Information processing method, information processing apparatus, program product, and electronic device

Abstract

The invention provides an information processing method, an information processing device, a program product and electronic equipment, and relates to the technical field of computers. The method comprises the steps of obtaining first dialogue information of a dialogue between a target user and a first agent; generating user portrait information of the target user according to the first dialogue information; and generating second dialogue information according to the user portrait information, wherein the second dialogue information is used for a dialogue between a second intelligent agent and the target user. The information sharing of different agents is realized, the second agent can provide the targeted second dialogue information according to the personalized characteristics of the target user, and the dialogue service quality and the user experience can be improved.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the field of computer technology, and more specifically, the embodiments of the present disclosure relate to an information processing method, an information processing device, a computer program product, and an electronic device. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the disclosure that are recited in the claims, and no description herein is admitted to be prior art by inclusion in this section.

[0003] In service scenarios such as dialogue and question-and-answer sessions provided by artificial intelligence models, corresponding reply information is usually generated based on the information input by the user. Since the information that the model can obtain is relatively limited, the generated reply information is usually of low quality, which affects the user experience. Summary of the invention

[0004] Artificial intelligence models such as LLM (Large Language Model) usually generate corresponding reply information based on the prompt information input by the user, thereby realizing service scenarios such as intelligent dialogue and answering. However, the information that the prompt information can provide is relatively limited, and the reply information generated by the model based on it is usually of low quality. Moreover, in open dialogue scenarios (that is, the relevant information in the topic does not have a standard answer, such as a chat scenario), the model cannot provide adaptive information based on the user's personalized characteristics, and the experience it brings to the user is very different from the real dialogue experience.

[0005] In view of one or more of the above problems, embodiments of the present disclosure provide an information processing method, an information processing device, a computer program product, and an electronic device.

[0006] According to a first aspect of the present disclosure, there is provided an information processing method, comprising: obtaining first dialogue information of a conversation between a target user and a first intelligent agent; generating user portrait information of the target user based on the first dialogue information; generating second dialogue information based on the user portrait information, the second dialogue information being used for a conversation between a second intelligent agent and the target user.

[0007] In one embodiment, generating the user portrait information of the target user based on the first conversation information includes: dividing the first conversation information into blocks to obtain one or more conversation blocks; extracting classified memory information corresponding to one or more memory categories from the conversation blocks; and generating the user portrait information based on the classified memory information.

[0008] In one embodiment, dividing the first conversation information into blocks to obtain one or more conversation blocks includes: dividing the first conversation information into one or more conversation fragment information according to a first segmentation parameter; determining a corresponding second segmentation parameter according to a conversation frequency of the conversation fragment information, and segmenting the corresponding conversation fragment information into conversation blocks according to the second segmentation parameter.

[0009] In one embodiment, extracting classified memory information corresponding to one or more memory categories from the dialogue block includes: obtaining first prompt information corresponding to each memory category; combining the first prompt information corresponding to each dialogue block and each memory category respectively to obtain second prompt information corresponding to each dialogue block and each memory category; and processing the second prompt information using a memory extraction model to obtain classified memory information corresponding to each dialogue block and each memory category.

[0010] In one embodiment, generating the user portrait information based on the classified memory information includes: determining the basic information of the target user based on the classified memory information; matching the classified memory information with the identifier of the target user and the identifier of the first intelligent entity, and combining the basic information to obtain the user portrait information.

[0011] In one embodiment, the one or more memory categories include: a long-term memory category, a short-term memory category, and an appointment category.

[0012] In one embodiment, generating the second dialogue information based on the user portrait information includes: obtaining context information during the dialogue between the target user and the second agent; obtaining third prompt information by combining the context information and the user portrait information; and using the second agent to process the third prompt information to obtain the second dialogue information.

[0013] According to a second aspect of the present disclosure, there is provided an information processing device, comprising: a first dialogue information acquisition module, configured to acquire first dialogue information of a dialogue between a target user and a first intelligent agent; a user portrait information generation module, configured to generate user portrait information of the target user based on the first dialogue information; and a second dialogue information generation module, configured to generate second dialogue information based on the user portrait information, wherein the second dialogue information is used for a dialogue between a second intelligent agent and the target user.

[0014] According to a third aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method of the first aspect and possible implementation methods thereof are implemented.

[0015] According to a fourth aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the method of the above-mentioned first aspect and its possible implementation methods by executing the executable instructions.

[0016] The technical solution disclosed in this disclosure has the following technical effects:

[0017] The first dialogue information between the target user and the first agent is obtained, and the user portrait information of the target user is generated according to the first dialogue information. The user portrait information is an effective summary of the first dialogue information. The second agent generates the second dialogue information according to the user portrait information, which is equivalent to reusing the dialogue memory between the target user and the first agent in the second agent. Especially when the target user and the second agent have their first dialogue, the user portrait information provides sufficient reference information, realizing information sharing between the first agent and the second agent. The second agent can provide targeted second dialogue information according to the personalized characteristics of the target user, which is conducive to improving the dialogue service quality and user experience of the second agent. In addition, this scheme fully mines the dialogue information of the target user and improves the information utilization rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A flowchart showing an information processing method in this exemplary embodiment.

[0019] Figure 2 A flowchart for generating user portrait information in this exemplary embodiment is shown.

[0020] Figure 3 A schematic diagram showing a dialog segmentation in this exemplary embodiment.

[0021] Figure 4 A flowchart for obtaining classified memory information in this exemplary embodiment is shown.

[0022] Figure 5 A schematic flowchart showing an information processing method in this exemplary embodiment.

[0023] Figure 6 A schematic diagram showing the structure of an information processing device in this exemplary embodiment is shown.

[0024] Figure 7 A schematic structural diagram of an electronic device in this exemplary embodiment is shown.

[0025] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0026] The following is a detailed description of the principles and spirit of the present disclosure with reference to several representative embodiments of the present disclosure. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present disclosure, and are not intended to limit the scope of the present disclosure in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0027] The embodiments of the present disclosure may be implemented as a system, device, apparatus, method or computer program product. Therefore, the present disclosure may be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software. SUMMARY OF THE INVENTION

[0029] In the related art, artificial intelligence models such as LLM usually generate corresponding reply information based on the prompt information input by the user, thereby realizing service scenarios such as intelligent dialogue and answering. However, the information that the prompt information can provide is relatively limited, and the reply information generated by the model based on it is usually of low quality. Moreover, in open dialogue scenarios (i.e., the relevant information in the topic does not have a standard answer, such as a chat scenario), the model cannot provide adaptive information based on the user's personalized characteristics, and the experience it brings to the user is very different from the real dialogue experience.

[0030] In view of the above problems, the present disclosure provides an information processing method, an information processing device, a computer program product and an electronic device. The first dialogue information between the target user and the first agent is obtained, and the user portrait information of the target user is generated according to the first dialogue information. The user portrait information is an effective summary of the first dialogue information. The second agent generates the second dialogue information according to the user portrait information, which is equivalent to reusing the dialogue memory between the target user and the first agent in the second agent. Especially when the target user and the second agent have the first dialogue, the user portrait information provides sufficient reference information, realizing the information sharing between the first agent and the second agent. The second agent can provide targeted second dialogue information according to the personalized characteristics of the target user, which is conducive to improving the dialogue service quality and user experience of the second agent. In addition, this scheme fully mines the dialogue information of the target user and improves the information utilization rate.

[0031] After introducing the basic principles of the present disclosure, various non-limiting embodiments of the present disclosure are described in detail below.

[0032] Application Scenario Overview

[0033] It should be noted that the following application scenarios are only shown to facilitate understanding of the spirit and principle of the present disclosure, and the embodiments of the present disclosure are not limited in this respect. On the contrary, the embodiments of the present disclosure can be applied to any applicable scenario.

[0034] The embodiments of the present disclosure can be applied to diversified service scenarios. For example, some Internet service providers provide diversified services, such as providing music, customer service, chat and other services in one App (Application), or providing chat services in which agents play different roles. Different agents can be set in different services to provide corresponding intelligent services. Based on the embodiments of the present disclosure, different agents can share the conversation memory between users and agents, so that personalized service content can be generated for a specific user when serving the user, improving the user experience.

[0035] In one embodiment, the first agent and the second agent in this exemplary embodiment can be applied to the following scenarios: a chat robot in a role-playing scenario; an emotional companion robot that provides personalized care and companionship to users who need psychological support and deeply understands the user's emotions and needs; an educational counseling robot that provides customized teaching content and counseling suggestions based on students' interests and learning progress; a customer service robot that provides users with efficient and personalized customer service and quickly responds to users' questions and feedback; and a smart home assistant that understands users' living habits and preferences in a smart home environment and provides personalized services and suggestions.

[0036] Exemplary Methods

[0037] Figure 1 The exemplary process of the information processing method is shown, including the following steps S110 to S130. Each step is described in detail below.

[0038] refer to Figure 1 In step S110, first dialogue information between the target user and the first agent is obtained.

[0039] An intelligent agent refers to an intelligent system based on artificial intelligence technology. The core part of the intelligent agent is the machine learning model. In addition, it can also include relevant data (such as knowledge base, memory bank), tools (such as text language conversion programs that can be called by the model), etc. Through the combination of models, data, tools, etc., a system is formed that can receive external input information, make autonomous decisions and output information. In this exemplary embodiment, the first agent and the second agent are two different agents, and the difference between them includes but is not limited to any one or more of the following: different application scenarios, for example, the first agent is applied to an intelligent customer service scenario or a question and answer scenario in a specific field, and the second agent is applied to a chat scenario, or the first agent is applied to a general dialogue scenario, and the second agent is applied to a role-playing chat scenario, such as the second agent plays a friend of the user with a specific identity, or plays a specific game NPC (Non-Player Character); different machine learning models are used, such as the first agent includes a first language model, and the second agent includes a second language model, and the structures of the first language model and the second language model are different; different data training is used, for example, the first agent is fine-tuned and trained with question and answer sample data in customer service scenarios or specific fields on the basis of pre-training, and the second agent is trained with chat sample data on the basis of pre-training; different knowledge bases are provided, such as the first agent has a knowledge base for intelligent customer service scenarios or specific fields, and the second agent has a knowledge base common to chat scenarios (which may include relevant knowledge of language habits).

[0040] The target user may be any user with an agent dialogue record. The first dialogue information between the target user and the first agent is obtained, which may be all historical dialogue information between the target user and the first agent, or dialogue information within a certain range.

[0041] In one embodiment, the number of first agents may be multiple. For example, if the target user has had conversations with multiple agents, each of the agents is used as a first agent, and first conversation information between the target user and each first agent is obtained.

[0042] Continue to refer Figure 1 In step S120, user portrait information of the target user is generated according to the first conversation information.

[0043] The user portrait information may be information obtained by summarizing and refining the first conversation information, and may represent the conversation memory between the target user and the first agent. For example, the user portrait information may be used to describe the characteristics of the user, such as the user's gender, nickname, age, region, hobbies, language habits, etc.

[0044] In one implementation, the summary information of the first conversation information may be generated by extracting a summary, etc., as the user profile information of the target user. For example, a specific language model (such as LLM) may be used to process the task of extracting a summary of the first conversation information, and output the summary information, i.e., the user profile information.

[0045] In one embodiment, reference Figure 2 As shown, the above-mentioned generation of user portrait information of the target user according to the first conversation information may include the following steps S210 to S230:

[0046] Step S210: divide the first dialogue information into blocks to obtain one or more dialogue blocks.

[0047] For example, if a memory retrieval model (which may be a language model, such as LLM, or the first language model in the first agent) is used to generate user portrait information, the first dialogue information may be divided into blocks according to the processing capability of the memory retrieval model, so that the length of a single dialogue block meets the processing capability of the memory retrieval model. Alternatively, the first dialogue information includes multiple dialogues between the target user and the first agent, and each dialogue (multiple dialogues that are continuous in time are regarded as one dialogue, and if the time interval between two dialogues exceeds the time threshold, they can be regarded as two dialogues) is regarded as a dialogue block.

[0048] In one implementation, dividing the first dialog information into blocks to obtain one or more dialog blocks may include the following steps:

[0049] Dividing the first conversation information into one or more conversation fragment information according to the first segmentation parameter;

[0050] The corresponding second segmentation parameter is determined according to the conversation frequency of the conversation segment information, and the corresponding conversation segment information is segmented into conversation blocks according to the second segmentation parameter.

[0051] The first conversation information is actually divided into two levels according to the first segmentation parameter and the second segmentation parameter. The first conversation information is divided into conversation fragment information according to the first segmentation parameter, and then the conversation fragment information is divided into conversation blocks according to the second segmentation parameter.

[0052] The first segmentation parameter is a preliminary or basic segmentation parameter, which can be a relatively rough segmentation parameter. For example, the first segmentation parameter can be a first unit time length, such as one day, which means that the first conversation information is divided into one or more conversation fragments in units of days, and each conversation fragment includes one day's conversation information. Alternatively, the first segmentation parameter can be a time threshold. If the time interval between two adjacent conversations in the first conversation information exceeds the time threshold, the conversation is segmented between the two conversations. Through the first-level segmentation, it is easy to identify the changes in the interests or emotions of the target user.

[0053] Different conversation segment information may contain different amounts of information. Therefore, further segmentation may be performed according to the second segmentation parameter. The second segmentation parameter may be a more detailed segmentation parameter. In one embodiment, the correspondence between the conversation frequency of the conversation segment information and the second segmentation parameter may be preset, such as a linear function relationship, a segmented correspondence relationship, etc. According to the correspondence, the corresponding second segmentation parameter is determined according to the conversation frequency of each conversation segment information.

[0054] Exemplarily, the second segmentation parameter may be the number of segments, which is determined according to the conversation frequency of the conversation segment information. If the conversation frequency is high, the corresponding number of segments is large, indicating that the conversation segment information with a high conversation frequency is segmented into more conversation blocks. If the conversation frequency is low, the corresponding number of segments is small, indicating that the conversation segment information with a low conversation frequency is segmented into fewer conversation blocks or even not segmented (i.e., some conversation segment information can be directly used as conversation blocks). Alternatively, the second segmentation parameter may be the second unit duration, which is determined according to the conversation frequency of the conversation segment information. If the conversation frequency is high, the corresponding second unit duration is short, which tends to segment the conversation segment information into more conversation blocks. If the conversation frequency is low, the corresponding second unit duration is long, which tends to segment the conversation segment information into fewer conversation blocks or even not segmented. It can be seen that the exemplary embodiment adopts an adaptive second segmentation parameter, so that the segmentation result of each conversation segment information is adapted to its conversation frequency, distinguishes high-frequency conversations from low-frequency conversations, captures the density of target user interaction, and avoids too much or too little information contained in each conversation block.

[0055] Through the above two-level block division method, the first dialogue information can be fully divided into blocks, and it is ensured that each dialogue block has a suitable amount of information, which is convenient for subsequent processing.

[0056] In one implementation, based on dividing the first dialogue information into blocks, dialogue related information can be added to the dialogue blocks. Figure 3As shown, after the first conversation information is divided into blocks according to time and conversation frequency, the user ID (user_id), agent ID (robot_id), start time (start_time) and end time (end_time) of the conversation block (chunk), specific conversation content (content), keywords (keyLabel) and label groups (Labels) extracted from the conversation content can be added to each conversation block to form a conversation block with relatively complete information.

[0057] Step S220, extracting classified memory information corresponding to one or more memory categories from the dialogue block.

[0058] Since the first conversation information may contain information from different aspects, the key information therein may be classified, extracted and summarized. The memory category is a pre-set information category of the conversation memory. Exemplarily, the one or more memory categories may include but are not limited to the following three categories:

[0059] The long-term memory category covers the target user’s personal information, long-term preferences, and stable interests. For example, the target user’s occupation, hobbies, frequently mentioned topics, etc.

[0060] Short-term memory categories reflect the target user’s recent emotional state, temporary needs, and short-term interests. For example, events that the target user has recently paid attention to, questions that have been raised recently, etc.

[0061] The agreement category includes the agreements, stipulations or matters of common understanding between the target user and the first agent, such as the actions that the target user requires the first agent to take in a specific situation, special instructions agreed upon by both parties, etc.

[0062] According to the information dimensions contained in different memory categories, relevant information can be extracted from the dialogue block to obtain classified memory information corresponding to different memory categories.

[0063] In one embodiment, reference Figure 4 As shown, the above-mentioned extraction of classified memory information corresponding to one or more memory categories from the dialogue block includes the following steps S410 to 430:

[0064] Step S410, obtaining the first prompt information corresponding to each memory category.

[0065] The first prompt information may be a prompt template for extracting classified memory information for a specific memory category. The first prompt information may be pre-set for each memory category. For example, the first prompt information corresponding to the long-term memory category may be:

[0066] “Please identify the user’s gender, nickname, age, region, occupation, hobbies, long-term preferences, and stable interests based on the following conversation information;

[0067] Dialog message: {dialog block}."

[0068] Step S420, respectively combine the first prompt information corresponding to each dialogue block and each memory category to obtain the second prompt information corresponding to each dialogue block and each memory category.

[0069] Among them, each dialogue block can be combined with the first prompt information corresponding to each memory category. For example, if three memory categories are set, corresponding to three kinds of first prompt information, each dialogue block can be combined with three kinds of first prompt information. For example, the dialogue block can be filled in the specified position in the first prompt information (such as the {dialogue block} position in the first prompt information corresponding to the long-term memory category above) to form three second prompt information. In this way, the second prompt information corresponding to each dialogue block and each memory category can be obtained.

[0070] In one implementation, the first prompt information corresponding to some memory categories (such as long-term memory categories, agreed categories, etc.) may allow multiple dialog blocks to be filled in, so as to facilitate the identification of long-term information.

[0071] Step S430, using the memory retrieval model to process the second prompt information to obtain classified memory information corresponding to each dialogue block and each memory category.

[0072] The second prompt information can be input into the memory retrieval model, and the memory retrieval model can extract corresponding information according to the specified information category in the second prompt information, thereby obtaining the classified memory information corresponding to each second prompt information, that is, the classified memory information corresponding to each dialogue block and each memory category.

[0073] based on Figure 4 The method adopts different first prompt information and combines dialogue blocks to obtain second prompt information. The memory retrieval model processes the second prompt information and can better understand which aspects of information need to be extracted, thereby obtaining high-quality classified memory information and achieving a comprehensive and sufficient memory summary.

[0074] Step S230, generating user portrait information based on the classified memory information.

[0075] For example, the classified memory information corresponding to different memory categories can be annotated (each piece of classified memory information belongs to which memory category) and then combined to obtain user portrait information. Alternatively, after combining the classified memory information corresponding to different memory categories, key information can be further extracted from them to obtain user portrait information.

[0076] In one implementation, the above-mentioned generation of user portrait information based on the classified memory information may include the following steps:

[0077] Determine the basic information of the target user based on the classified memory information;

[0078] The classified memory information is matched with the identifier of the target user and the identifier of the first agent, and the basic information is combined to obtain user portrait information.

[0079] The basic information of the target user includes, but is not limited to, gender, nickname, age, region, hobbies, etc. The basic information of the target user can be determined from the classified memory information corresponding to the long-term memory category or the agreed category. The basic information can be the information highlighted in the user portrait information. For example, in the user portrait information, a corresponding entry or field can be established for each basic information to highlight each basic information.

[0080] The classified memory information can be matched with the identifier of the target user and the identifier of the first agent to indicate which user and which agent each piece of classified memory information is the conversation memory between. For example, a label "the event between user X and bot1" can be added to each piece of classified memory information (user X represents the target user, bot1 represents the first agent) or<user,bot> .

[0081] After matching different classified memory information with the target user's identifier and the first agent's identifier, the basic information is combined to obtain user portrait information. This provides structured user portrait information, ensuring the comprehensiveness and orderliness of the user portrait information, and facilitating the management and extraction of dialogue memory, so that the second agent can accurately understand the user portrait information in the subsequent process.

[0082] In one embodiment, the memory information may be organized in a hierarchical structure, such as storing classified memory information corresponding to long-term memory categories, short-term memory categories, and agreed categories in different levels, thereby enhancing the management and retrieval efficiency of information.

[0083] based on Figure 2 The method adopts the method of dialogue segmentation, which effectively reduces the amount of data processed at a time and improves the processing efficiency of massive dialogue information. On the basis of dialogue segmentation, classified memory information is extracted according to memory categories for different dialogue blocks to ensure accurate extraction of key information from complex dialogues, comprehensively and fully construct and extract dialogue memory, and generate user profile information from it, so that the user profile information covers all aspects of the target user. This processing mechanism enables the system to quickly build and update large-scale user profile information with limited computing resources.

[0084] In one embodiment, user portrait information can be constructed in the form of a knowledge graph. For example, after extracting memory information (such as classified memory information), the memory information is stored in the knowledge graph in the form of nodes and edges to obtain a knowledge graph representing the user portrait, which facilitates relationship reasoning and information expansion.

[0085] Continue to refer Figure 1 In step S130, second dialogue information is generated according to the user portrait information, and the second dialogue information is used for the second agent to communicate with the target user.

[0086] For example, user portrait information can be input into the second intelligent agent as dialogue memory, so that the second intelligent agent can fully consider and utilize the user portrait information when communicating with the target user. Even if it has not communicated with the target user, it can still more accurately understand the target user's underlying meaning and possible preferred communication methods, thereby providing high-quality second dialogue information.

[0087] In one implementation, the generating of the second conversation information according to the user portrait information may include the following steps:

[0088] Obtaining context information during the conversation between the target user and the second agent;

[0089] By combining the context information and the user portrait information, third prompt information is obtained;

[0090] The third prompt information is processed by the second agent to obtain second dialogue information.

[0091] The context information may include existing dialogue information during the dialogue process between the target user and the second agent, such as dialogue information input by the user. The context information and user profile information are combined and added to a specific prompt template to form the third prompt information. For example, the prompt template is as follows:

[0092] “Please respond to the information entered by the user. When responding, please refer to the characteristics of the user and provide a response that is suitable for the user;

[0093] Information entered by the user: {context information};

[0094] The characteristics of this user: {user portrait information};”

[0095] The context information and user portrait information are filled into the above prompt template to obtain the third prompt information. The third prompt information can then be input into the second agent, and the second agent generates the second dialogue information. The second dialogue information fully considers the dialogue memory between the target user and the first agent, conforms to the personalized characteristics of the target user, and has a high quality.

[0096] In one implementation, the user portrait information can be updated based on the conversation information between the target user and the second agent, such as periodically summarizing the conversation information between the target user and the second agent, and updating the user portrait information based on the summarized information, such as adding new memory (such as classified memory information) to the user portrait information, or modifying the original memory, thereby maintaining the conversation and service quality.

[0097] Figure 5 A schematic flow chart of this exemplary embodiment is shown. The chat information (i.e., the first conversation information) between the target user and the first agent (bot1) is divided into blocks, and the classified memory information corresponding to different memory categories is extracted from the conversation blocks through LLM, thereby obtaining the user portrait information. The second agent (bot2) has no chat history with the target user, and can chat with the target user in combination with the user portrait information, which can achieve a higher chat quality.

[0098] Moreover, this exemplary embodiment takes into account both offline batch processing and online real-time processing. The historical chat data is processed offline to generate user portrait information. At the same time, it supports real-time analysis of newly added chat records and dynamically updates user portrait information. This dual-line parallel processing mechanism ensures that the system can capture changes in user interests and needs in a timely manner, avoids the omission of newly added user information, and improves the timeliness and accuracy of user portrait information.

[0099] Exemplary Devices

[0100] The exemplary embodiment of the present disclosure also provides an information processing device. Figure 6 As shown, the information processing device 600 may include the following program modules:

[0101] A first dialogue information acquisition module 610 is configured to acquire first dialogue information of a dialogue between a target user and a first agent;

[0102] A user portrait information generating module 620, configured to generate user portrait information of a target user according to the first conversation information;

[0103] The second dialogue information generation module 630 is configured to generate second dialogue information based on the user portrait information, and the second dialogue information is used for the second agent to communicate with the target user.

[0104] In one embodiment, generating the user portrait information of the target user based on the first conversation information includes: dividing the first conversation information into blocks to obtain one or more conversation blocks; extracting classified memory information corresponding to one or more memory categories from the conversation blocks; and generating the user portrait information based on the classified memory information.

[0105] In one embodiment, dividing the first conversation information into blocks to obtain one or more conversation blocks includes: dividing the first conversation information into one or more conversation fragment information according to a first segmentation parameter; determining a corresponding second segmentation parameter according to a conversation frequency of the conversation fragment information, and segmenting the corresponding conversation fragment information into conversation blocks according to the second segmentation parameter.

[0106] In one embodiment, extracting classified memory information corresponding to one or more memory categories from the dialogue block includes: obtaining first prompt information corresponding to each memory category; combining the first prompt information corresponding to each dialogue block and each memory category respectively to obtain second prompt information corresponding to each dialogue block and each memory category; and processing the second prompt information using a memory extraction model to obtain classified memory information corresponding to each dialogue block and each memory category.

[0107] In one embodiment, generating the user portrait information based on the classified memory information includes: determining the basic information of the target user based on the classified memory information; matching the classified memory information with the identifier of the target user and the identifier of the first intelligent entity, and combining the basic information to obtain the user portrait information.

[0108] In one embodiment, the one or more memory categories include: a long-term memory category, a short-term memory category, and an appointment category.

[0109] In one embodiment, generating the second dialogue information based on the user portrait information includes: obtaining context information during the dialogue between the target user and the second agent; obtaining third prompt information by combining the context information and the user portrait information; and using the second agent to process the third prompt information to obtain the second dialogue information.

[0110] In addition, other specific details of the embodiments of the present disclosure have been described in detail in the embodiments of the above method and will not be repeated here.

[0111] Exemplary Program Products

[0112] The exemplary embodiments of the present disclosure also provide a computer program product, which includes a computer program, and the computer program implements the above method when executed by a processor.

[0113] In one embodiment, the computer program product may be a tangible product containing a computer program, such as a computer-readable storage medium storing a computer program. The readable storage medium may be a storage medium based on electrical, magnetic, optical, electromagnetic, infrared, or other signals, including but not limited to: random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory (Flash), mechanical hard disk (HDD), solid-state drive (SSD), and the like. Exemplarily, the computer program product may be implemented as a non-volatile storage medium storing a computer program, such as a read-only memory, a NAND flash memory, and the like.

[0114] In one embodiment, the computer program product may be an intangible product including a computer program. Exemplarily, the computer program product may be implemented as a virtual digital product, such as a digital file storing an executable file, an installation package, etc. of the computer program.

[0115] The code of the computer program can be written in one or more programming languages. Programming languages ​​such as C language, Java, C++, etc. The program code can be executed entirely on the user computing device, or partially on the user computing device, or as a separate software package, or partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, such as a local area network (LAN), a wide area network (WAN), etc., or can be connected to an external computing device (e.g., an Internet connection provided by an operator).

[0116] The computer program may be carried or transmitted via electrical, magnetic, optical, electromagnetic, infrared, or other signals. The electronic device may convert the signal carrying the computer program into a digital signal, thereby running the computer program. When the computer program is run on an electronic device, its code is used to enable the electronic device to execute (more specifically, the processor of the electronic device may execute) the method steps of various exemplary embodiments of the present disclosure, such as any one or more steps in the above method.

[0117] The above method steps are implemented by a computer program to obtain the first dialogue information between the target user and the first agent, and generate the user portrait information of the target user based on the first dialogue information. The user portrait information is an effective summary of the first dialogue information. The second agent generates the second dialogue information based on the user portrait information, which is equivalent to reusing the dialogue memory between the target user and the first agent in the second agent. Especially when the target user and the second agent have their first dialogue, the user portrait information provides sufficient reference information, realizing information sharing between the first agent and the second agent. The second agent can provide targeted second dialogue information based on the personalized characteristics of the target user, which is conducive to improving the dialogue service quality and user experience of the second agent. In addition, this solution fully mines the dialogue information of the target user and improves the information utilization rate.

[0118] Exemplary Electronic Devices

[0119] The exemplary embodiments of the present disclosure also provide an electronic device. The electronic device may include a processor and a memory. The memory stores executable instructions of the processor, such as a computer program. The processor executes the method steps of various exemplary embodiments of the present disclosure by executing the executable instructions.

[0120] Reference below Figure 7 , the electronic device is exemplarily described in the form of a general-purpose computing device. It should be understood that Figure 7 The electronic device 700 shown is merely an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0121] like Figure 7 As shown, the electronic device 700 may include: a processor 710 , a memory 720 , a bus 730 , an I / O (input / output) interface 740 , and a network adapter 750 .

[0122] The memory 720 may include a volatile memory, such as a RAM 721, a cache unit 722, and may also include a non-volatile memory, such as a ROM 723. The memory 720 may also include one or more program modules 724, such program modules 724 include but are not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or a combination thereof may include the implementation of a network environment. For example, the program module 724 may include each module in the above-mentioned device.

[0123] The processor 710 may include one or more processing units. For example, the processor 710 may include an AP (Application Processor), a modem processor, a GPU (Graphics Processing Unit), an ISP (Image Signal Processor), a controller, an encoder, a decoder, a DSP (Digital Signal Processor), a baseband processor and / or an NPU (Neural-Network Processing Unit) and other processing units.

[0124] The processor 710 may be configured to execute executable instructions stored in the memory 720, such as executing any one or more steps in the above method.

[0125] The processor 710 executes the above method steps to obtain the first dialogue information between the target user and the first agent, and generates the user portrait information of the target user based on the first dialogue information. The user portrait information is an effective summary of the first dialogue information. The second agent generates the second dialogue information based on the user portrait information, which is equivalent to reusing the dialogue memory between the target user and the first agent in the second agent. Especially when the target user and the second agent have their first dialogue, the user portrait information provides sufficient reference information, realizing information sharing between the first agent and the second agent. The second agent can provide targeted second dialogue information based on the personalized characteristics of the target user, which is conducive to improving the dialogue service quality and user experience of the second agent. In addition, this solution fully mines the dialogue information of the target user and improves the information utilization rate.

[0126] The bus 730 is used to realize the connection between different components of the electronic device 700, and may include a data bus, an address bus, and a control bus.

[0127] The electronic device 700 can communicate with one or more external devices 800 (eg, a keyboard, a mouse, an external controller, etc.) through the I / O interface 740 .

[0128] The electronic device 700 can communicate with one or more networks through the network adapter 750. For example, the network adapter 750 can provide mobile communication solutions such as 3G / 4G / 5G, or wireless communication solutions such as wireless LAN, Bluetooth, near field communication, etc. The network adapter 750 can communicate with other modules of the electronic device 700 through the bus 730.

[0129] although Figure 7Not shown, other hardware and / or software modules may also be provided in the electronic device 700, including but not limited to: a display, a microcode, a device driver, a redundant processor, an external disk drive array, a tape drive, and a data backup storage system.

[0130] As can be seen from the above, the technical solution of the present disclosure can be implemented as a method, an apparatus, a system, a computer program product, a storage medium, an electronic device, etc. Those skilled in the art can understand that various aspects of the present disclosure can be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation combining hardware and software, such as being respectively referred to as a "circuit", "module" or "system".

[0131] It should be understood that the present disclosure is not limited to the specific method steps or structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from the scope thereof. Those skilled in the art will easily think of other embodiments based on the specific embodiments provided by the present disclosure. Therefore, the specific embodiments provided by the present disclosure are only exemplary, and the scope and spirit of the present disclosure are indicated by the claims, and any variations, uses or adaptive changes of the present disclosure should be covered, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the technical field that are not disclosed in the present disclosure.

Claims

1. An information processing method, characterized in that: include: Acquire first dialogue information of a dialogue between a target user and a first agent; generating user portrait information of the target user according to the first conversation information; Second dialogue information is generated based on the user portrait information, and the second dialogue information is used for the second agent to communicate with the target user.

2. The method according to claim 1, characterized in that The generating the user portrait information of the target user according to the first conversation information includes: Divide the first dialogue information into blocks to obtain one or more dialogue blocks; extracting classified memory information corresponding to one or more memory categories from the dialogue block; The user portrait information is generated according to the classified memory information.

3. The method according to claim 2, characterized in that The step of dividing the first dialogue information into blocks to obtain one or more dialogue blocks includes: Dividing the first conversation information into one or more conversation fragment information according to a first segmentation parameter; A corresponding second segmentation parameter is determined according to the conversation frequency of the conversation segment information, and the corresponding conversation segment information is segmented into conversation blocks according to the second segmentation parameter.

4. The method according to claim 2, characterized in that: The step of extracting classified memory information corresponding to one or more memory categories from the conversation block includes: Obtaining first prompt information corresponding to each memory category; Combining the first prompt information corresponding to each dialogue block and each memory category respectively to obtain the second prompt information corresponding to each dialogue block and each memory category; The second prompt information is processed using a memory retrieval model to obtain classified memory information corresponding to each dialogue block and each memory category.

5. The method according to claim 2, characterized in that: The generating the user portrait information according to the classified memory information includes: Determine the basic information of the target user according to the classified memory information; The classified memory information is matched with the identifier of the target user and the identifier of the first intelligent agent, and the basic information is combined to obtain the user portrait information.

6. The method according to any one of claims 2 to 5, characterized in that: The one or more memory categories include: long-term memory category, short-term memory category, and appointment category.

7. The method according to claim 1, characterized in that The generating the second dialogue information according to the user portrait information includes: Acquiring context information during a conversation between the target user and the second agent; Obtaining third prompt information by combining the context information and the user portrait information; The third prompt information is processed by the second agent to obtain the second dialogue information.

8. An information processing device, characterized in that: include: A first dialogue information acquisition module is configured to acquire first dialogue information of a dialogue between a target user and a first agent; A user portrait information generating module, configured to generate user portrait information of the target user according to the first conversation information; The second dialogue information generation module is configured to generate second dialogue information based on the user portrait information, and the second dialogue information is used for the second agent to communicate with the target user.

9. A computer program product having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 7 by executing the executable instructions.

Images