belongs to a first threshold range; if not, determining a call task logic to be started according to the unauthenticated telephone number;

step 808, invoking the call task logic to initiate a call to the unauthenticated phone number, and generating first dialogue data about inquiring attribute information of the calling phone number.

For example, the unauthenticated phone number is judged based on a virtual number rule, if the unauthenticated phone number belongs to the virtual number, subsequent processing is not performed, and if the unauthenticated phone number does not belong to the virtual number, the call grade of the number is judged, for example, the unauthenticated phone number can be divided into three grades according to the call feasibility of the number, for example, a general mobile phone number section belongs to grade 1, a general fixed number section belongs to grade 2, and an out-of-domain number belongs to grade 3; and when the unidentified number belongs to the level 3, the conversation data is obtained in an automatic callback mode.

Further, attribute information of the handset about the unauthenticated phone number is further based on the dialogue data of the call.

Further preferably, when the call level of the unauthenticated phone number belongs to a first threshold range, accessing the intelligent interaction platform through an open interface of the intelligent interaction platform, and issuing a function acquisition request of the unauthenticated phone number;

and acquiring response information of the function acquisition request based on the unauthenticated telephone number, which is shared by the third party through the intelligent interaction platform.

EXAMPLE seven

With reference to fig. 9, on the basis of the foregoing embodiment, the present embodiment further provides an automatic assistant training device, which is characterized in that the device includes:

the calling acquisition module is used for acquiring calling telephone number information during incoming call, classifying and storing the calling numbers and setting a calling number attribute matching relation table aiming at the calling numbers;

the storage management module is used for collecting call voice information and classifying and storing the voice information according to the calling telephone number;

the data processing module is used for identifying the voice information, distinguishing a conversation main body and marking the conversation data in the voice information according to a telephone number corresponding to the conversation main body;

the data clustering module is used for clustering the conversation data according to the number attribute corresponding to the calling telephone number in the number attribute matching relation table;

the logic control module is used for analyzing the conversation data in each cluster, extracting key data in the conversation and creating a first initial answering task logic based on the cluster; associating the first initial listening task logic with the corresponding calling number and number attribute information.

Preferably, the calling number attribute matching relationship table includes a calling number, a number attribute and an association class.

Preferably, the data processing module further comprises a storage module for storing the dialog data in the form of marked text in an incoming call data repository, wherein the telephone number of the calling party is used as a primary index, and the call end time of the call incoming by the calling party is used as a secondary index.

Preferably, the device further comprises

And the function switch module is used for authorizing the starting of the automatic assistant training function and authorizing the calling authority of the application program of the user terminal, and comprises the starting of the automatic assistant training function, the authorization of answering or dialing the call aiming at the call module and the recording aiming at the call.

Preferably, the device further comprises

The timing monitoring module is also used for acquiring the subsequent operation of the user after the call of the call incoming by the calling party is finished; the subsequent operation comprises collecting operation logs of a user, determining a plurality of operations of the user after the call of the calling incoming call is finished, analyzing the time difference between the operation starting time and the call finishing time, obtaining the user operation of which the time difference is less than or equal to a first preset time threshold, and determining the operation process of the user through the operation logs of the user operation of which the time difference is less than or equal to the first preset time threshold.

Example eight

Referring to fig. 10, on the basis of the foregoing embodiments, the present embodiment further provides an intelligent response apparatus based on identity confirmation, where the apparatus includes:

Preferably, the attribute information determination module is further configured to

Calling a contact list, inquiring whether the calling number is in the contact list, and if so, determining the attribute information of the calling phone number through the contact list; if not, inquiring a calling number attribute matching relation table, determining attribute information of the calling phone number, and if the calling phone number is not contained in the calling number attribute matching relation table, sending a number inquiry request to a cloud server to inquire the attribute information of the calling number; and if the cloud server cannot determine the attribute information of the calling number, setting the attribute information of the calling number as an unauthenticated telephone number.

Preferably, the control module is further configured to preset a dialog processing interface to start a hotkey or an item prompt, and enter the dialog processing interface based on a user clicking the hotkey or the item prompt in a call takeover process.

Preferably, the device further comprises

And the display module presents the first dialogue data and second dialogue data generated based on the first dialogue data in the dialogue processing interface.

Preferably, the control module is further configured to receive second instruction information of the user, where the second instruction information is instruction information of a user intention to intervene in a dialog;

acquiring third dialogue data provided by a user;

and the first answering task logic performs subsequent dialogue processing based on the extracted key data of the third dialogue data.

Example nine

Referring to fig. 11, on the basis of the foregoing embodiment, this embodiment further provides an automatic learning apparatus for a voice assistant, where the apparatus includes:

the calling acquisition module acquires called telephone number information during calling, acquires a called number of the calling and determines the number attribute of the called number;

the storage management module is used for storing the call voice information after the call is connected and storing the call voice information in a classified manner according to the called telephone number;

the timing monitoring module is used for inquiring the application log and determining whether the operation record and/or the notification record of the application exist in a first preset time range before the call starting time; if yes, sending a notification message to the control module;

the control module acquires detailed operation data and/or notification records of the application within a second preset time range; analyzing the detailed operation data and/or the notification record of the application in a second preset time range and the dialogue data, and judging whether the detailed operation data and/or the notification record of the application in the second preset time range and the dialogue data have an association relation or not; when the detailed operation data and/or the notification record of the application in the second preset time range are in an association relation with the dialogue data, extracting the detailed operation data of the application in the second preset time range and key data in the dialogue data, and creating a first initial association logic.

Preferably, the call acquisition module is further used for

and searching a called number attribute relation matching relation table according to the called number, and if the called number attribute relation matching relation table has the number attribute corresponding to the called number, extracting the number attribute of the called number.

Preferably, the data processing module is further configured to store the dialog data in the form of a marked text in a call data repository, where a called phone number is used as a primary index, and a call start time is used as a secondary index.

Preferably, the control module is further configured to determine a cluster corresponding to the number attribute according to the number attribute corresponding to the called phone number, determine a corresponding call task logic based on the cluster, extract key data in a session according to analysis of the session data, and update slot setting and slot candidate information of the call task logic.

Preferably, the control module is further configured to, when the detailed operation data of the application in the second predetermined time range has an association relationship with the session data, extract key data in the detailed operation data of the application in the second predetermined time range and the session data, acquire user state information corresponding to the session data, and create a first initial association logic based on the key data and the user state information.

Example ten

Referring to fig. 12, on the basis of the foregoing embodiment, this embodiment further provides an intelligent calling device of a voice assistant, where the device includes:

the condition judgment module is used for determining the current application scene and judging whether the triggering condition of the call task is met or not according to the application scene; if the triggering condition of the call task is met, sending a notification message to a data extraction module;

the data extraction module is used for acquiring key data information of the current application scene and providing the key data information for a first association logic;

the control module is used for determining a first call task logic to be enabled through the first association logic according to the key data information; pushing the key data information to the first call task logic through the first association logic;

the control module controls the first call task logic to initiate a call for a target user; generating first dialogue data according to the received key data information; and after the call is established, playing the voice information corresponding to the first call data to the target user.

Preferably, the condition judgment module is further used for

Acquiring information of a current application program of a user terminal, and determining a current application scene according to the information of the current application program of the user terminal;

the information of the current application program of the user terminal comprises a notification message that the user receives the application program or a message that the user opens the application to view.

Preferably, the condition judgment module is further used for

Determining candidate task types according to the information of the current application program of the user terminal and a processing rule; judging whether the candidate task category belongs to a first calling task cluster; and if the cluster belongs to the first call task cluster, meeting the triggering condition of the call task.

Preferably, the control module further comprises

After the conversation starts, starting a recording function, and storing the conversation data of the called number in a conversation data resource library; the called telephone number is used as a primary index, and the call start time is used as a secondary index.

Preferably, the control module further comprises

And receiving second voice data replied by the called party, generating second dialogue data through voice recognition, extracting key data information in the second dialogue data according to the slot position setting of the relevant dialogue rule of the first call task logic, and generating next dialogue data.

EXAMPLE eleven

Referring to fig. 13, on the basis of the foregoing embodiment, the present embodiment further provides an AR-based intelligent call device, where the device includes:

the scene acquisition module is used for acquiring scene information of a position area where a user is located in real time through the AR equipment;

a target determination module to determine the augmented reality AR device type, determine one or more candidate targets based on the augmented reality AR device type;

the control module is used for acquiring the attribute information of the candidate target object and starting a first association logic based on the attribute information of the candidate target object;

the control module is further used for determining a first call task logic to be enabled through the first association logic according to the attribute information of the candidate target object;

the control module further determines a phone number corresponding to the candidate target according to the attribute information of the candidate target object, and initiates a call to the phone number through the first call task logic.

Preferably, the object determination module is further for

Preferably, the control module is also used for

Determining a candidate calling task logic type according to the attribute information of the candidate target object and the user attribute information;

Example twelve

Referring to fig. 14, on the basis of the foregoing embodiment, the present embodiment further provides an automatic learning apparatus for a voice assistant, where the apparatus includes:

the calling acquisition module is used for acquiring calling telephone number information during incoming call, classifying and storing the calling numbers, and setting/updating a calling number attribute matching relation table aiming at the calling numbers;

the attribute information determining module is used for determining the attribute information of the calling telephone number according to the calling telephone number;

the control module is used for judging whether an unauthenticated telephone number exists or not, and if the unauthenticated telephone number exists, acquiring dialogue data corresponding to the voice information marked according to the unauthenticated telephone number;

the judging module is used for determining whether key data information related to number attributes exists or not according to the dialogue data corresponding to the voice information marked by the unauthenticated telephone number; if yes, sending a notice to a data management module;

the data management module acquires the number attribute of the unauthenticated telephone number and updates the number attribute to a calling number attribute matching relation table;

and the sharing module shares the number attribute of the unauthenticated phone number to the intelligent interaction platform.

Preferably, the call acquisition module is further used for

And judging whether a calling number attribute matching relation table exists or not, and if not, creating the calling number attribute matching relation table.

Determining whether a number attribute corresponding to the calling number exists in a calling number attribute matching relation table; if the attribute information of the calling phone number exists, determining the attribute information of the calling phone number according to a calling number attribute matching relation table, if the calling phone number is not contained in the calling number attribute matching relation table, calling a contact list, inquiring whether the calling phone number is in the contact list, if so, determining the attribute information of the calling phone number through the contact list, and if not, sending a number inquiry request to a cloud server to inquire the attribute information of the calling phone number; and if the cloud server does not register or mark the attribute information of the calling number, determining that the attribute information of the calling number is an unauthenticated telephone number.

Preferably, the judging module is further configured to

If the key data information related to the number attribute does not exist, sending a notification message to the judging module;

the judging module is further used for judging whether the unauthenticated number is a virtual number or not, and if not, acquiring the calling grade of the unauthenticated telephone number; judging whether the calling grade of the unauthenticated telephone number belongs to a first threshold range; if not, sending a notification message to the control module;

the control module is further configured to determine a call task logic to be enabled according to the unauthenticated phone number.

Preferably, the control module is also used for

Calling the call task logic to initiate a call for the unauthenticated telephone number, and generating first dialogue data about attribute information of the inquired calling telephone number.

The invention also provides a terminal device, characterized in that it comprises a processor and a memory, in which a computer program is stored that is executable on the processor, said computer program implementing the method as described above when executed by the processor.

The terminal equipment comprises but is not limited to a computer, a mobile phone, a tablet personal computer, a vehicle machine, a vehicle-mounted terminal, an intelligent sound box, a set top box and an intelligent household appliance.

The invention provides a computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, which computer program is executable on a processor, and when executed implements a method as described above.

Any combination of one or more computer-readable media may be employed. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. The computer-readable storage medium may include: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), a flash memory, an erasable programmable read-only memory (EPROM), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

Computer program code for carrying out operations of the present invention may be written in one or more programming languages, or a combination thereof.

The above description is only an example for the convenience of understanding the present invention, and is not intended to limit the scope of the present invention. In the specific implementation, a person skilled in the art may change, add, or reduce the components of the apparatus according to the actual situation, and may change, add, reduce, or change the order of the steps of the method according to the actual situation without affecting the functions implemented by the method.

While embodiments of the invention have been shown and described, it will be understood by those skilled in the art that: various changes, modifications, substitutions and alterations can be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents, and all changes that come within the meaning and range of equivalency of the claims are to be embraced within their scope.

Claims

1. An intelligent response method based on identity confirmation, which is characterized by comprising the following steps:

step 307, determining candidate information of a relevant slot position in a first answer task logic according to the first answer data, wherein the relevant slot position is a corresponding slot position set based on a relevant dialog rule of the first answer task logic.

2. The method of claim 1, wherein the step 302 further comprises

3. The method as claimed in claim 1, further comprising

And presetting a dialogue processing interface to start a hot key or item prompt, and entering the dialogue processing interface based on that a user clicks the hot key or the item prompt in the process of executing call takeover.

4. The method of claim 3, further comprising

5. The method of claim 3, further comprising

Receiving second instruction information of a user, wherein the second instruction information is instruction information of user intention intervention conversation;

acquiring third dialogue data provided by a user;

6. An intelligent answering device based on identity confirmation, characterized in that the device comprises:

and the control module determines candidate information of a relevant slot position in a first answering task logic according to the first answering task data, wherein the relevant slot position is a corresponding slot position set based on a relevant conversation rule of the first answering task logic.

7. The apparatus of claim 6, wherein the attribute information determination module is further configured to determine the attribute information

8. The device of claim 6, wherein the control module is further configured to preset a dialog handling interface to initiate a hotkey or an item prompt, and enter the dialog handling interface based on a user clicking the hotkey or the item prompt during the call takeover process.

9. The apparatus of claim 6, further comprising

And the display module presents the first dialogue data and second dialogue data generated based on the first dialogue data in a dialogue processing interface.

10. The apparatus of claim 7, wherein the control module is further configured to receive second instruction information of the user, where the second instruction information is instruction information of a user intending to intervene in a dialog;

acquiring third dialogue data provided by a user;

11. A terminal device, characterized in that the terminal device comprises a processor and a memory, in which a computer program is stored which is executable on the processor, which computer program, when being executed by the processor, realizes the method according to any one of claims 1 to 5.

12. A computer-readable storage medium, in which a computer program that is executable on a processor is stored, which computer program, when being executed, carries out the method according to any one of claims 1 to 5.