Conversation method, conversation device, electronic device and storage medium

Abstract

A conversation method, a conversation device, an electronic device and a storage medium, belonging to the field of artificial intelligence technology. The method comprises: obtaining target user information of a target object. Constructing a user portrait according to the target user information to obtain target portrait information of the target object. Inputting the target portrait information into a pre-trained interest prediction model to perform interest prediction to obtain target interest information of the target object. Generate a target conversation model according to the target interest information and the role information of the first initiating object, the target conversation model including at least two target conversation types. Generate conversation information of each target conversation type according to the target conversation model and the target interest information, the conversation information is used for the first initiating object to interact with the conversation terminal, thereby achieving targeted conversation guidance and improving conversation effect.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a conversation method, a conversation device, an electronic device and a storage medium. Background Art

[0002] In order to attract outstanding talents, companies usually organize interviews with recruitment targets (such as job applicants or new employees) to promote the company's situation to the recruitment targets and understand their work needs. In the current interview method, the company's recruiters mainly determine the content of communication based on the job situation and the resume information of the recruitment targets, and are responsible for guiding the conversation topics. However, this method is easily influenced by the subjective influence of the recruiters and lacks targeted conversation skills, resulting in poor conversation results. Summary of the invention

[0003] The main purpose of the embodiments of the present application is to propose a conversation method, a conversation device, an electronic device and a storage medium, aiming to achieve targeted conversation guidance and improve conversation effects.

[0004] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application proposes a conversation method, the method comprising:

[0005] Get the target user information of the target object;

[0006] Constructing a user portrait according to the target user information to obtain target portrait information of the target object;

[0007] Inputting the target portrait information into a pre-trained interest prediction model to perform interest prediction, thereby obtaining target interest information of the target object;

[0008] Generate a target conversation model according to the target interest information and the role information of the first initiating object, wherein the target conversation model includes at least two target conversation types;

[0009] According to the target conversation model and the target interest information, the conversation information of each target conversation type is generated; wherein the conversation information is used for the first initiating object to perform conversation interaction with the conversation terminal.

[0010] In some implementations, the session information includes target speech information of the target session type, and generating the session information of each target session type according to the target session model and the target interest information includes:

[0011] According to the target conversation type, reference speech information is obtained from a preset conversation information library; the reference speech information and the target interest information are input into a pre-trained first evaluation model for a first evaluation process to obtain a first matching degree; wherein the first matching degree is used to represent the matching degree between the reference speech information and the target interest information; according to the first matching degree, the target speech information is filtered out from the reference speech information; wherein the target speech information is speech information used to reply to the target object.

[0012] In some implementations, the session information further includes tool screening information of the target session type; after screening the target speech information from the reference speech information according to the first matching degree, the method further includes:

[0013] According to the target conversation type, tool description information is obtained from the conversation information library; wherein the tool description information is used to represent the description information of the conversation assistance tool; the tool description information and the target speech information are input into a pre-trained second evaluation model for a second evaluation process to obtain a second matching degree; wherein the second matching degree is used to represent the matching degree between the tool description information and the target speech information; according to the second matching degree, the tool screening information is screened out from the tool description information; wherein the tool screening information is used to screen out the target assistance tool from the conversation assistance tool.

[0014] In some implementations, the method further includes training the interest prediction model, specifically including:

[0015] Acquire training user information of the training object; construct a user portrait based on the training user information to obtain the training portrait information of the training object; acquire training interest information of the training object; input the training portrait information and the training interest information into a preset initial prediction model for training processing to obtain the interest prediction model.

[0016] In some implementations, after generating the session information of each target session type according to the target session model and the target interest information, the method further includes:

[0017] Acquire a session instruction triggered by the first initiating object; filter out a specified session type from the target session types according to the session instruction; and output the session information according to the specified session type.

[0018] In some implementations, after acquiring the target user information of the target object, the method further includes: acquiring input information; and updating the target user information according to the input information.

[0019] In some implementations, generating a target conversation model according to the target interest information and the role information of the first initiating object includes:

[0020] Obtain a list of session initiators; wherein the list of session initiators includes at least two session initiators arranged in order, and a first initiating object is one of the session initiators in the list of session initiators; obtain position information of the first initiating object in the list of session initiators to obtain the initiator position; if the initiator position is the first position, generate the target session model according to the target interest information and the role information of the first initiating object; if the initiator position is the second position, determine the session initiator arranged before the first initiating object from the list of session initiators as the second initiating object; obtain an evaluation result of the second initiating object on the target object and a historical session model generated when the second initiating object conducts a session with the target object, wherein the historical session model includes at least two historical session types; generate a target session model according to the target interest information and the role information of the first initiating object, and in combination with the historical session model and the evaluation result.

[0021] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides a conversation device, which is applied to a conversation terminal, and the device includes:

[0022] An acquisition module is used to acquire target user information of a target object;

[0023] A construction module, used to construct a user portrait according to the target user information to obtain the target portrait information of the target object;

[0024] A prediction module, used for inputting the target portrait information into a pre-trained interest prediction model to perform interest prediction and obtain target interest information of the target object;

[0025] A first generating module, configured to generate a target conversation model according to the target interest information and the role information of the first initiating object, wherein the target conversation model includes at least two target conversation types;

[0026] The second generating module is used to generate session information of each target session type according to the target session model and the target interest information; wherein the session information is used for the first initiating object to perform a session interaction with the session terminal.

[0027] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory, a processor, a program stored on the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, the session method described in the first aspect is realized.

[0028] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a storage medium, which is a computer-readable storage medium used for computer-readable storage, and the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the conversation method described in the first aspect above.

[0029] The conversation method, conversation device, electronic device and storage medium proposed in the present application construct target portrait information based on the target user information of the target object, and input the target portrait information into a pre-trained interest prediction model for interest prediction, so as to obtain the target interest information of the target object, which is convenient for targeted analysis of the interest tendency of the target object. According to the role information and target interest information of the first initiating object, a target conversation model is constructed, which effectively sorts out the target conversation type suitable for the current conversation, and can simultaneously meet the interest tendency of the target object and the conversation needs of the first initiating object. Based on this, in combination with the target conversation model and the target interest information, the conversation information corresponding to each target conversation type is generated for the first initiating object to have a conversation with the conversation terminal, thereby providing the first initiating object with conversation skills, achieving a targeted conversation guidance effect, and helping to improve the conversation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flowchart of a conversation method provided in an embodiment of the present application;

[0031] Figure 2 It is a flowchart of a training interest prediction model in an embodiment of the present application;

[0032] Figure 3 yes Figure 1 Specific flow diagram of step S105;

[0033] Figure 4 is a schematic diagram of a process for generating tool screening information in an embodiment of the present application;

[0034] Figure 5 yes Figure 1 Specific flow diagram of step S104;

[0035] Figure 6 is a schematic diagram of a first display page in an embodiment of the present application;

[0036] Figure 7 is a schematic diagram of a second display page in an embodiment of the present application;

[0037] Figure 8 is a structural diagram of a conversation device provided in an embodiment of the present application;

[0038] Fig. 9 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0040] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0042] First, some nouns involved in this application are analyzed:

[0043] Artificial Intelligence (AI): It is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. AI is a branch of computer science. AI attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Research in this field includes robots, language recognition, image recognition, natural language processing and expert systems. AI can simulate the information process of human consciousness and thinking. AI is also a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0044] Natural Language Processing (NLP): NLP uses computers to process, understand, and apply human languages ​​(such as Chinese, English, etc.). NLP is a branch of artificial intelligence and an interdisciplinary subject between computer science and linguistics. It is often referred to as computational linguistics. Natural language processing includes grammatical analysis, semantic analysis, and text understanding. Natural language processing is commonly used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis, and opinion mining. It involves data mining related to language processing, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computing.

[0045] Information Extraction: A text processing technology that extracts specified types of entity, relationship, event and other factual information from natural language text and forms structured data output. Information extraction is a technology that extracts specific information from text data. Text data is composed of some specific units, such as sentences, paragraphs, and chapters. Text information is composed of some small specific units, such as characters, words, phrases, sentences, paragraphs, or a combination of these specific units. Extracting noun phrases, names, place names, etc. from text data are all text information extraction. Of course, the information extracted by text information extraction technology can be various types of information.

[0046] User portrait is a labeled user model abstracted from the user's personal attributes (such as age, height, weight, native place, permanent residence, etc.), behavioral attributes (such as living habits and personal preferences, etc.) and social attributes (such as educational background, marital status, income level and asset status, etc.). It concretizes the user image by utilizing labels obtained from highly refined user information, thereby providing users with targeted services and facilitating computer processing.

[0047] Based on this, the embodiments of the present application provide a conversation method, a conversation device, an electronic device and a storage medium, which aim to achieve targeted conversation guidance and improve conversation effects.

[0048] The conversation method, conversation device, electronic device and storage medium provided in the embodiments of the present application are specifically described through the following embodiments. First, the conversation method in the embodiments of the present application is described.

[0049] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, basic artificial intelligence technologies generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software technology mainly includes computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0050] The conversation method provided in the embodiment of the present application relates to the field of artificial intelligence technology. The conversation method provided in the embodiment of the present application can be applied to a conversation terminal, or it can be software running in the conversation terminal. In some embodiments, the conversation terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the software can be an application that implements the conversation method, etc., but is not limited to the above forms.

[0051] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0052] It should be noted that in each specific implementation of the present application, when it comes to the need to perform relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with the relevant laws, regulations, and standards of the relevant countries and regions. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present application will be obtained.

[0053] Figure 1 is a flowchart of a conversation method provided in an embodiment of the present application. Figure 1The method may include but is not limited to steps S101 to S105.

[0054] Step S101: Acquire target user information of a target object.

[0055] The embodiments of the present application are applicable to various application scenarios such as recruitment interviews and employee interviews. For example, the target object may be a job seeker or a new employee. The target user information may include the resume information of the target object, such as age, gender, income, family situation and work experience. In some cases, the target user information may also include the behavior data and financial situation of the target object. The behavior data includes but is not limited to the target object's use of enterprise products or internal management software (or system). For example, the behavior data includes the target object's login status in the internal management software, the health section activity, the life section activity and the financial section activity, etc. The financial situation includes but is not limited to life insurance, property insurance, securities, bank asset status and transaction status, etc., so as to facilitate multi-faceted mining of the potential needs of the target object. It can be understood that the embodiments of the present application do not specifically limit the type of target user information.

[0056] Specifically, in step S101, the method of obtaining the target user information includes but is not limited to: obtaining the resume information of the target object from the talent database, and extracting the target user information of the target object from the resume information; or, collecting the target user information of the target object through the embedded point method of authorized clients such as iOS, Android, mini-programs and websites; or, obtaining file data manually imported or pushed by a third-party application system, and then extracting the target user information from the file data, and the file data may include but is not limited to operation logs, third-party open application data, form information and other business data.

[0057] Step S102: construct a user portrait based on the target user information to obtain target portrait information of the target object.

[0058] In an embodiment of the present application, the target portrait information includes at least two feature tags generated according to the target user information, which are used to label the target object. Specifically, the method of constructing the user portrait includes but is not limited to a data integration step and a tag calculation step. Among them, the data integration step includes dirty data cleaning, format conversion and user identification merging processing of the target user information. Dirty data cleaning is used to filter data in the wrong format, format conversion is used to convert data from different data sources into a standard encoding format, and user identification merging processing is used to integrate user information of the same user, such as binding user information through user ID. The tag calculation step is used to perform batch calculations on the target user information after data integration according to the specified tag rules through a batch offline tag processing engine to generate feature tags.

[0059] In some implementations, data warehouse tools (such as hive, etc.) can be used to perform ELT (Extract-Load-Transform) processing to develop user feature label tables to facilitate the construction and management of user portraits.

[0060] Step S103: Input the target portrait information into a pre-trained interest prediction model to perform interest prediction, and obtain the target interest information of the target object.

[0061] In the embodiment of the present application, the interest prediction model is used to predict interests based on user portraits to obtain topic categories that the relevant users are interested in. The topic categories may include but are not limited to income, working hours, team atmosphere, financial situation and development prospects, etc. It can be understood that in step S103, the target interest information is used to indicate the topic categories that the target object is interested in, which can reflect the personal interest tendency of the target object.

[0062] Step S104: Generate a target conversation model according to the target interest information and the role information of the first initiating object, where the target conversation model includes at least two target conversation types.

[0063] In an embodiment of the present application, the role information of the first initiating object is a category predefined based on the position or some other criteria assigned to the first initiating object, such as the role information of the first initiating object is any one of a recommender, a personnel supervisor, a department supervisor, a department manager and a general manager. The target conversation model is used to represent the conversation flow framework adopted in the current conversation, wherein each target conversation type represents a different conversation flow node in the conversation flow framework. More specifically, the target conversation model may also include target conversation steps for each target conversation type. For example, the target conversation model may include two target conversation types, a contact interview and an attraction interview, and the conversation steps of the contact interview may at least include greetings and praises, a preliminary understanding of business topics, advantages to attract interest, invitations to cloud interviews and invitations to participate in activities, and the conversation steps of the attraction interview may at least include an in-depth understanding of the target object, the first initiating object's self-introduction, willingness communication, corporate prospect description, invitation to sign up for job training and objection handling.

[0064] In an optional implementation, step S104 specifically includes: obtaining a preset conversation model, which includes at least two preset conversation types. Obtaining keyword information for each preset conversation type, performing content matching processing on the target interest information and the role information of the first initiating object with each keyword information, obtaining a first matching result, and filtering out at least two target conversation types from all preset conversation types according to the first matching result. The first matching result indicates the degree of matching between the target interest information and the role information and the corresponding keyword information; the content matching processing can adopt the BM25 algorithm, Fast text, and a BERT-based text matching model, etc., without specific limitation.

[0065] Optionally, if the first matching result reaches a specified matching degree, the target session type can be determined according to the corresponding keyword information. Optionally, all preset session types can be sorted in descending order according to the first matching result, and the first N preset session types are taken as the target session type, where N can be a positive integer specified and adjusted manually, such as 2, 3 or 4, etc., without limitation.

[0066] In another optional implementation, step S104 may also be: obtaining a preset conversation model, wherein the preset conversation model includes at least two preset conversation types. According to the role information of the first initiating object, determine the historical users who meet the role information, and filter out the conversation type adopted by the historical users in the conversation from all preset conversation types as the target conversation type. Further, obtain all conversation steps of the target conversation type and the interest tags of each conversation step, wherein the interest tags include the M most frequently used interest information by the historical users in the conversation, where M is a positive integer. If the interest tag includes the target interest information, take the corresponding conversation step as the target conversation step.

[0067] Step S105: Generate session information of each target session type according to the target session model and the target interest information; wherein the session information is used for the first initiating object to conduct a session with the session terminal.

[0068] Steps S101 to S105 shown in the embodiment of the present application, by generating the target interest information of the target object, facilitate targeted analysis of the interest tendency of the target object, and then construct a target conversation model based on the role information and target interest information of the first initiating object, effectively sort out the target conversation type suitable for the current conversation, and can simultaneously meet the interest tendency of the target object and the conversation needs of the first initiating object. Based on this, combined with the target conversation model and the target interest information, the conversation information corresponding to each target conversation type is generated for the first initiating object to interact with the conversation terminal, thereby providing the first initiating object with conversation skills, achieving targeted conversation guidance, and helping to improve the conversation effect. In addition, it can also improve the objectivity of the conversation content, prevent the conversation initiator from exaggerating the company's profits, and thus guide reasonable recruitment.

[0069] Before step S103 in some embodiments, the conversation method further includes pre-training an interest prediction model, which may be a transformer model, an XGBoost model, or a LightGBM model, etc., without specific limitation. Figure 2 , Figure 2 It is a flowchart of training an interest prediction model in an embodiment of the present application. The training steps of the interest prediction model may include the following steps S201 to S204.

[0070] Step S201: Acquire training user information of a training object.

[0071] The training object may be a user who meets the specified screening criteria, which may include a specified user identity (such as an employee who has been employed for 1 year or an employee who has been employed for 2 years), user behavior (such as personal performance and outstanding contribution, etc.), and other people's evaluation, etc., without specific limitation. For example, the training object may be an employee who has been employed and has outstanding performance.

[0072] Step S202: construct a user portrait based on the training user information to obtain training portrait information of the training object.

[0073] The training portrait information includes at least two feature tags generated according to the training user information, which are used to label the training object. It should be noted that steps S201 and S202 may refer to the description of the above steps S101 and S102, and will not be repeated here.

[0074] Step S203: Acquire the training interest information of the training object.

[0075] The training interest information includes the topic categories that the training object is interested in.

[0076] Step S204: input the training portrait information and the training interest information into a preset initial prediction model for training processing to obtain an interest prediction model.

[0077] Step S204 can specifically be: input the training portrait information and the training interest information into a preset initial prediction model to obtain the predicted interest information. The similarity between the predicted interest information and the training interest information is calculated through the loss function of the initial prediction model, the loss function of the initial prediction model is optimized according to the similarity, the model loss of the loss function is back-propagated, and the model parameters are continuously adjusted until the similarity is greater than or equal to the similarity threshold, and the optimization of the initial prediction model is stopped to obtain an interest prediction model that meets the requirements. It should be noted that the loss function can be any one of a 0-1 loss function, an absolute value loss function, a logarithmic loss function, an exponential loss function, and a Hinge loss function, which can be selected and adjusted according to actual needs, and will not be repeated here.

[0078] It can be understood that after constructing and training the interest prediction model, the interest prediction model can also be taken as a new initial prediction model, and further training processing can be performed through steps S201 to S204, so as to continuously optimize the accuracy of the interest prediction model. Based on this, in step S203, the training interest information can be generated in the historical session in which the training object participates, and the generation method is: the training user information of the training object is input into the interest prediction model used in the historical session to perform interest prediction, and obtain the training interest information.

[0079] It can be seen that through steps S201 to S204, the interest prediction model that meets the specified screening criteria can be trained by taking the training object as a reference standard.

[0080] In step S105 of some embodiments, the session information includes target speech information of the target session type. Figure 3 , Figure 3 yes Figure 1 The specific flow chart of step S105 is shown in FIG. 1 , where step S105 may include but is not limited to the following steps S301 to S303 .

[0081] Step S301: according to the target conversation type, obtain reference speech information from a preset conversation information library.

[0082] The session information library is used to store auxiliary information used in the session, including but not limited to speech information, tool description information and multimedia information (such as pictures, audio or video, etc.). In addition, the auxiliary information can be classified according to different session types to obtain classification results for each session type, such as speech classification results, tool classification results and multimedia classification results for each session type. In step S301, the speech classification results corresponding to the target session type are retrieved from the session information library, which can be used as reference speech information.

[0083] Step S302: Input the reference speech information and the target interest information into a pre-trained first evaluation model for a first evaluation process to obtain a first matching degree.

[0084] Among them, the first matching degree is used to represent the matching degree between the reference speech information and the target interest information. The first evaluation model can be a logistic regression model, a BP neural network model, a Kalman filter prediction model, etc., which will not be repeated here. In some optional implementations, the training step of the first evaluation model includes: obtaining the historical speech information used in the historical conversation with the training object, and inputting the historical speech information and the training interest information into a preset initial evaluation model for training processing to obtain the first evaluation model.

[0085] Step S303: According to the first matching degree, target speech information is filtered out from the reference speech information.

[0086] Among them, the target speech information is the speech information used to reply to the target object.

[0087] It can be understood that the reference speech information includes at least two speech templates. In step S302, each speech template and the target interest information can be respectively input into the first evaluation model for the first evaluation process to obtain the matching degree corresponding to each speech template. Correspondingly, in step S303, as an optional implementation, speech templates with a matching degree reaching a first preset value can be screened out from the reference speech information as the target speech information. The first preset value can be manually specified and adjusted, such as the first preset value is 85% or 90%, etc., without specific limitation. As another optional implementation, all speech templates can also be sorted in order from large to small in matching degree to obtain a first sorting result, and the speech templates ranked in the first K1 positions from the first sorting result are obtained as the target speech information. K1 can be a positive integer manually set and adjusted.

[0088] It can be seen that through steps S301 to S303, the speech that fits the interest tendency of the target object can be screened out, which is conducive to achieving a better conversation guidance effect.

[0089] Furthermore, in some optional implementations, the session information also includes tool screening information for the target session type. Figure 4 , Figure 4 It is a flowchart of generating tool screening information in an embodiment of the present application. After step S105, steps S401 to S403 may also be included.

[0090] Step S401: Obtain tool description information from a session information library according to the target session type.

[0091] Among them, the tool description information is used to represent the description information of the conversation assistance tool. The conversation assistance tool can be software or plug-in running in the conversation terminal, and the conversation assistance tool can include AI interview assistance tools, career development prediction tools, income estimation tools and other personalized assistance tools, etc., without limitation. Correspondingly, the description information can be word tags (such as tool names and tool function tags, etc.), phrases and sentences, etc., also without limitation.

[0092] Step S402: Input the tool description information and the target speech information into a pre-trained second evaluation model for a second evaluation process to obtain a second matching degree.

[0093] Among them, the second matching degree is used to represent the matching degree between the tool description information and the target speech information. The second evaluation model can be a logistic regression model, a BP neural network model, a Kalman filter prediction model, etc., which will not be repeated here. In some optional implementations, the training step of the second evaluation model includes: obtaining historical auxiliary tools used in historical conversations with the training object, thereby determining the description information of the historical auxiliary tools, and then inputting the above-mentioned historical speech information and the description information of the historical auxiliary tools into a preset initial evaluation model for training processing to obtain a second evaluation model.

[0094] Step S403: According to the second matching degree, filter out tool screening information from the tool description information.

[0095] The tool screening information is used to screen out target assistive tools from conversation assistive tools.

[0096] It can be understood that if the conversation assistance tool includes at least two auxiliary tools, the tool description information includes the description information of each auxiliary tool. In step S402, the description information of each auxiliary tool can be respectively input into the second evaluation model with the target speech information for the second evaluation process to obtain the matching degree corresponding to the description information of each auxiliary tool. Correspondingly, in step S403, as an optional implementation, the description information with a matching degree reaching a second specified value can be screened out from the tool description information as tool screening information. The second specified value can be manually specified and adjusted, such as specifying a matching degree of 85% or 90%, etc., without specific limitation. As another optional implementation, the description information of all auxiliary tools can also be sorted in order of matching degree from large to small to obtain a second sorting result, and the description information arranged in the first K2 positions from the second sorting result is obtained as the tool screening information. K2 can be a positive integer manually set and adjusted.

[0097] It can be seen that through steps S401 to S403, auxiliary tools can be screened in combination with speech information suitable for the target object, which can fully utilize the extended functions of the auxiliary tools and further improve the conversation guidance effect.

[0098] In step S105 of some embodiments, the session information may also include target multimedia information of the target session type. Similarly, reference multimedia information may be obtained from the session information library according to the target session type. The target portrait information and the reference multimedia information are input into a pre-trained third evaluation model for third evaluation processing to obtain a third matching degree, which is used to represent the matching degree between the target portrait information and the reference multimedia information. According to the third matching degree, the target multimedia information is screened out from the reference multimedia information, and personalized audio and video recommendations can be implemented according to the user portrait of the target object. The above-mentioned implementation method of generating the target multimedia information can refer to the description of the above-mentioned steps S301 to S303, which will not be repeated here.

[0099] In step S104 of some embodiments, please refer to Figure 5 , Figure 5 yes Figure 1 Detailed flowchart of step S104, step S104 includes but is not limited to the following steps S501 to S506.

[0100] Step S501: Obtain a list of session initiators.

[0101] In an embodiment of the present application, the list of session initiators includes at least two session initiators arranged in order, and the first initiating object is one of the session initiators in the list of session initiators. By adjusting the arrangement order of different session initiators in the list of session initiators, the session initiation order of the session initiators can be changed, which is suitable for a more flexible session process. For example, if the list of session initiators is [referrer, department head, general manager], it means that the session is initiated to the target object in the order of the referrer, department head and general manager.

[0102] Step S502: Acquire the location information of the first initiating object in the list of session initiators, obtain the location of the initiator, and execute step S503 or step S504.

[0103] It can be understood that the initiator position is used to indicate the arrangement order of the first initiating object in the session initiator list.

[0104] Step S503: If the initiator's location is the first location, a target conversation model is generated according to the target interest information and the role information of the first initiating object.

[0105] Step S504: If the initiator position is the second position, a session initiator arranged before the first initiator object is determined from the session initiator list as the second initiator object, and steps S505 to S506 are executed.

[0106] Step S505: obtaining the evaluation result of the second initiating object on the target object and the historical conversation model generated when the second initiating object and the target object have a conversation, wherein the historical conversation model includes at least two historical conversation types.

[0107] Step S506: Generate a target conversation model based on the target interest information and the role information of the first initiating object, and in combination with the historical conversation model and the evaluation result.

[0108] The first position may be any arrangement position in the list of session initiators, and the second position indicates an arrangement position in the list of session initiators other than the first position. In an optional implementation, the first position may be the first arrangement position in the list of session initiators. In this case, if the initiator position is the first position, it indicates that the first initiating object is the first object to initiate a session with the target object, so a new target session model is directly generated through step S503; if the initiator position is the second position, it indicates that a second initiating object that has a session with the target object exists before the first initiating object, so the target session model can be generated by combining the evaluation result of the second initiating object on the target object and the adopted historical session model, thereby comprehensively considering the existing evaluation situation and dialogue situation, and further improving the pertinence and comprehensiveness of the target session model.

[0109] It is understandable that step S506 can refer to the description of step S104 and will not be described in detail. For example, in some optional implementations, after obtaining the preset conversation model and the keyword information of each preset conversation type in the preset conversation model, the target interest information, the role information of the first initiating object, the historical conversation model and the evaluation result are matched with each keyword information to obtain a second matching result, thereby filtering out at least two target conversation types from all preset conversation types according to the second matching result. The evaluation result includes but is not limited to the evaluation scores of the target object's past resume, communication skills, image and temperament, personal connections, learning ability and willingness to join the company.

[0110] It should be noted that, in actual applications, the embodiments of the present application are at least applicable to the following conversation scenarios: In some cases, if the current conversation belongs to an offline interview, the first initiator uses the conversation terminal and / or the display terminal to conduct a face-to-face interview with the target object, wherein the display terminal may refer to a terminal located in the offline interview venue, and the display terminal includes but is not limited to a projector, an interactive screen, and a speaker, etc. In other cases, if the current conversation belongs to an online interview, the first initiator uses the conversation terminal and the target object uses the communication terminal to conduct an online conversation, and the communication terminal may be a smart phone, a tablet computer, a laptop computer, or a desktop computer, etc. For example, the first initiator conducts a voice conversation with the target object through the conversation terminal, and for another example, the first initiator creates a live broadcast room on the conversation terminal, and the target object logs in to the live broadcast room on the communication terminal to achieve a live broadcast conversation effect.

[0111] Optionally, after step S105, at least one of the following information display steps may be performed: 1. Outputting session information on the session terminal for display. Specifically, target guidance information may be output on the session terminal, target multimedia information may be output on the session terminal, and corresponding target auxiliary tools may be called on the session terminal according to tool screening information, etc., mainly to guide the session of the first initiating object. 2. Sending session information to the display terminal for display, for example, sending the target multimedia information to a projector for display, so as to facilitate information sharing in an offline session location. 3. Sending session information to the communication terminal of the target object, for example, directly sending the target multimedia information to the communication terminal for display, so as to facilitate information sharing in an online session.

[0112] After step S105 in some embodiments, a session instruction triggered by the first initiating object may be obtained, and a specified session type may be selected from the target session types according to the session instruction, so as to output session information according to the specified session type. The session instruction may be an instruction generated by the session terminal in response to a trigger operation of the first initiating object, and the trigger operation may be a preset control operation captured through an operation interface of the session terminal, for example, the preset control operation may be a user clicking a specified position in the operation interface.

[0113] Based on this, in some optional implementations, a first display page can also be generated in the operation interface of the conversation terminal, and a selection control corresponding to each target conversation type can be generated in the first display page. By obtaining a conversation instruction, the target control is filtered out from all selection controls according to the conversation instruction. At this time, the conversation instruction is an instruction generated by the conversation terminal in response to the triggering operation of the first initiating object on the target control. The target conversation type corresponding to the target control is taken as the designated conversation type, so that the conversation information of the designated conversation type is output in the first display page, so that the first initiating object can choose to view the conversation information of different target conversation types according to its own needs.

[0114] For example, see Figure 6 , Figure 6 is a schematic diagram of a first display page in an embodiment of the present application. Figure 6 As shown, the conversation terminal generates two target controls of the target conversation type in the first display page, namely, a first control 601 for contact interview and a second control 602 for attracting interview. When the first initiating object clicks the first control 601, the conversation terminal outputs the conversation information of the contact interview in the first display page, such as Figure 6 Similarly, when the first initiating object clicks the second control 602, the conversation terminal can output the conversation information (not shown) of attracting the interview in the first display page.

[0115] In some embodiments, the role switching instruction triggered by the first initiating object can also be obtained, and the third initiating object can be filtered out from the list of session initiators according to the role switching instruction, so as to generate a new target session model according to the target interest information and the role information of the third initiating object, and then execute step S105 to achieve flexible switching of the session subject. The role switching instruction can be an instruction generated by the session terminal in response to the switching operation of the first initiating object, and the switching operation can be the user selecting the switching position corresponding to the third initiating object in the first display page. Figure 6 For example, when the first initiating object clicks "director", the conversation terminal generates a role switching instruction and takes the director as the third initiating object.

[0116] Furthermore, in some embodiments, after step S101, input information may be obtained, so as to update the target user information according to the input information. Specifically, the input information may refer to information input by the first initiator through the operation interface of the conversation terminal.

[0117] In one implementation, Figure 6 For example, the conversation terminal generates a first input area 603 for updating occupational information in the first display page. The conversation terminal can detect the new occupational information input by the first initiator in the first input area 603, and update the target user information according to the new occupational information.

[0118] In another implementation, the conversation terminal may also generate a second display page in the operation interface, and the second display page is used to display the target user information and evaluation results of the target initiator. The conversation terminal obtains input information through the second display page, which plays a role in timely information update. Figure 7 , Figure 7 is a schematic diagram of a second display page in an embodiment of the present application. Figure 7 In the example, when the first initiator clicks the first editing control 701, the target user information can be edited, and the conversation terminal obtains input information according to the editing situation. Alternatively, when the first initiator clicks the second editing control 702, the evaluation result can be edited, and the conversation terminal can also update the evaluation result according to the editing situation.

[0119] See also Figure 8 The embodiment of the present application also provides a conversation device, which is applied to a conversation terminal and can implement the above conversation method. The conversation device includes an acquisition module 801, a construction module 802, a prediction module 803, a first generation module 804 and a second generation module 805, wherein:

[0120] The acquisition module 801 is used to acquire target user information of a target object.

[0121] The construction module 802 is used to construct a user portrait according to the target user information to obtain the target portrait information of the target object.

[0122] The prediction module 803 is used to input the target portrait information into a pre-trained interest prediction model to perform interest prediction and obtain the target interest information of the target object.

[0123] The first generating module 804 is used to generate a target conversation model according to the target interest information and the role information of the first initiating object, where the target conversation model includes at least two target conversation types.

[0124] The second generating module 805 is used to generate session information of each target session type according to the target session model and the target interest information; wherein the session information is used for the first initiating object to perform a session interaction with the session terminal.

[0125] The specific implementation of the conversation device is basically the same as the specific implementation of the conversation method described above, and will not be described in detail here.

[0126] The embodiment of the present application also provides an electronic device, the electronic device comprising: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory, and the program is executed by the processor to realize the above-mentioned conversation method. The electronic device can be any intelligent terminal including a mobile terminal, a tablet computer, a vehicle-mounted computer, etc.

[0127] See also Fig. 9 , Fig. 9 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0128] The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0129] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 902, and the processor 901 calls and executes the session method of the embodiment of this application;

[0130] Input / output interface 903, used to implement information input and output;

[0131] Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized by wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.);

[0132] A bus 905 that transmits information between various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);

[0133] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0134] An embodiment of the present application also provides a storage medium, which is a computer-readable storage medium used for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the above-mentioned conversation method.

[0135] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0136] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0137] It can be understood by those skilled in the art that Figure 1-9 The technical solutions shown in the figure do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than those shown in the figure, or a combination of certain steps, or different steps.

[0138] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0139] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0140] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0141] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0142] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0143] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0144] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0145] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store programs.

[0146] The preferred embodiments of the present application are described above with reference to the accompanying drawings, but the scope of the rights of the present application is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present application should be within the scope of the rights of the present application.

Claims

1. A conversation method, applied to a conversation terminal, characterized in that: The method comprises: Get the target user information of the target object; Constructing a user portrait according to the target user information to obtain target portrait information of the target object; Inputting the target portrait information into a pre-trained interest prediction model to perform interest prediction, thereby obtaining target interest information of the target object; Generate a target conversation model according to the target interest information and the role information of the first initiating object, wherein the target conversation model includes at least two target conversation types; According to the target conversation type, obtaining reference speech information from a preset conversation information library; Inputting the reference speech information and the target interest information into a pre-trained first evaluation model for first evaluation processing to obtain a first matching degree; wherein the first matching degree is used to indicate the matching degree between the reference speech information and the target interest information; According to the first matching degree, target speech information is filtered out from the reference speech information; wherein the target speech information is speech information used to reply to the target object; According to the target conversation type, acquiring tool description information from the conversation information library; wherein the tool description information is used to represent description information of the conversation assistance tool; Inputting the tool description information and the target speech information into a pre-trained second evaluation model for second evaluation processing to obtain a second matching degree; wherein the second matching degree is used to indicate the matching degree between the tool description information and the target speech information; According to the second matching degree, tool screening information is screened out from the tool description information; wherein the tool screening information is used to screen out a target auxiliary tool from the conversation auxiliary tools.

2. The method according to claim 1, characterized in that: The method further includes training the interest prediction model, specifically including: Obtain training user information of the training object; Constructing a user portrait according to the training user information to obtain training portrait information of the training object; Acquiring training interest information of the training object; The training portrait information and the training interest information are input into a preset initial prediction model for training processing to obtain the interest prediction model.

3. The method according to claim 1, characterized in that After filtering out target speech information from the reference speech information according to the first matching degree, the method further includes: Obtaining a session instruction triggered by the first initiating object; Filtering a specified session type from the target session types according to the session instruction; The session information is output according to the specified session type.

4. The method according to any one of claims 1 to 3, characterized in that: After obtaining the target user information of the target object, the method further includes: Get input information; The target user information is updated according to the input information.

5. The method according to any one of claims 1 to 3, characterized in that: The generating a target conversation model according to the target interest information and the role information of the first initiating object includes: Obtaining a list of session initiators; wherein the list of session initiators includes at least two session initiators arranged in order, and the first initiating object is one of the session initiators in the list of session initiators; Acquire the location information of the first initiating object in the list of session initiators to obtain the initiator location; If the initiator's location is the first location, generating the target conversation model according to the target interest information and the role information of the first initiating object; If the initiator position is the second position, determining a session initiator arranged before the first initiator object from the session initiator list as the second initiator object; Acquire an evaluation result of the target object by the second initiating object and a historical conversation model generated when the second initiating object and the target object have a conversation, wherein the historical conversation model includes at least two historical conversation types; A target conversation model is generated according to the target interest information and the role information of the first initiating object, in combination with the historical conversation model and the evaluation result.

6. A conversation device, characterized in that: Applied to a conversation terminal, the conversation device comprises: An acquisition module is used to acquire target user information of a target object; A construction module, used to construct a user portrait according to the target user information to obtain the target portrait information of the target object; A prediction module, used for inputting the target portrait information into a pre-trained interest prediction model to perform interest prediction and obtain target interest information of the target object; A first generating module, configured to generate a target conversation model according to the target interest information and the role information of the first initiating object, wherein the target conversation model includes at least two target conversation types; The second generation module is used to obtain reference speech information from a preset session information library according to the target session type; input the reference speech information and the target interest information into a pre-trained first evaluation model for first evaluation processing to obtain a first matching degree; wherein the first matching degree is used to represent the matching degree between the reference speech information and the target interest information; according to the first matching degree, the target speech information is screened out from the reference speech information; wherein the target speech information is speech information used to reply to the target object; After the target speech information is screened out from the reference speech information according to the first matching degree, the conversation device further includes a module for executing the following steps: According to the target conversation type, acquiring tool description information from the conversation information library; wherein the tool description information is used to represent description information of the conversation assistance tool; Inputting the tool description information and the target speech information into a pre-trained second evaluation model for second evaluation processing to obtain a second matching degree; wherein the second matching degree is used to indicate the matching degree between the tool description information and the target speech information; According to the second matching degree, tool screening information is screened out from the tool description information; wherein the tool screening information is used to screen out a target auxiliary tool from the conversation auxiliary tools.

7. An electronic device, characterized in that: The electronic device includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, the steps of the conversation method as described in any one of claims 1 to 5 are realized.

8. A storage medium, the storage medium being a computer-readable storage medium, used for computer-readable storage, characterized in that: The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the conversation method in any one of claims 1 to 5.

Images