Information processing method, question information processing method, and information processing system

Abstract

Embodiments of the present specification provide an information processing method, a question information processing method and an information processing system, wherein the information processing method comprises: determining question information submitted by a user, and obtaining knowledge information associated with the question information; determining at least one information unit based on the question information, and constructing prompt information based on the question information and question processing information contained in the at least one information unit; inputting the prompt information and the knowledge information into an information processing model to obtain at least one set of candidate answer information, and sending the at least one set of candidate answer information to a target user; receiving feedback information submitted by the target user in response to the at least one set of candidate answer information, and determining target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information.

Description

Technical Field

[0001] The embodiments in this specification relate to the field of information processing technology, and in particular to information processing methods, problem information processing methods, and information processing systems. Background Technology

[0002] In online Q&A scenarios such as e-commerce customer service, education, and consultation, user questions are typically answered by AI based on content from a pre-set knowledge base. This approach often results in answers that don't match the user's question, negatively impacting the user experience. While human Q&A can improve the quality of answers, it involves high training costs and low problem-solving efficiency. Furthermore, it struggles to provide timely answers to unfamiliar questions. Therefore, a more effective information processing method is urgently needed to address these issues. Summary of the Invention

[0003] In view of this, embodiments of this specification provide information processing methods. One or more embodiments of this specification also relate to an information processing system, an information processing apparatus, a problem information processing method, a problem information processing device, a computing device, a computer-readable storage medium, and a computer program product, to address the technical deficiencies existing in the prior art.

[0004] According to a first aspect of the embodiments of this specification, an information processing method is provided, comprising: Determine the question information submitted by the user, and obtain the knowledge information associated with the question information; Based on the problem information, at least one information unit is determined, and a prompt message is constructed based on the problem information and the problem processing information contained in the at least one information unit; The prompt information and the knowledge information are input into the information processing model to obtain at least one set of candidate answer information, and the at least one set of candidate answer information is sent to the target user; Receive feedback information submitted by the target user in response to the at least one set of candidate answer information, and determine the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information.

[0005] According to a second aspect of the embodiments of this specification, another information processing method is provided, applied to an information processing terminal, including: Receive problem information sent by the platform server and obtain the knowledge information associated with the problem information; Based on the problem information, at least one information unit is determined, and a prompt message is constructed based on the problem information and the problem processing information contained in the at least one information unit; The prompt information and the knowledge information are input into the information processing model to obtain at least one set of candidate answer information, and the at least one set of candidate answer information is sent to the target user; Receive feedback information submitted by the target user for the at least one set of candidate answer information, and determine the target answer information corresponding to the question information based on the feedback information in the at least one set of candidate answer information.

[0006] According to a third aspect of the embodiments of this specification, an information processing system is provided, including a platform server, a user terminal, an information processing terminal, and a user server, comprising: The platform server is used to receive the problem information submitted by the user terminal and send the problem information to the information processing terminal and the user server. The information processing terminal is used to acquire knowledge information associated with the question information; determine at least one information unit based on the question information, and construct prompt information based on the question information and the question processing information contained in the at least one information unit; input the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information, and send the at least one set of candidate answer information to the user server. The user server is configured to determine the target answer information corresponding to the question information from the at least one set of candidate answer information, and send the target answer information to the user terminal.

[0007] According to a fourth aspect of the embodiments of this specification, an information processing apparatus is provided, comprising: The determination module is configured to determine the question information submitted by the user and obtain the knowledge information associated with the question information; The construction module is configured to determine at least one information unit based on the problem information, and to construct prompt information based on the problem information and the problem handling information contained in the at least one information unit; The input module is configured to input the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information, and send the at least one set of candidate answer information to the target user; The receiving module is configured to receive feedback information submitted by the target user in response to the at least one set of candidate answer information, and determine the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information.

[0008] According to a fifth aspect of the embodiments of this specification, another information processing apparatus is provided, comprising: The receiving module is configured to receive problem information sent by the platform server and obtain the knowledge information associated with the problem information; The construction module is configured to determine at least one information unit based on the problem information, and to construct prompt information based on the problem information and the problem handling information contained in the at least one information unit; The input module is configured to input the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information; The determination module is configured to receive feedback information submitted by a service user in response to the at least one set of candidate answer information, and determine the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information.

[0009] According to a sixth aspect of the embodiments of this specification, a problem information processing method is provided, comprising: Determine the question information submitted by the user, and obtain the knowledge information associated with the question information; Based on the problem information, at least one information unit is determined, and a prompt message is constructed based on the problem information and the problem processing information contained in the at least one information unit; The prompt information and the knowledge information are input into the information processing model to obtain at least one set of candidate answer information, and the at least one set of candidate answer information is sent to the customer service terminal. The customer service terminal is used to generate candidate answer prompt cards based on the at least one set of candidate answer information, and push the candidate answer prompt cards to the customer service through a virtual customer service object. The system receives feedback information submitted by the customer service representative to the candidate answer hint card, determines the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information, and sends the target answer information to the user.

[0010] According to a seventh aspect of the embodiments of this specification, a problem information processing apparatus is provided, comprising: The acquisition module is configured to determine the question information submitted by the user and acquire the knowledge information associated with the question information; The construction module is configured to determine at least one information unit based on the problem information, and to construct prompt information based on the problem information and the problem handling information contained in the at least one information unit; The input module is configured to input the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information, and send the at least one set of candidate answer information to the customer service terminal, wherein the customer service terminal is used to generate candidate answer prompt cards based on the at least one set of candidate answer information, and push the candidate answer prompt cards to the customer service through a virtual customer service object; The receiving module is configured to receive feedback information submitted by the customer service representative to the candidate answer prompt card, determine the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information, and send the target answer information to the user.

[0011] According to an eighth aspect of the embodiments of this specification, a computing device is provided, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the above-described information processing method.

[0012] According to a ninth aspect of the embodiments of this specification, a computer-readable storage medium is provided that stores computer-executable instructions, which, when executed by a processor, implement the steps of the information processing method described above.

[0013] According to a tenth aspect of the embodiments of this specification, a computer program product is provided, including a computer program or instructions that, when executed by a processor, implement the steps of the information processing method described above.

[0014] This specification provides an embodiment of an information processing method that determines user-submitted question information and obtains knowledge information associated with the question information. Based on the question information, at least one information unit is determined, and prompt information is constructed based on the question information and the question processing information contained in the at least one information unit. Determining information units matching the question information allows for on-demand selection of information units. The prompt information and knowledge information are input into an information processing model to obtain at least one set of candidate answer information, which is then sent to the target user. The target user can select from the at least one set of candidate answer information according to the actual situation and submit feedback information based on the selection result. Upon receiving the feedback information submitted by the target user for the at least one set of candidate answer information, the target answer information corresponding to the question information can be determined from the at least one set of candidate answer information based on the feedback information. Determining the target answer information from the at least one set of candidate answer information based on the feedback information submitted by the target user assists the target user in answering the user-submitted question information, improving the efficiency and accuracy of the target user's answering of question information. It provides a self-service answering method for the target user, increasing the target user's willingness to answer user-submitted questions and improving the target user's response efficiency to question information. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating an information processing method provided in one embodiment of this specification; Figure 2This is a technical link architecture diagram of an information processing method provided in one embodiment of this specification; Figure 3 This is an order illustration of an information processing method provided in one embodiment of this specification; Figure 4 This is a schematic diagram illustrating a direct derivation of an information processing method provided in one embodiment of this specification; Figure 5 This is a schematic diagram showing an adjustment of an information processing method provided in one embodiment of this specification; Figure 6 This is a schematic diagram of information processing interaction provided in one embodiment of the information processing method of this specification; Figure 7 This is a schematic diagram of the structure of an information processing system provided in one embodiment of this specification; Figure 8 This is a schematic diagram illustrating the problem information processing interaction of an information processing system provided in one embodiment of this specification; Figure 9 This is a flowchart of another information processing method provided in one embodiment of this specification; Figure 10 This is a flowchart illustrating a problem information processing method provided in one embodiment of this specification; Figure 11 This is a schematic diagram of the structure of an information processing device provided in one embodiment of this specification; Figure 12 This is a schematic diagram of the structure of another information processing device provided in one embodiment of this specification; Figure 13 This is a schematic diagram of the structure of a problem information processing device provided in one embodiment of this specification; Figure 14 This is a structural block diagram of a computing device provided in one embodiment of this specification. Detailed Implementation

[0016] Many specific details are set forth in the following description to provide a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0017] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of this specification. The singular forms “a,” “described,” and “the” as used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0018] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this specification, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0019] Furthermore, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties. Moreover, the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0020] The technical solutions provided in this application can employ deep learning models with relatively large parameter scales. However, this large model is merely an example; this application does not limit the number of model parameters supported by the deep learning model used, aiming to meet actual needs. The deep learning models involved in this application can be artificial intelligence-based language models (LM) or multimodal models (MM).

[0021] First, the terms and concepts used in one or more embodiments of this specification will be explained.

[0022] Automatic Speech Recognition (ASR) refers to the technological process of automatically converting human speech signals into corresponding text or processable instructions. Its core lies in utilizing acoustic models, language models, and signal processing algorithms to extract features, decode phonemes, and match words in audio, addressing the challenges of different accents, speaking speeds, and background noise.

[0023] OCR (Optical Character Recognition) refers to the technology of automatically detecting, segmenting, and recognizing printed or handwritten text in images or paper documents, and converting it into editable and searchable electronic text. This process encompasses steps such as image preprocessing (e.g., binarization, denoising, skew correction), layout analysis, character feature extraction, and classifier matching, while relying on deep learning models to improve robustness to complex fonts, blurry photos, and artistic fonts.

[0024] Prompts are carefully designed input texts or instruction templates used by users to guide Large Language Models (LLMs) or other generative AIs to produce specific outputs. Their function is to clarify task objectives, set response formats, inject contextual constraints, or stimulate the model's reasoning chain, thereby controlling the style, depth, and accuracy of the generated content.

[0025] In the field of artificial intelligence, an agent is a software entity or system capable of autonomously perceiving its environment, making decisions, and executing actions to achieve predefined goals. It typically integrates the reasoning capabilities of a large language model, memory modules, tool calling interfaces (such as APIs and databases), and planning strategies. It can break down complex tasks into multi-step sub-tasks, dynamically select external tools to obtain information or perform operations, and continuously adjust its behavior based on feedback.

[0026] FAQ (Frequently Asked Questions): In a collective context, FAQs refer to standardized information entries prepared and organized in a question-and-answer format by the operations or technical support team for a specific product, service, or area. Each FAQ typically includes a typical user question (or a variation of the question) and its corresponding authoritative standard answer, along with category tags, applicable versions, and a keyword index.

[0027] To address the aforementioned technical problems, this specification provides an information processing method. This specification also relates to an information processing system, an information processing apparatus, a problem information processing method, a problem information processing apparatus, a computing device, a computer-readable storage medium, and a computer program product, which will be described in detail in the following embodiments.

[0028] See Figure 1 , Figure 1 A flowchart of an information processing method according to an embodiment of this specification is shown, which specifically includes the following steps.

[0029] Step 102: Determine the problem information submitted by the user and obtain the knowledge information associated with the problem information.

[0030] Specifically, users can be the recipients of services in online Q&A scenarios such as e-commerce customer service, teaching, and consultation. In e-commerce scenarios, users can be consumers or merchants on e-commerce platforms; in teaching scenarios, users can be users of teaching applications; and in online consultation scenarios, users can be the inquirers. Correspondingly, the question information refers to the user's inquiry that needs to be resolved or requires an accurate answer. The knowledge information associated with the question information includes, but is not limited to, the consultation recipient included in or corresponding to the question information, detailed information about the consultation recipient, user behavior information, user interaction information, the answer rules corresponding to the question information, general knowledge, and domain knowledge.

[0031] Based on this, after determining the question information submitted by the user, the associated knowledge information can be obtained. This associated knowledge information includes, but is not limited to, domain knowledge related to the question, the person being consulted in the question, detailed information about that person, user behavior information, and user interaction information. The purpose of obtaining this associated knowledge information is to subsequently search or retrieve auxiliary information within this knowledge database to find help in answering the question.

[0032] In practical applications, within e-commerce scenarios, user-submitted questions can be linked to products and services offered by the e-commerce platform. When the question concerns a product, the associated knowledge information includes, but is not limited to, product knowledge graphs and the user's historical orders. For example... Figure 2 As shown, the information processing method includes an input layer, a parsing layer, a understanding layer, a retrieval layer, a generation layer, an evaluation layer, an output layer, and an optimization layer. After determining the question information submitted by the user, the question information can be input into the input layer as a consumer multimodal message. The message types of the consumer multimodal message include, but are not limited to: text, images, video, and audio.

[0033] Step 104: Determine at least one information unit based on the problem information, and construct a prompt message based on the problem information and the problem handling information contained in the at least one information unit.

[0034] Specifically, after determining the user-submitted question information and obtaining the associated knowledge information, at least one information unit can be identified based on the question information. Then, prompt information is constructed based on the question information and the question processing information contained in at least one information unit. Here, an information unit refers to a skill service unit, which is a standardized service unit that can be executed independently. Each skill encapsulates the complete processing logic for a specific domain task scenario, i.e., the question processing information. The question processing information specifically includes: intent matching rules (defining the intent type and triggering conditions of the skill's response), knowledge retrieval scope (specifying the knowledge graph subgraph, knowledge base partition, and FAQ set associated with the skill), a Prompt template (predefining prompt words, role settings, output format requirements, etc. for this scenario), a response generation strategy (including parameters such as style, content emphasis, and length control), and post-processing rules (sensitive word filtering rules, compliance verification rules, etc.). The prompt information is the model prompt word used to input into the information processing model and guide the model in generating the answer.

[0035] Based on this, after determining the user-submitted question information and obtaining the associated knowledge information, at least one information unit is determined based on the question information. Each information unit is a skill unit related to the question information, and each information unit contains question-processing information to assist in answering the question. Based on the question information and the question-processing information contained in at least one information unit, prompt information is constructed. This prompt information can then be input into the information processing model to assist in generating candidate answer information.

[0036] Furthermore, considering that the question information is directly provided by the user, it needs to be structured to convert the question information into structured question text. This allows for rapid intent recognition and object recognition based on the structured question text. The specific implementation is as follows: The problem information is converted into problem text, and the problem text is structured to obtain structured problem text corresponding to the problem text; intent recognition is performed on the structured problem text to obtain intent information, and object recognition is performed on the structured problem text to obtain object information; initial prompt information and processing logic information are determined from the problem processing information contained in the at least one information unit; the initial prompt information is updated based on the intent information, the object information, and the processing logic information to obtain the prompt information.

[0037] Specifically, the question information can be multimodal, including but not limited to voice, text, images, and video. When the question information contains voice content, ASR (Automatic Speech Recognition) is required to convert the voice content into text. When the question information contains images, OCR (Optical Character Recognition) or image recognition is required to convert the images into text. When the question information contains video content, keyframes need to be extracted from the video content, followed by image recognition to convert it into text. Structured processing of the question text refers to extracting and arranging key information according to preset fields. Preset fields include, but are not limited to, message type (mixed text and image), text content, recognized entities, and original modalities. Intent recognition of the structured question text aims to identify the consultation intent contained in the question information, including but not limited to product selection, returns, and after-sales consultation. Object recognition of the structured question text refers to identifying the consultation object contained or associated with the question information. Consultation objects include, but are not limited to, products and orders. Initial prompt information refers to the prompt template, i.e., the Prompt template. Processing logic information includes, but is not limited to, the knowledge retrieval scope, response generation strategy, and post-processing rules.

[0038] Based on this, when the question information includes image, audio, and / or video information, the question information is converted into question text, and the question text is structured according to preset fields to obtain the structured question text. The structured question text contains key information arranged in a structured manner, such as message type, text content, identified entity, and original modality. Intent recognition of the structured question text can obtain the user's intent information and clarify the intended consultation. Object recognition of the structured question text can obtain the object information of the user's consultation object. Initial prompt information and processing logic information are determined from the question processing information contained in at least one information unit, and the corresponding information fields in the initial prompt information are filled based on the intent information, object information, and processing logic information to obtain the prompt information. When at least one information unit consists of two or more information units, a prompt information can be constructed based on the initial prompt information of each information unit, resulting in two or more sets of prompt information.

[0039] For example, in an e-commerce scenario, the user is the consumer. If a consumer sends a product image with the text "Can this be returned or exchanged?", the content sent is the question information. The question information contains both text and an image. OCR recognition is needed to extract the product name / model from the image, while the text is directly retained. The text and the extracted product name / model are then integrated into a structured semantic representation, i.e., structured question text. Structured question text includes fields such as message type (text and image mix), text content, identified entity, and original modality. The advantage of structuring question information is that subsequent intent recognition, object (entity) extraction, and knowledge retrieval can directly read fixed fields without repeatedly parsing the question information, reducing coupling and improving processing efficiency. After determining the structured question text, intent recognition and object recognition can be performed on it. The recognized intent information can be product after-sales service, and the recognized object information can be the product model. If at least one information unit is an after-sales service information unit, the after-sales prompt word template (initial prompt information) and after-sales processing logic (processing logic information) can be determined based on the after-sales service information unit. By filling the intent information, object information, and processing logic information into the corresponding fields of the prompt word template, you can obtain the prompt word (prompt information).

[0040] In summary, by updating the initial prompt information based on intent information, object information, and processing logic information, and obtaining prompt information, the system can populate the initial prompt information with key information based on intent information, object information, and processing logic information, thus obtaining prompt information that can be input into the information processing model and ensuring the richness of the prompt information.

[0041] Furthermore, considering that user intent is only identified in the form of keywords, subtle intents that cannot be obtained through text recognition may be missed. Therefore, user intent information can be analyzed and represented from multiple dimensions to improve the accuracy of intent information. The specific implementation is as follows: The structured question text is input into the intent recognition model to obtain the intent type and the intent score corresponding to the intent type; the user's question status information is determined based on the structured question text, and the intent type, the intent score, and the question status information are used as the intent information.

[0042] Specifically, the intent recognition model can be a large language model used to identify the intent type of a user's submitted question based on structured question text, as well as the corresponding intent score. The intent type represents the purpose of the user submitting the question, while the intent score represents the intensity of the user's expectation of an answer. The user's question status information can indicate the urgency or necessity of the user's expectation of an answer. In e-commerce scenarios, the intent recognition model can be a large language model fine-tuned for the e-commerce domain.

[0043] Based on this, the structured question text is input into the intent recognition model. The intent recognition model can identify the user's intent type and the corresponding intent score based on the structured question text. The user's question status information, indicating the urgency of the question, is then determined based on the structured question text. The intent type, intent score, and question status information are then used as the intent information.

[0044] Continuing with the previous example, after semantically encoding the structured question text using a finely tuned large language model from the e-commerce domain, the encoded vector is input into a multi-level intent classifier, which outputs the intent classification and confidence score. The intent classification represents the user's intent type, and the confidence score refers to the intent score. User state recognition is then performed to determine the current question's state and urgency.

[0045] In summary, using intent type, intent score, and question status information as intent information enriches the representation of user intent. Using intent score and question status information as intent information facilitates quick and accurate identification and timely handling of urgent questions, thereby improving the user's question-and-answer experience.

[0046] Furthermore, considering that the structured question text does not directly contain the user's consultation object, it is necessary to determine the entity information corresponding to the structured question text, and then query the object database based on the entity information before determining the object information. As a stage-wise processing result corresponding to the question information, the object information can be sent to the user's corresponding service user before obtaining the answer information corresponding to the question information. The specific implementation is as follows: The process involves: identifying entity information in the structured question text; querying an object database based on the entity information to obtain the object information associated with the entity information; and after obtaining the object information by performing object recognition on the structured question text, the process further includes: identifying the service user corresponding to the user and displaying the object information on the service user's service user page.

[0047] Specifically, entity information refers to the object attributes of the object being asked in the structured question text. These object attributes include, but are not limited to, order number, order time, and order status. The object database can be a product database or an order database. Object information is essentially product information, including but not limited to product ID. A service user is a service provider who answers a user's question; a service user is a customer service representative who answers a user's question. The service user page refers to the dialogue page between the service user and the user.

[0048] Based on this, entity recognition can be performed on structured question text to identify entity information. By querying object databases such as order databases and / or product databases based on this entity information, the associated object information can be obtained. After obtaining object information through object recognition of the structured question text, the corresponding service user can also be identified, and the object information can be displayed on the service user's page. Service users can promptly obtain the interim processing results of the question information on their service user page, facilitating their understanding of the processing progress.

[0049] Continuing with the previous example, after structuring the user-submitted question information to obtain the structured question text "Where is the package for order 2025060001?", entity information is extracted from the structured question text: Order Number = 2025060001, Inquiry Type = Logistics Inquiry. Based on the order number, the specific order and logistics information are retrieved and matched in the order database. Simultaneously, the corresponding product is associated in the product knowledge graph to determine the object information corresponding to the product. The object information includes order details and product information. This object information is then pushed to... Figure 3 The panel showing the contact information for service users on the right side of the customer service workbench is displayed. This allows service users to easily track the progress of their issues.

[0050] In summary, by identifying the corresponding service user for each user and displaying the object information on that user's service user page, the system effectively feeds back the object information obtained from real-time problem identification to the user's corresponding service user. This allows the service user to promptly grasp the processing progress and interim results of the problem information.

[0051] Step 106: Input the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information, and send the at least one set of candidate answer information to the target user.

[0052] Specifically, after identifying at least one information unit based on the question information and constructing prompt information based on the question information and the problem-solving information contained in the at least one information unit, the prompt information and knowledge information can be input into the information processing model to obtain at least one set of candidate answer information. This at least one set of candidate answer information is then sent to the target user. The information processing model can be a large language model used to generate solutions and / or answers corresponding to the user-submitted question information. The candidate answer information is the solutions and / or answers provided to the target user for identification and selection. The target user is the person who solves the problem information for the user; in the case of a consumer on an e-commerce platform, the target user is the customer service representative.

[0053] Based on this, after determining at least one information unit based on the question information and constructing prompt information based on the question information and the question processing information contained in the at least one information unit, the prompt information and knowledge information are input into the information processing model. Utilizing the solution generation and / or answer generation capabilities of the information processing model, at least one set of candidate answer information is generated. This at least one set of candidate answer information is then sent to the target user for identification and selection.

[0054] Furthermore, after determining the knowledge information, a queryable knowledge base can be identified based on this information. By querying the knowledge base, intent knowledge fragments and object knowledge fragments can be obtained, which are used to construct accurate target prompts. This ensures that the information processing model can output accurate and highly usable candidate answer information. The specific implementation is as follows: Determine at least one knowledge base and object state data associated with the knowledge information; query the at least one knowledge base based on the intent information and the object information to obtain intent knowledge fragments and object knowledge fragments; update the prompt information to target prompt information based on the object state data, the intent knowledge fragments, and the object knowledge fragments, and input the target prompt information into the information processing model to obtain the at least one set of candidate answer information.

[0055] Specifically, the knowledge base includes, but is not limited to, product knowledge bases, order knowledge bases, industry knowledge bases, and custom knowledge bases. Object state data can be dynamic product data. Intent knowledge fragments are intent knowledge in the knowledge base that matches intent information, and can represent the analysis and interpretation of intent information. Object knowledge fragments can be extended information or detailed information about the object.

[0056] Based on this, at least one knowledge base and object state data associated with the knowledge information are identified. At least one knowledge base is queried based on the intent information and object information to obtain intent knowledge fragments and object knowledge fragments. The object state field, intent field, and object field are determined in the prompt information. Based on the object state data, intent knowledge fragments, and object knowledge fragments, the prompt information is updated to the target prompt information, and the target prompt information is input into the information processing model to obtain at least one set of candidate answer information.

[0057] Following the previous example, after determining the knowledge information, knowledge bases such as the product knowledge graph, industry knowledge base, and store-customized knowledge base can be established. When the question information corresponds to a return / exchange inquiry, return / exchange rules are retrieved from the product knowledge graph based on object information (product name, etc.), and relevant knowledge fragments are output. General refund processes are retrieved from the industry knowledge base based on intent information, and industry standard fragments are output. Furthermore, store-specific after-sales rules can be retrieved from the store-customized knowledge base based on the merchant ID, and store rule fragments are output. All these fragments are assembled into a unified Prompt (target prompt information), which is then input into a large language model to generate a recommended response—that is, at least one set of candidate answer information.

[0058] In summary, by updating the prompt information to target prompt information based on object state data, intent knowledge fragments, and object knowledge fragments, and by incorporating more intent and object knowledge fragments into the target prompt information, the richness of the target prompt information is enhanced. When the target prompt information is input into the information processing model, it guides the model to generate at least one set of candidate answer information with a high degree of matching with the question information.

[0059] Furthermore, considering that the information processing model may have logical flaws in its processing of prompts and knowledge information, and is prone to model illusions, it is necessary to detect at least one set of candidate answer information output by the information processing model. The specific implementation is as follows: The prompt information and the knowledge information are input into an information processing model to obtain at least one set of initial candidate answer information; at least one answer detection dimension is determined, and the at least one set of initial candidate answer information is detected according to the at least one answer detection dimension to obtain detection information corresponding to each of the at least one set of initial candidate answer information; based on the detection information corresponding to each of the at least one set of initial candidate answer information, the at least one set of candidate answer information is determined. The detection information corresponding to any set of initial candidate answer information can represent the display status and / or display priority of the initial candidate answer information. Specifically, at least one set of initial candidate answer information is directly output by the information processing model and has not undergone accuracy testing. Answer testing dimensions include, but are not limited to, factual accuracy, sensitive information detection, state fit, format conformity, and confidence score.

[0060] Based on this, the prompt information and knowledge information are input into the information processing model to obtain at least one set of initial candidate answer information directly output by the information processing model. Considering that the information processing model may experience model illusions or have logical flaws during the reasoning process, at least one answer detection dimension can be determined. At least one set of initial candidate answer information is then detected according to this dimension, obtaining detection information corresponding to each set of initial candidate answer information. Each set of initial candidate answer information and its corresponding detection information constitutes an information pair. Any information pair can be used to adjust the initial candidate answer information based on the detection information. Adjustment methods include, but are not limited to, not displaying to the target user or downgrading the display. Based on the detection information corresponding to each set of initial candidate answer information, at least one set of candidate answer information can be determined. The determined set of candidate answer information can be displayed in a fixed order, where the display order indicates the display priority.

[0061] Following the previous example, the prompts and knowledge information are input into the information processing model to obtain at least one set of initial candidate answer information directly output by the model. At least one answer detection dimension is determined, including but not limited to: ① Factual accuracy, used to detect whether at least one set of initial candidate answer information is consistent with retrieved knowledge / real-time data; ② Sensitive information detection, used to detect whether at least one set of initial candidate answer information contains prohibited words, information that cannot be disclosed, or inappropriate promises; ③ Attitude fit, used to detect whether at least one set of initial candidate answer information matches the consumer's emergency situation; ④ Format compliance, used to detect whether the length, punctuation, and layout of at least one set of initial candidate answer information conform to customer service standards; ⑤ Confidence score, used to calculate the quality score corresponding to each of the above dimensions, and candidate answer information below a preset score threshold is not pushed to customer service or is downgraded in display. For each set of initial candidate answer information, detection can be performed sequentially according to the detection dimensions to obtain detection information. Based on the detection information, the display priority, whether to display, and other attributes of the initial candidate answer information can be adjusted to obtain candidate answer information.

[0062] In summary, by detecting at least one set of initial candidate answer information according to at least one answer detection dimension, the detection information corresponding to each of the at least one set of initial candidate answer information is obtained. Based on the detection information corresponding to each of the at least one set of initial candidate answer information, the at least one set of initial candidate answer information can be filtered to obtain at least one set of candidate answer information. This minimizes the impact of model illusion and ensures the accuracy of at least one set of candidate answer information.

[0063] Furthermore, considering that the process from determining the question information to identifying the target candidate answer information may take a considerable amount of time, resulting in long waiting times for users, a sending condition can be set. When the target candidate answer information meets the sending condition, it can be directly sent to the user without requiring manual sending by the service provider. The specific implementation is as follows: If the at least one set of candidate answer information includes target candidate answer information that meets the sending conditions, an answer information component is generated based on the target candidate answer information; and the answer information component is displayed on the user's Q&A page.

[0064] Specifically, the sending conditions include, but are not limited to, at least one of the following: confidence level, user waiting time, and quality detection information of candidate answer information. The answer information component contains target candidate answer information. If the target candidate answer information contains at least two options, then at least two options can be displayed in the answer information component as touchable buttons. These touchable buttons can correspond to return confirmation buttons and compensation confirmation buttons.

[0065] Based on this, if at least one set of candidate answer information contains target candidate answer information that meets the sending conditions, an answer information component is generated based on the target candidate answer information. This answer information component is then displayed on the user's Q&A page. The answer information component can display touchable buttons corresponding to the options available to the user, facilitating feedback from the user regarding the target candidate answer information.

[0066] Continuing with the previous example, such as Figure 4 As shown, after determining the target candidate answer information, "Dear user, the failure to heat up when powered on is generally related to poor contact or circuit problems. You can try the following steps...", the target candidate answer information can be tested. If the target candidate answer information meets the sending conditions, the answer information component of the target candidate answer information can be directly generated and sent to the user. The user can view and operate the answer information control on the user Q&A page, such as accepting a refund plan by operating the answer information control.

[0067] In summary, when at least one set of candidate answer information contains the target candidate answer information that meets the sending conditions, the answer information component can be directly displayed on the user's Q&A page without requiring any user intervention. This ensures that users can promptly obtain the answer information component containing the target candidate answer information. Presenting the target candidate answer to the user in the form of an answer information component provides an entry point for user action, thereby improving the user's Q&A experience.

[0068] Step 108: Receive feedback information submitted by the target user for the at least one set of candidate answer information, and determine the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information.

[0069] Specifically, after inputting the prompts and knowledge information into the information processing model to obtain at least one set of candidate answer information, the system can receive feedback information submitted by the target user in response to the at least one set of candidate answer information. Based on the feedback information, the system determines the target answer information corresponding to the question information from the at least one set of candidate answer information. The feedback information can be the selection result of the target user among the at least one set of candidate answer information. The target answer information is either the set of candidate answer information selected by the user or at least one set of candidate answer information determined from the at least one set of candidate answer information based on the feedback information provided by the target user.

[0070] Based on this, after inputting the prompt information and knowledge information into the information processing model to obtain at least one set of candidate answer information, feedback information submitted by the target user regarding at least one set of candidate answer information is received. The feedback information submitted by the target user represents the selection result submitted by the target user after identifying and selecting from at least one set of candidate answer information. Based on the feedback information, the target answer information corresponding to the question information can be determined from at least one set of candidate answer information, thus determining the set of candidate answer information or at least one set of candidate answer information ultimately selected by the target user.

[0071] Furthermore, considering that after sending the target answer information to the service user, there might be a mismatch between the target answer information and the user's interaction scenario, in order to ensure that the target user can quickly respond to the questions raised by the user, the service user can be provided with the function of adjusting the target answer information. After receiving the adjustment information submitted by the service user regarding the target answer information, the target answer information can be adjusted based on the adjustment information and then sent to the user. The specific implementation is as follows: The system receives adjustment information from the service user regarding the target answer information; updates the target answer information based on the adjustment information to obtain the answer information to be sent; and sends the answer information to be sent to the user.

[0072] Specifically, the adjustment information represents the user's suggestions for modifying the target answer information. This adjustment information can be used to adjust the text content, the image content, and the visual display dimensions of the target answer information, including but not limited to color and the size of the answer card. The answer information to be sent is the result of adjusting the target answer information in at least one dimension based on the adjustment information. Adjustment dimensions include, but are not limited to, text, image, and visual display dimensions.

[0073] Based on this, the system receives adjustment requests from service users regarding the target answer information. The corresponding dialog page for the service user provides adjustment controls for the target answer information, allowing the service user to submit adjustments. The system updates the target answer information based on these adjustments, obtains the answer information to be sent, and then sends the answer information to the user.

[0074] Continuing with the previous example, such as Figure 5 The page shown is the dialogue page visible to the service user (customer service personnel). Customer service personnel can see two sets of candidate answer information, Option A and Option B, on the dialogue page. If the customer service personnel choose Option A as the target answer and need to adjust it, they can submit adjustment information (e.g., changing to a vertical striped four-piece suit). The AI ​​digital human engine will then update "striped four-piece suit" in the target answer information to "vertical striped four-piece suit," obtaining the answer information to be sent. The AI ​​digital human engine can directly send the answer information to the user, or it can send it to the customer service personnel, display it on the dialogue page, and then the customer service personnel will send it to the user. The intelligent service assistant panel is located on the right side of the customer service workbench dialogue window on the dialogue page, containing a section for frequently asked questions and an AI-recommended solution section. Recommended solutions are presented in card format, including a preview of the reply content, a confidence level indicator, knowledge source annotations, and operation buttons. The recommended solutions use a dynamic sorting mechanism based on confidence level, relevance, and customer service preferences.

[0075] In summary, the target answer information is updated based on the adjustment information to obtain the answer information to be sent, and then the answer information to be sent to the user. Before being sent to the user, the target answer information can be reviewed and adjusted by the service user to ensure the accuracy and usability of the answer information to be sent.

[0076] Furthermore, after the answer information to be sent to the user is delivered, the processing of the user's question information is completed. To ensure the effectiveness of subsequent processing of other questions, relevant information such as the answers to this question and user evaluations can be collected to construct a good sample and optimize the information processing model. The specific implementation is as follows: Determine the user's evaluation information regarding the answer to be sent, and construct training samples based on the evaluation information, the answer to be sent, and the question information; fine-tune the information processing model based on the training samples.

[0077] Based on this, user evaluation information regarding the answer to be sent is determined, representing the user's level of satisfaction. Training samples are constructed based on the evaluation information, the answer to be sent, and the question information. The user's level of satisfaction with the answer to be sent can be determined based on the evaluation information. Training samples containing the answer to be sent and the question information can be constructed using the evaluation information as a reference. In the training samples, the question information can be used as sample data, and the answer to be sent can be used as label data. The training samples can be determined as positive or negative samples based on the evaluation information. The information processing model is then fine-tuned based on the training samples.

[0078] In summary, the information processing model is fine-tuned and trained using training samples constructed based on evaluation information, information to be sent, and question information to ensure that the answer generation performance of the information processing model remains at a high level.

[0079] Furthermore, at least one information unit originates from the information unit set. Considering the diversity and uncertainty of user needs, in order to meet users' question-and-answer needs as much as possible, the information unit set can be continuously supplemented. The specific implementation is as follows: Based on the unit requirement information, determine the unit knowledge, unit intent, and unit prompt information, and construct the target information unit based on the unit knowledge, unit intent, and unit prompt information; determine the information unit set associated with the at least one information unit, and add the target information unit to the information unit set.

[0080] Specifically, unit requirement information can be either processing capacity expansion requirements or scenario expansion requirements. In an e-commerce scenario, unit requirement information can be the unit requirement information corresponding to product pairing recommendation capabilities. The unit title image is the product pairing intent recognition rule, and the unit prompt information can be a product pairing recommendation prompt word template. Unit knowledge can be product pairing knowledge or the product pairing knowledge retrieval scope. The target unit information is the product pairing skill service unit, i.e., the product pairing skill. The information unit collection stores various types of skill service units available in the task platform.

[0081] Based on this, for the information unit set, the target user can update the information unit set according to actual needs. This includes, but is not limited to, updating or deleting existing unit information in the information unit set, and adding new unit information. If the target user submits unit requirement information, it indicates that the target user has an update requirement for the information unit set. The user can then determine the unit knowledge, unit intent, and unit prompt information submitted in response to the unit requirement information, and construct the target information unit based on these information. The unit information, unit intent, and unit prompt information can also be generated using a large model if the user submits a unit type. At least one information unit set associated with the information unit is identified, and the target information unit is added to the information unit set. Adjustments can also be made to existing information units in the information unit set.

[0082] Continuing with the previous example, if an e-commerce platform's customer service has a request for a new product pairing recommendation function, based on the customer service's provided information, a target information unit can be constructed that includes product pairing recommendation unit knowledge, product pairing recommendation unit intent, and product pairing recommendation unit prompts. This is known as the product pairing recommendation information unit – the product pairing recommendation Skill. This information unit is stored in an information unit set. Subsequent user-submitted questions about product pairings can then be matched to the relevant product pairing recommendation information unit in the set. This standardized encapsulation of Skills allows for the rapid replication and promotion of high-quality service capabilities across different merchants and scenarios, achieving the large-scale accumulation and sharing of service capabilities.

[0083] In summary, by adding the target information units generated based on the unit requirement information to the information unit set, subsequent user-submitted questions can be matched against information units in the updated information unit set. Existing information units in the information unit set can also be appropriately adjusted. Adjusting the information unit set achieves the goal of expanding service scenarios and enhancing service capabilities without altering the underlying development pipeline, greatly improving the scalability and response speed of domain tasks.

[0084] This specification provides an embodiment of an information processing method that determines user-submitted question information and obtains knowledge information associated with the question information. Based on the question information, at least one information unit is determined, and prompt information is constructed based on the question information and the question processing information contained in the at least one information unit. Determining information units matching the question information allows for on-demand selection of information units. The prompt information and knowledge information are input into an information processing model to obtain at least one set of candidate answer information, which is then sent to the target user. The target user can select from the at least one set of candidate answer information according to the actual situation and submit feedback information based on the selection result. Upon receiving the feedback information submitted by the target user for the at least one set of candidate answer information, the target answer information corresponding to the question information can be determined from the at least one set of candidate answer information based on the feedback information. Determining the target answer information from the at least one set of candidate answer information based on the feedback information submitted by the target user assists the target user in answering the user-submitted question information, improving the efficiency and accuracy of the target user's answering of question information. It provides a self-service answering method for the target user, increasing the target user's willingness to answer user-submitted questions and improving the target user's response efficiency to question information.

[0085] The following is in conjunction with the appendix Figure 6 Taking the application of the information processing method provided in this specification in the user question-and-answer field of e-commerce as an example, the information processing method will be further explained. Figure 6 This specification illustrates an information processing interaction diagram of an information processing method according to an embodiment of the present specification, which specifically includes the following steps.

[0086] Step 602: Determine the problem information submitted by the user and obtain the knowledge information associated with the problem information.

[0087] In practical applications, users can be the recipients of services in online Q&A scenarios such as e-commerce customer service, teaching, and consultation. In e-commerce scenarios, users can be consumers or merchants on e-commerce platforms. The question information refers to the user's inquiry that needs to be resolved or requires an accurate answer. The knowledge information associated with the question information includes, but is not limited to, the consultation recipient included in or corresponding to the question information, detailed information about the consultation recipient, user behavior information, user interaction information, the corresponding answering rules, general knowledge, and domain knowledge.

[0088] Step 604: Convert the problem information into problem text, and perform structured processing on the problem text to obtain the structured problem text corresponding to the problem text.

[0089] Step 606: Perform intent recognition on the structured question text to obtain intent information, and perform object recognition on the structured question text to obtain object information.

[0090] Step 608: Based on the problem information, determine at least one information unit, and in the problem processing information contained in the at least one information unit, determine the initial prompt information and the processing logic information.

[0091] Step 610: Update the initial prompt information based on the intent information, object information, and processing logic information to obtain the prompt information.

[0092] In practical applications, question information can be multimodal, including but not limited to speech, text, images, and video. When the question information contains speech, ASR (Automatic Speech Recognition) is required to convert the speech into text. If the question information contains images, OCR (Optical Character Recognition) or image recognition is required to convert the images into text. If the question information contains video, keyframes need to be extracted from the video content, followed by image recognition to convert it into text. Structured processing of the question text involves extracting and arranging key information according to preset fields. These preset fields include, but are not limited to, message type (mixed text and image), text content, recognized entities, and original modalities. Intent recognition of structured question text aims to identify the consultation intent contained in the question information, including but not limited to product selection, returns, and after-sales consultation. Object recognition of structured question text involves identifying the consultation object contained or associated with the question information. Consultation objects include, but are not limited to, products and orders. Initial prompt information refers to the prompt template, i.e., the Prompt template. Processing logic information includes, but is not limited to, the knowledge retrieval scope, response generation strategy, and post-processing rules.

[0093] By deploying multiple specialized AI agents (intent analysis agent, knowledge retrieval agent, solution generation agent, quality assessment agent, and personalization adaptation agent) and orchestrating them through a coordinator, complex customer service tasks are handled through specialized division of labor. Compared to a single large model solution, the multi-agent array improves intent recognition accuracy by 15%-20% and significantly outperforms the direct output of a general-purpose model in terms of the professionalism and completeness of solution generation. Each agent can be independently iterated and upgraded, and local optimization does not affect the overall system stability, thus achieving an evolvable system architecture with "high cohesion and low coupling".

[0094] Step 612: Determine at least one knowledge base and object state data associated with the knowledge information, and query at least one knowledge base based on the intent information and object information to obtain intent knowledge fragments and object knowledge fragments.

[0095] Step 614: Update the prompt information to the target prompt information based on the object state data, intent knowledge fragments, and object knowledge fragments, and input the target prompt information into the information processing model to obtain at least one set of candidate answer information.

[0096] Once the knowledge information is determined, knowledge bases such as the product knowledge graph, industry knowledge base, and store-customized knowledge base can be established. When the question information corresponds to a return / exchange inquiry, return / exchange rules are retrieved from the product knowledge graph based on object information (product name, etc.), and relevant knowledge fragments are output. General refund processes are retrieved from the industry knowledge base based on intent information, and industry standard fragments are output. Furthermore, store-specific after-sales rules can be retrieved from the store-customized knowledge base based on the merchant ID, and store rule fragments are output. All these fragments are assembled into a unified Prompt (target prompt information), which is then input into a large language model to generate recommended responses—that is, at least one set of candidate answer information.

[0097] Step 616: Send at least one set of candidate answer information to the target user and receive feedback information submitted by the target user regarding at least one set of candidate answer information.

[0098] Step 618: Based on the feedback information, determine the target answer information corresponding to the question information from at least one set of candidate answer information, and send the target answer information to the user.

[0099] The target user is the person who provides the information to resolve the user's problem. In the case of a consumer on an e-commerce platform, the target user is the customer service representative. Feedback information can be the target user's choice from at least one set of candidate answers. The target answer information is either the set of candidate answers chosen by the user or at least one set of candidate answers determined from the feedback provided by the target user. Once the target answer information is determined, it can be sent to the user as feedback on their question.

[0100] In addition, customer service personnel can see two sets of candidate answer information, Option A and Option B, on the chat page. If the customer service personnel choose Option A as the target answer and need to adjust it, they can submit the adjustment information (e.g., changing to a vertical striped four-piece suit). The AI ​​digital human engine will then update "striped four-piece suit" in the target answer information to "vertical striped four-piece suit," obtaining the answer information to be sent. The AI ​​digital human engine can either directly send the answer information to the user, or it can send it to the customer service personnel, display it on the chat page, and then the customer service personnel will subsequently send it to the user.

[0101] When e-commerce platforms have customer service requests for new product pairing recommendations, a target information unit—a Product Pairing Recommendation Skill—can be constructed based on the customer service's recommendations. This skill unit includes product pairing recommendation knowledge, intent, and prompts. The product pairing recommendation information unit is stored in an information unit set, allowing subsequent user-submitted questions about product pairings to be matched with relevant skill units. This standardized encapsulation of Skills enables rapid replication and promotion of high-quality service capabilities across different merchants and scenarios, achieving large-scale accumulation and sharing of service capabilities. By abstracting service capabilities into standardized Skill units, "create once, reuse many times" service orchestration is achieved. Merchants, platform operators, and customer service managers can all create and configure Skills through a visual interface. When adding a new service scenario, only the corresponding Skill needs to be added, without modifying the underlying development pipeline. This mechanism shortens the launch cycle of new service scenarios from weeks in traditional development models to hours, significantly improving scalability and response speed. Meanwhile, Skill's standardized encapsulation enables high-quality service capabilities to be quickly replicated and promoted across different merchants and scenarios, achieving the large-scale accumulation and sharing of service capabilities.

[0102] This specification provides an embodiment of an information processing method that introduces a reusable orchestration mechanism for Skills (skill service units) as the core service organization form of the AI ​​digital human engine layer. This mechanism allows merchants, platform operators, and customer service administrators to independently create, configure, and manage Skill service units according to their respective needs, achieving modular encapsulation and flexible reuse of service capabilities. Skill Definition and Standardized Encapsulation: A Skill is an independently executable standardized service unit. Each Skill encapsulates complete processing logic for a specific scenario, including: intent matching rules (defining the intent type and triggering conditions of the Skill's response), knowledge retrieval scope (specifying the knowledge graph subgraphs, knowledge base partitions, and FAQ sets associated with the Skill), Prompt templates (predefining prompts, role settings, output format requirements, etc. for the scenario), response generation strategies (including parameters such as tone style, content emphasis, and length control), and post-processing rules (sensitive word filtering rules, compliance verification rules, etc.). Through a standardized Skill description protocol, the input and output interfaces of each Skill remain unified, ensuring that Skills can be combined and replaced with each other. Multi-role Skill Creation System: Provides skill creation capabilities for three types of roles: a) Platform-level Skills: Created and maintained by the e-commerce platform operations team, covering common industry scenarios such as logistics tracking, return and exchange processes, price protection, and other common after-sales scenarios, which can be directly referenced by all merchants; b) Merchant-level Skills: Created by merchants or their authorized customer service administrators, customized for specific store scenarios, such as consultation on the selection of specific products, explanations of exclusive store promotional activities, and brand-specific service processes.

[0103] Skill creation and management are entirely completed through a visual configuration interface, without requiring modification to the underlying code or development pipeline. When new service scenarios or requirements arise, relevant roles only need to create a new Skill in the Skill management platform, configure its intent matching rules, associated knowledge scope, and Prompt template to complete the deployment. The core advantages of this mechanism are: adding a Skill does not affect the operation of existing Skills, does not require redeployment, and does not require modification to the core processing flow of the AI ​​digital human engine; deployed Skills can be referenced and reused by other merchants or teams at any time, forming a Skill market ecosystem; the Skill version management mechanism supports A / B testing, allowing new versions of Skills to be run in parallel with old versions to compare effects before a full switch. After consumer messages undergo intent recognition, the Skill orchestration and scheduling engine is responsible for querying the list of matching candidate Skills from the Skill registry center, selecting and activating one or more Skills for parallel or serial execution based on intent classification results, confidence scores, and priority rules. When multiple Skills are executed in parallel, the scheduling engine is responsible for aggregating the output results of each Skill and generating a final recommended solution ranking list based on confidence weighting and deduplication strategies. The scheduling engine also supports cascading calls between Skills, meaning the output of one Skill can be used as the input of another, enabling multi-step processing in complex scenarios. It provides full lifecycle management capabilities for Skills, including: creation and configuration, testing and verification (providing a sandbox testing environment and simulated dialogue functionality), deployment and release, monitoring and statistics (real-time tracking of each Skill's trigger count, adoption rate, consumer satisfaction, and other metrics), optimization and iteration (continuously adjusting Skill configuration parameters based on performance data), and decommissioning and archiving.

[0104] It can also build an independent incremental personalized learning layer for each merchant's store, continuously learning from the merchant's unique customer service data to achieve deep adaptation of recommendation schemes from general to personalized. Based on the general language model, a set of lightweight personalized adaptation parameters is added to each merchant's store. These personalized parameters are independent of the general model and only affect the generation of recommendation schemes for that merchant. These include: language style parameters (controlling the wording habits and colloquialisms in responses) and content emphasis parameters (controlling the weighting of dimensions such as price, quality, and functionality). Incremental learning data comes from customer service adoption behavior, editing difference analysis, consumer satisfaction feedback, and merchant knowledge update signals. Personalized parameters can be fine-tuned in real time through online learning algorithms based on each customer service adoption / editing behavior, with update latency controlled at the minute level. The performance metrics of each merchant's personalized model can be continuously monitored, including recommendation scheme adoption rate, consumer satisfaction, and response quality score. When the performance of the personalized model is detected to be lower than that of the general model, a fallback mechanism is automatically triggered, temporarily switching back to the general model and reinitializing the personalized parameters. Simultaneously, the output of the personalized model still needs to pass through a unified quality assessment module and sensitive information filtering to ensure that personalization does not introduce non-compliant content.

[0105] This specification provides an embodiment of an information processing method that determines user-submitted question information and obtains knowledge information associated with the question information. Based on the question information, at least one information unit is determined, and prompt information is constructed based on the question information and the question processing information contained in the at least one information unit. Determining information units matching the question information allows for on-demand selection of information units. The prompt information and knowledge information are input into an information processing model to obtain at least one set of candidate answer information, which is then sent to the target user. The target user can select from the at least one set of candidate answer information according to the actual situation and submit feedback information based on the selection result. Upon receiving the feedback information submitted by the target user for the at least one set of candidate answer information, the target answer information corresponding to the question information can be determined from the at least one set of candidate answer information based on the feedback information. Determining the target answer information from the at least one set of candidate answer information based on the feedback information submitted by the target user assists the target user in answering the user-submitted question information, improving the efficiency and accuracy of the target user's answering of question information. It provides a self-service answering method for the target user, increasing the target user's willingness to answer user-submitted questions and improving the target user's response efficiency to question information.

[0106] Corresponding to the above method embodiments, this specification also provides information processing system embodiments. Figure 7 A schematic diagram of the structure of an information processing system according to one embodiment of this specification is shown. Figure 7As shown, the information processing system 700 includes a platform server 710, a user terminal 740, an information processing terminal 720, and a user server 730. The system includes: the platform server 710, which receives question information submitted by the user terminal 740 and sends the question information to the information processing terminal 720 and the user server 730; the information processing terminal 720, which acquires knowledge information associated with the question information; determines at least one information unit based on the question information and constructs prompt information based on the question information and the question processing information contained in the at least one information unit; inputs the prompt information and the knowledge information into an information processing model to obtain at least one set of candidate answer information and sends the at least one set of candidate answer information to the user server 730; and the user server 730, which determines the target answer information corresponding to the question information from the at least one set of candidate answer information and sends the target answer information to the user terminal 740.

[0107] An embodiment of this specification provides an information processing system including a platform server, a user terminal, an information processing terminal, and a user server. The platform server receives question information submitted by the user terminal and sends the question information to the information processing terminal and the user server. The information processing terminal determines the question information submitted by the user and obtains knowledge information associated with the question information. Based on the question information, it determines at least one information unit and constructs prompt information based on the question information and the question processing information contained in the at least one information unit. Determining the information unit matching the question information allows for on-demand selection of information units. The prompt information and knowledge information are input into an information processing model to obtain at least one set of candidate answer information, which is then sent to the user server. The target user on the user server can select from the at least one set of candidate answer information according to the actual situation and submit feedback information based on the selection result. Upon receiving the feedback information submitted by the target user on the user server for the at least one set of candidate answer information, the platform server can determine the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information and send the target answer information to the user terminal. Based on feedback submitted by target users, the system identifies the target answer from at least one set of candidate answers. This assists target users in answering their submitted questions, improving the efficiency and accuracy of their responses. It provides a self-service solution for target users, increasing their willingness to answer questions and improving their response efficiency.

[0108] In practical applications, information processing systems can correspond to e-commerce Q&A scenarios, such as... Figure 8As shown, the information processing system includes a consumer, an intelligent routing module, human customer service, an AI digital human engine, a knowledge and data layer, and a customer service workbench. The consumer corresponds to the user terminal, the human customer service corresponds to the user server, the AI ​​digital human engine corresponds to the information processing terminal, and the intelligent routing module corresponds to the platform server. When a consumer initiates a customer service inquiry, they provide the question information. The intelligent routing module routes the question information to the human customer service representative and simultaneously forwards it to the AI ​​digital human engine in real time. The AI ​​digital human engine performs multimodal message parsing, intent and state analysis, entity extraction, and inquiry target identification on the question information. It then submits a knowledge retrieval request to the knowledge and data layer, which returns matching knowledge and real-time data. The AI ​​digital human engine obtains the question information containing the above analysis results and constructs a prompt based on the question information. The prompt and the knowledge information retrieved from the knowledge and data layer are input into the information processing module (LLM) to generate at least one set of candidate answer information (solutions to the question). After evaluating the quality of the solutions recommended by the AI ​​digital human engine, the recommended solutions are pushed to the customer service workbench, along with the inquiry target information. The customer service workbench displays recommended solution cards to live agent representatives. The agent can then click "send" or edit and send the solution to the workbench. Based on the received solution information, the workbench sends a response message corresponding to the customer's question. The workbench records the live agent's actions, and after the session, the customer can rate their satisfaction with the service. This satisfaction rating serves as feedback data on the question-answering service, used to optimize the information processing model.

[0109] See Figure 9 , Figure 9 A flowchart of another information processing method according to an embodiment of this specification is shown. The information processing method is applied to an information processing terminal and specifically includes the following steps: Step 902: Receive the problem information sent by the platform server and obtain the knowledge information associated with the problem information; Step 904: Determine at least one information unit based on the problem information, and construct a prompt message based on the problem information and the problem handling information contained in the at least one information unit; Step 906: Input the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information, and send the at least one set of candidate answer information to the target user; Step 908: Receive feedback information submitted by the target user for the at least one set of candidate answer information, and determine the target answer information corresponding to the question information based on the feedback information in the at least one set of candidate answer information.

[0110] This specification provides an embodiment of an information processing method applied to an information processing terminal. It receives question information sent by a platform server and acquires knowledge information associated with the question information. Based on the question information, it determines at least one information unit and constructs prompt information based on the question information and the question processing information contained in the at least one information unit. Determining the information unit matching the question information allows for on-demand selection of information units. The prompt information and knowledge information are input into an information processing model to obtain at least one set of candidate answer information, which is then sent to a target user. The target user can select from the at least one set of candidate answer information according to the actual situation and submit feedback information based on the selection result. Upon receiving the feedback information submitted by the target user for the at least one set of candidate answer information, the target answer information corresponding to the question information can be determined from the at least one set of candidate answer information based on the feedback information. Determining the target answer information from the at least one set of candidate answer information based on the feedback information submitted by the target user assists the target user in answering the question information submitted by the user, improving the efficiency and accuracy of the target user's answering of the question information. It provides a self-service answering method for the target user, increasing the target user's willingness to answer the question information raised by the user and improving the target user's response efficiency to the question information.

[0111] See Figure 10 , Figure 10 A flowchart of a problem information processing method according to an embodiment of this specification is shown, which specifically includes the following steps: Step 1002: Determine the problem information submitted by the user and obtain the knowledge information associated with the problem information; Step 1004: Determine at least one information unit based on the problem information, and construct a prompt message based on the problem information and the problem processing information contained in the at least one information unit; Step 1006: Input the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information, and send the at least one set of candidate answer information to the customer service terminal, wherein the customer service terminal is used to generate candidate answer prompt cards based on the at least one set of candidate answer information, and push the candidate answer prompt cards to the customer service through a virtual customer service object; Step 1008: Receive feedback information submitted by the customer service representative to the candidate answer prompt card, determine the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information, and send the target answer information to the user.

[0112] This specification provides a problem information processing method according to one embodiment, which determines the problem information submitted by a user and obtains the knowledge information associated with the problem information. Based on the problem information, at least one information unit is determined, and prompt information is constructed based on the problem information and the problem processing information contained in the at least one information unit. Determining the information unit matching the problem information allows for on-demand selection of information units. The prompt information and knowledge information are input into an information processing model to obtain at least one set of candidate answer information, and this at least one set of candidate answer information is sent to a customer service terminal. The customer service terminal can generate candidate answer prompt cards based on the at least one set of candidate answer information and push the candidate answer prompt cards to customer service representatives through a virtual customer service object.

[0113] As a structured auxiliary card, the candidate answer hint card is pushed to the customer service representative on the Q&A page between the representative and the user by the AI ​​digital human corresponding to the virtual customer service representative. The representative can then view at least one set of candidate answer information in the hint card. The AI ​​digital human can assist the customer service representative in answering user questions in real time. After receiving feedback from the customer service representative on the candidate answer hint card, the AI ​​digital human can determine the target answer information corresponding to the question information from at least one set of candidate answer information based on the feedback information and send the target answer information to the user. Determining the target answer information from at least one set of candidate answer information based on the feedback information submitted by the customer service representative assists the customer service representative in answering user questions, improving the efficiency and accuracy of customer service in answering questions. It provides customer service representatives with a self-service way to answer questions, increases their willingness to answer user questions, and improves their response efficiency.

[0114] Corresponding to the above method embodiments, this specification also provides embodiments of an information processing apparatus. Figure 11 A schematic diagram of the structure of an information processing apparatus according to one embodiment of this specification is shown. Figure 11 As shown, the device includes: The determination module 1102 is configured to determine the problem information submitted by the user and obtain the knowledge information associated with the problem information; The construction module 1104 is configured to determine at least one information unit based on the problem information, and to construct prompt information based on the problem information and the problem processing information contained in the at least one information unit; The input module 1106 is configured to input the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information, and send the at least one set of candidate answer information to the target user; The receiving module 1108 is configured to receive feedback information submitted by the target user in response to the at least one set of candidate answer information, and determine the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information.

[0115] In an optional embodiment, constructing the prompt information based on the problem information and the problem handling information contained in the at least one information unit includes: The problem information is converted into problem text, and the problem text is then processed in a structured manner to obtain the structured problem text corresponding to the problem text. The structured question text is subjected to intent recognition to obtain intent information, and the structured question text is subjected to object recognition to obtain object information; Initial prompt information and processing logic information are determined from the problem processing information contained in the at least one information unit; The initial prompt information is updated based on the intent information, the object information, and the processing logic information to obtain the new prompt information.

[0116] In an optional embodiment, the step of performing intent recognition on the structured question text to obtain intent information includes: The structured question text is input into the intent recognition model to obtain the intent type and the intent score corresponding to the intent type; The user's question status information is determined based on the structured question text, and the intent type, intent score, and question status information are used as the intent information.

[0117] In an optional embodiment, the step of performing object recognition on the structured question text to obtain object information includes: Identify entity information in the structured question text; Based on the entity information, query the object database to obtain the object information associated with the entity information; After performing object recognition on the structured question text to obtain object information, the method further includes: Identify the service user corresponding to the user, and display the object information on the service user's service user page.

[0118] In an optional embodiment, inputting the prompt information and the knowledge information into an information processing model to obtain at least one set of candidate answer information includes: Determine at least one knowledge base and object state data associated with the knowledge information; Based on the intent information and the object information, query the at least one knowledge base to obtain intent knowledge fragments and object knowledge fragments; Based on the object state data, the intent knowledge fragment, and the object knowledge fragment, the prompt information is updated to the target prompt information, and the target prompt information is input into the information processing model to obtain the at least one set of candidate answer information.

[0119] In an optional embodiment, inputting the prompt information and the knowledge information into an information processing model to obtain at least one set of candidate answer information includes: The prompt information and the knowledge information are input into the information processing model to obtain at least one set of initial candidate answer information; Determine at least one answer detection dimension, and detect the at least one set of initial candidate answer information according to the at least one answer detection dimension to obtain the detection information corresponding to the at least one set of initial candidate answer information respectively; The at least one set of candidate answer information is determined based on the detection information corresponding to the at least one set of initial candidate answer information.

[0120] In an optional embodiment, after inputting the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information, the method further includes: If the at least one set of candidate answer information includes target candidate answer information that meets the sending conditions, an answer information component is generated based on the target candidate answer information. The answer information component is displayed on the user's Q&A page.

[0121] In an optional embodiment, after determining the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information, the method further includes: Receive adjustment information from the service user regarding the target answer information; The target answer information is updated based on the adjustment information to obtain the answer information to be sent, and the answer information to be sent is sent to the user.

[0122] In an optional embodiment, after sending the answer information to be sent to the user, the method further includes: Determine the user's evaluation information regarding the answer to be sent, and construct training samples based on the evaluation information, the answer to be sent, and the question information; The information processing model is fine-tuned based on the training samples.

[0123] In an optional embodiment, the method further includes: The unit knowledge, unit intent, and unit prompt information are determined based on the unit requirement information, and the target information unit is constructed based on the unit knowledge, the unit intent, and the unit prompt information. Determine the set of information units associated with the at least one information unit, and add the target information unit to the set of information units.

[0124] An embodiment of this specification provides an information processing apparatus that determines user-submitted question information and obtains knowledge information associated with the question information. Based on the question information, at least one information unit is determined, and prompt information is constructed based on the question information and the question processing information contained in the at least one information unit. Determining information units matching the question information allows for on-demand selection of information units. The prompt information and knowledge information are input into an information processing model to obtain at least one set of candidate answer information, which is then sent to a target user. The target user can select from the at least one set of candidate answer information according to the actual situation and submit feedback information based on the selection result. Upon receiving the feedback information submitted by the target user for the at least one set of candidate answer information, the target answer information corresponding to the question information can be determined from the at least one set of candidate answer information based on the feedback information. Determining the target answer information from the at least one set of candidate answer information based on the feedback information submitted by the target user assists the target user in answering the user-submitted question information, improving the efficiency and accuracy of the target user's answering of question information. It provides a self-service answering method for the target user, increasing the target user's willingness to answer user-submitted questions and improving the target user's response efficiency to question information.

[0125] The above is an illustrative scheme of an information processing device according to this embodiment. It should be noted that the technical solution of this information processing device and the technical solution of the information processing method described above belong to the same concept. For details not described in detail in the technical solution of the information processing device, please refer to the description of the technical solution of the information processing method described above.

[0126] Corresponding to the above method embodiments, this specification also provides embodiments of an information processing apparatus. Figure 12 A schematic diagram of another information processing apparatus provided in one embodiment of this specification is shown. Figure 12 As shown, the device includes: The receiving module 1202 is configured to receive problem information sent by the platform server and obtain knowledge information associated with the problem information; The construction module 1204 is configured to determine at least one information unit based on the problem information, and to construct prompt information based on the problem information and the problem handling information contained in the at least one information unit; The input module 1206 is configured to input the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information, and send the at least one set of candidate answer information to the target user; The determining module 1208 is configured to receive feedback information submitted by the target user for the at least one set of candidate answer information, and determine the target answer information corresponding to the question information based on the feedback information in the at least one set of candidate answer information.

[0127] This specification provides an embodiment of an information processing apparatus applied to an information processing terminal. It receives question information sent by a platform server and acquires knowledge information associated with the question information. Based on the question information, it determines at least one information unit and constructs prompt information based on the question information and the question processing information contained in the at least one information unit. Determining the information unit matching the question information allows for on-demand selection of information units. The prompt information and knowledge information are input into an information processing model to obtain at least one set of candidate answer information, which is then sent to a target user. The target user can select from the at least one set of candidate answer information according to the actual situation and submit feedback information based on the selection result. Upon receiving the feedback information submitted by the target user for the at least one set of candidate answer information, the target answer information corresponding to the question information can be determined from the at least one set of candidate answer information based on the feedback information. Determining the target answer information from the at least one set of candidate answer information based on the feedback information submitted by the target user assists the target user in answering the question information submitted by the user, improving the efficiency and accuracy of the target user's answering of the question information. It provides a self-service answering method for the target user, increasing the target user's willingness to answer the question information raised by the user and improving the target user's response efficiency to the question information.

[0128] The above is an illustrative scheme of another information processing device according to this embodiment. It should be noted that the technical solution of this information processing device and the technical solution of the other information processing method described above belong to the same concept. For details not described in detail in the technical solution of the information processing device, please refer to the description of the technical solution of the information processing method described above.

[0129] Corresponding to the above method embodiments, this specification also provides embodiments of a problem information processing device. Figure 13 A schematic diagram of a problem information processing device according to one embodiment of this specification is shown. Figure 13 As shown, the device includes: The acquisition module 1302 is configured to determine the question information submitted by the user and acquire the knowledge information associated with the question information; The construction module 1304 is configured to determine at least one information unit based on the problem information, and to construct prompt information based on the problem information and the problem handling information contained in the at least one information unit; The input module 1306 is configured to input the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information, and send the at least one set of candidate answer information to the customer service terminal, wherein the customer service terminal is used to generate candidate answer prompt cards based on the at least one set of candidate answer information, and push the candidate answer prompt cards to the customer service through a virtual customer service object; The receiving module 1308 is configured to receive feedback information submitted by the customer service representative to the candidate answer prompt card, determine the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information, and send the target answer information to the user.

[0130] This specification provides a problem information processing apparatus according to one embodiment, which determines the problem information submitted by a user and obtains knowledge information associated with the problem information. Based on the problem information, at least one information unit is determined, and prompt information is constructed based on the problem information and the problem processing information contained in the at least one information unit. By determining the information unit matching the problem information, on-demand selection of information units can be achieved. The prompt information and knowledge information are input into an information processing model to obtain at least one set of candidate answer information, and the at least one set of candidate answer information is sent to a customer service terminal. The customer service terminal can generate candidate answer prompt cards based on the at least one set of candidate answer information and push the candidate answer prompt cards to customer service representatives through a virtual customer service object.

[0131] As a structured auxiliary card, the candidate answer hint card is pushed to the customer service representative on the Q&A page between the representative and the user by the AI ​​digital human corresponding to the virtual customer service representative. The representative can then view at least one set of candidate answer information in the hint card. The AI ​​digital human can assist the customer service representative in answering user questions in real time. After receiving feedback from the customer service representative on the candidate answer hint card, the AI ​​digital human can determine the target answer information corresponding to the question information from at least one set of candidate answer information based on the feedback information and send the target answer information to the user. Determining the target answer information from at least one set of candidate answer information based on the feedback information submitted by the customer service representative assists the customer service representative in answering user questions, improving the efficiency and accuracy of customer service in answering questions. It provides customer service representatives with a self-service way to answer questions, increases their willingness to answer user questions, and improves their response efficiency.

[0132] The above is an illustrative scheme of a problem information processing device according to this embodiment. It should be noted that the technical solution of this problem information processing device and the technical solution of the problem information processing method described above belong to the same concept. For details not described in detail in the technical solution of the problem information processing device, please refer to the description of the technical solution of the problem information processing method described above.

[0133] Figure 14 A structural block diagram of a computing device 1400 according to one embodiment of this specification is shown. The components of the computing device 1400 include, but are not limited to, a memory 1410 and a processor 1420. The processor 1420 is connected to the memory 1410 via a bus 1430, and a database 1450 is used to store data.

[0134] The computing device 1400 also includes an access device 1440, which enables the computing device 1400 to communicate via one or more networks 1460. Examples of such networks include Public Switched Telephone Network (PSTN), Local Area Network (LAN), Wide Area Network (WAN), Personal Area Network (PAN), or combinations of communication networks such as the Internet. The access device 1440 may include one or more of any type of wired or wireless network interface (e.g., a network interface card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) wireless interface, a Wi-MAX (Worldwide Interoperability for Microwave Access) interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, or a Near Field Communication (NFC) interface.

[0135] In one embodiment of this specification, the above-described components of the computing device 1400 and Figure 14 Other components, not shown, can also be connected to each other, for example, via a bus. It should be understood that... Figure 14 The block diagram of the computing device shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0136] The computing device 1400 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or personal computers (PCs). The computing device 1400 can also be a mobile or stationary server.

[0137] The processor 1420 is configured to execute the following computer-executable instructions, which, when executed by the processor, implement the steps of the above-described information processing method.

[0138] The above is an illustrative scheme of a computing device according to this embodiment. It should be noted that the technical solution of this computing device and the technical solution of the information processing method described above belong to the same concept. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solution of the information processing method described above.

[0139] An embodiment of this specification also provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the above-described information processing method.

[0140] The above is an illustrative scheme of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium and the technical solution of the information processing method described above belong to the same concept. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solution of the information processing method described above.

[0141] An embodiment of this specification also provides a computer program product, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described information processing method.

[0142] The above is an illustrative scheme of a computer program product according to this embodiment. It should be noted that the technical solution of this computer program product and the technical solution of the information processing method described above belong to the same concept. For details not described in detail in the technical solution of the computer program product, please refer to the description of the technical solution of the information processing method described above.

[0143] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0144] The computer program / instructions include computer program code, which may be in the form of source code, object code, executable file, or certain intermediate forms. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium may be appropriately added or removed according to the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0145] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments in this specification are not limited to the described order of actions, because according to the embodiments in this specification, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the embodiments in this specification.

[0146] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0147] The preferred embodiments disclosed above are merely illustrative of this specification. Optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the embodiments described herein. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the embodiments, thereby enabling those skilled in the art to better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. An information processing method, comprising: Determine the problem information submitted by the user, and obtain the knowledge information associated with the problem information; Based on the problem information, at least one information unit is determined, and a prompt message is constructed based on the problem information and the problem handling information contained in the at least one information unit; The prompt information and the knowledge information are input into the information processing model to obtain at least one set of candidate answer information, and the at least one set of candidate answer information is sent to the target user; Receive feedback information submitted by the target user in response to the at least one set of candidate answer information, and determine the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information.

2. The information processing method according to claim 1, wherein constructing prompt information based on the problem information and the problem processing information contained in the at least one information unit includes: The problem information is converted into problem text, and the problem text is then processed in a structured manner to obtain the structured problem text corresponding to the problem text. The structured question text is subjected to intent recognition to obtain intent information, and the structured question text is subjected to object recognition to obtain object information; Initial prompt information and processing logic information are determined from the problem processing information contained in the at least one information unit; The initial prompt information is updated based on the intent information, the object information, and the processing logic information to obtain the new prompt information.

3. The information processing method according to claim 2, wherein performing intent recognition on the structured question text to obtain intent information includes: The structured question text is input into the intent recognition model to obtain the intent type and the intent score corresponding to the intent type; The user's question status information is determined based on the structured question text, and the intent type, intent score, and question status information are used as the intent information.

4. The information processing method according to claim 2, wherein the step of performing object recognition on the structured question text to obtain object information includes: Identify entity information in the structured question text; Based on the entity information, query the object database to obtain the object information associated with the entity information; After performing object recognition on the structured question text to obtain object information, the method further includes: Identify the service user corresponding to the user, and display the object information on the service user's service user page.

5. The information processing method according to claim 2, wherein inputting the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information includes: Determine at least one knowledge base and object state data associated with the knowledge information; Based on the intent information and the object information, query the at least one knowledge base to obtain intent knowledge fragments and object knowledge fragments; Based on the object state data, the intent knowledge fragment, and the object knowledge fragment, the prompt information is updated to the target prompt information, and the target prompt information is input into the information processing model to obtain the at least one set of candidate answer information.

6. The information processing method according to claim 1, wherein inputting the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information includes: The prompt information and the knowledge information are input into the information processing model to obtain at least one set of initial candidate answer information; Determine at least one answer detection dimension, and detect the at least one set of initial candidate answer information according to the at least one answer detection dimension to obtain the detection information corresponding to the at least one set of initial candidate answer information respectively; The at least one set of candidate answer information is determined based on the detection information corresponding to the at least one set of initial candidate answer information.

7. The information processing method according to claim 1, further comprising, after inputting the prompt information and the knowledge information into the information processing model to obtain at least one set of candidate answer information: If the at least one set of candidate answer information includes target candidate answer information that meets the sending conditions, an answer information component is generated based on the target candidate answer information. The answer information component is displayed on the user's Q&A page.

8. The information processing method according to claim 4, after determining the target answer information corresponding to the question information based on the feedback information from the at least one set of candidate answer information, further comprising: Receive adjustment information from the service user regarding the target answer information; The target answer information is updated based on the adjustment information to obtain the answer information to be sent, and the answer information to be sent is sent to the user.

9. The information processing method according to claim 8, further comprising, after sending the answer information to be sent to the user: Determine the user's evaluation information regarding the answer to be sent, and construct training samples based on the evaluation information, the answer to be sent, and the question information; The information processing model is fine-tuned based on the training samples.

10. The information processing method according to any one of claims 1-9, wherein the method further comprises: The unit knowledge, unit intent, and unit prompt information are determined based on the unit requirement information, and the target information unit is constructed based on the unit knowledge, the unit intent, and the unit prompt information. Determine the set of information units associated with the at least one information unit, and add the target information unit to the set of information units.

11. An information processing method, applied at an information processing terminal, comprising: Receive problem information sent by the platform server and obtain the knowledge information associated with the problem information; Based on the problem information, at least one information unit is determined, and a prompt message is constructed based on the problem information and the problem handling information contained in the at least one information unit; The prompt information and the knowledge information are input into the information processing model to obtain at least one set of candidate answer information, and the at least one set of candidate answer information is sent to the target user; Receive feedback information submitted by the target user for the at least one set of candidate answer information, and determine the target answer information corresponding to the question information based on the feedback information in the at least one set of candidate answer information.

12. A problem information processing method, comprising: Determine the problem information submitted by the user, and obtain the knowledge information associated with the problem information; Based on the problem information, at least one information unit is determined, and a prompt message is constructed based on the problem information and the problem handling information contained in the at least one information unit; The prompt information and the knowledge information are input into the information processing model to obtain at least one set of candidate answer information, and the at least one set of candidate answer information is sent to the customer service terminal. The customer service terminal is used to generate candidate answer prompt cards based on the at least one set of candidate answer information, and push the candidate answer prompt cards to the customer service through a virtual customer service object. The system receives feedback information submitted by the customer service representative to the candidate answer hint card, determines the target answer information corresponding to the question information from the at least one set of candidate answer information based on the feedback information, and sends the target answer information to the user.

13. An information processing system, comprising a platform server, a user terminal, an information processing terminal, and a user server, comprising: The platform server is used to receive the problem information submitted by the user terminal and send the problem information to the information processing terminal and the user server. The information processing terminal is used to obtain knowledge information associated with the problem information; Based on the problem information, at least one information unit is determined, and a prompt message is constructed based on the problem information and the problem handling information contained in the at least one information unit; The prompt information and the knowledge information are input into the information processing model to obtain at least one set of candidate answer information, and the at least one set of candidate answer information is sent to the user server. The user server is configured to determine the target answer information corresponding to the question information from the at least one set of candidate answer information, and send the target answer information to the user terminal.

14. A computing device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the method according to any one of claims 1 to 12.

15. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the method according to any one of claims 1 to 12.

16. A computer program product comprising a computer program or instructions which, when executed by a processor, implement the steps of the method according to any one of claims 1 to 12.

Images