Information processing method, electronic device, and readable storage medium

Abstract

The present disclosure provides an information processing method, an electronic device and a readable storage medium. The information processing method of the present disclosure comprises: in response to receiving inquiry information, determining a target user label contained in the inquiry information, the target user label representing the subject characteristics of a user initiating the inquiry information; in a business knowledge graph, determining the business domain knowledge applicable to the target user label, the business knowledge graph recording the association relationship between the user label and the business domain knowledge, and the association priority between the user label and the business domain knowledge having the association relationship; and generating response information applicable to the inquiry information based on the business domain knowledge having an association priority higher than a priority threshold with the target user label.

Description

Technical Field

[0001] This disclosure relates to the technical fields of data processing, and in particular to an information processing method, an electronic device, and a readable storage medium. Background Technology

[0002] In private domain business scenarios, enterprises rely heavily on accumulated customer data, business rules, and historical interaction habits to provide personalized, highly sticky closed-loop services. This places stringent requirements on intelligent agents, including strong knowledge dependence and multi-turn contextual coherence.

[0003] However, current mainstream solutions directly apply general-purpose large language models, but because their training corpus is limited to general data and does not closely match the enterprise's private domain knowledge, responses deviate from actual business. At the same time, the lack of effective modeling of customer tags and historical dialogues makes it difficult to maintain logical consistency in multi-turn interactions. Summary of the Invention

[0004] This disclosure provides an information processing method, an electronic device, and a readable storage medium.

[0005] According to one aspect of this disclosure, an information processing method is provided, comprising: in response to receiving query information, determining a target user tag contained in the query information, the target user tag representing the subject characteristics of the user initiating the query information; determining business domain knowledge applicable to the target user tag in a business knowledge graph, the business knowledge graph recording the association relationship between user tags and business domain knowledge, and the association priority between user tags and business domain knowledge with association relationships; and generating response information applicable to the query information based on business domain knowledge whose association priority with the target user tag is higher than a priority threshold.

[0006] In some implementations, the process prior to receiving the query information includes: constructing the business knowledge graph based on business data of the target business domain of interest in the query information; the business knowledge graph also records the relationships between multiple business domain knowledge and the association priority between two related business domain knowledge.

[0007] In some implementations, the business knowledge graph is constructed based on business data of the target business domain that the query information focuses on. This includes: using the main features that can be supported by the business rules in the business data as user tags, and using the business functions that can be supported by the business rules, the multiple process nodes in the business rules that implement the business functions, and the response templates in the business data that are applicable to the business functions as the business domain knowledge; and determining the association between the user tags and the business domain knowledge, and the association between multiple pieces of business domain knowledge, based on the business rules, to form the business knowledge graph.

[0008] In some implementations, constructing the business knowledge graph includes: when the business domain knowledge is a process node that implements a business function, determining the dependency relationship between the process node and other process nodes in the business process according to the business process recorded in the business rules, wherein the business process includes multiple process nodes that implement the business function and the execution order among the multiple process nodes.

[0009] In some implementations, after constructing the business knowledge graph, the process includes: updating the business knowledge graph using the new business data in response to the generation of new business data, and configuring new identifiers for business domain knowledge generated based on the new business data, wherein the new identifiers have a display period and disappear when the display period expires.

[0010] In some implementations, after constructing the business knowledge graph, the method further includes: generating a simulated information set based on the business knowledge graph, wherein the simulated information set is used to verify the response information obtained based on the business knowledge graph, the simulated information set is a simulated information set focusing on a single business function or a simulated information set focusing on multiple business functions, the simulated information set is multiple simulated query information with dialogue timing, and at least one background interference information is present in the multiple simulated query information.

[0011] In some implementations, generating a simulated information set based on the business knowledge graph includes: constructing target prompts for generating the simulated information set, wherein the target prompts include a business function attention quantity instruction, a background interference information generation instruction, and a target expression method; and, based on the business knowledge graph and the target prompts, calling an inquiry information generation model to generate a simulated information set containing the background interference information in the target expression method.

[0012] In some implementations, after generating a simulated information set containing the background interference information in the manner described in the target representation, the method includes: verifying the validity of the simulated information set.

[0013] In some implementations, verifying the validity of the simulated information set includes: determining whether the simulated information set contains business domain knowledge in the business knowledge graph; when the simulated information set contains the business domain knowledge, determining whether the simulated information set contains business domain knowledge with a new addition identifier in the business knowledge graph; when the simulated information set contains business domain knowledge with a new addition identifier, determining whether each simulated query in the simulated information set is a target expression; and when each simulated query is a target expression, determining that the simulated information set is valid.

[0014] In some implementations, after generating a simulated information set for verifying the business knowledge graph, the process includes: generating response information suitable for the simulated query information based on the business knowledge graph; and verifying the response information of the simulated query information, and generating alarm information about the response information of the simulated query information when the verification fails.

[0015] In some implementations, after generating alarm information regarding the response information of the simulated query information, at least one of the following is included: adjusting the business knowledge graph based on the alarm information; adjusting the first period corresponding to the interaction background information of the simulated query information based on the alarm information, wherein the interaction background information is historical query information in the first period.

[0016] In some implementations, the business knowledge graph is adjusted based on the alarm information, including: when the knowledge coverage of the response information to the simulated query is less than a coverage threshold, increasing the association priority between the target user tag corresponding to the simulated query and the business domain knowledge associated with the target user tag, and supplementing the attribute information of the business domain knowledge associated with the target user tag, wherein the attribute information is used to describe the business domain knowledge; when the process node of the response information to the simulated query is incorrect, correcting the dependency relationship of the business domain knowledge corresponding to the erroneous process node; and when the response information to the simulated query does not contain business domain knowledge with a new identifier, monitoring for the existence of new business data, and updating the business knowledge graph when the new business data exists.

[0017] In some implementations, the first period corresponding to the interactive background information of the simulated inquiry information is adjusted based on the alarm information, including: when the background correlation rate of the alarm information as the response information of the simulated inquiry information is less than the correlation rate threshold, the time span of the first period is increased.

[0018] In some implementations, in response to receiving the query information, the method further includes: determining the interaction background information of the query information, and using the main characteristics of the user contained in the interaction background information as the target user tag, wherein the interaction background information includes historical query content submitted by the user in a first period.

[0019] According to another aspect of this disclosure, an electronic device is provided, comprising: a memory storing execution instructions; and a processor executing the execution instructions stored in the memory, causing the processor to perform an information processing method according to any embodiment of this disclosure.

[0020] According to another aspect of this disclosure, a readable storage medium is provided that stores executable instructions, which, when executed by a processor, are used to implement an information processing method according to any embodiment of this disclosure.

[0021] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements an information processing method according to any embodiment of this disclosure. Attached Figure Description

[0022] The accompanying drawings illustrate exemplary embodiments of the present disclosure and, together with the description thereof, serve to explain the principles of the present disclosure. These drawings are included to provide a further understanding of the present disclosure and are incorporated in and constitute a part of this specification.

[0023] Figure 1 This is a schematic diagram illustrating an application scenario of the information processing method according to the embodiments of this disclosure.

[0024] Figure 2 This is a flowchart of an information processing method according to an embodiment of the present disclosure.

[0025] Figure 3 This is a flowchart illustrating the construction process of a business knowledge graph according to an embodiment of this disclosure.

[0026] Figure 4 This is a flowchart illustrating the construction process of another business knowledge graph according to an embodiment of this disclosure.

[0027] Figure 5 This is a flowchart illustrating the process of constructing a simulation information set according to an embodiment of the present disclosure.

[0028] Figure 6 This is a flowchart of the process for analyzing response information according to an embodiment of the present disclosure.

[0029] Figure 7This is a flowchart of another information processing method for determining response information according to an embodiment of the present disclosure.

[0030] Figure 8 This is a schematic block diagram of the structure of an information processing apparatus according to an embodiment of the present disclosure.

[0031] Figure 9 This is a schematic block diagram of an electronic device according to one embodiment of the present disclosure. Detailed Implementation

[0032] The present disclosure will now be described in further detail with reference to the accompanying drawings and examples. It should be understood that the specific examples described herein are for illustrative purposes only and are not intended to limit the scope of the disclosure. Furthermore, it should be noted that, for ease of description, only the parts relevant to the present disclosure are shown in the accompanying drawings.

[0033] It should be noted that, where there is no conflict, the embodiments and features described in this disclosure can be combined with each other. The technical solutions of this disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0034] Against the backdrop of rapid evolution in artificial intelligence technology, large language models have demonstrated strong application potential in various industries such as finance, healthcare, education, and real estate. More and more enterprises hope to introduce them into private domain business scenarios, using intelligent agents to automate customer service and intelligentize business processes, thereby improving operational efficiency and user experience. The core of private domain business lies in its highly personalized, strong customer stickiness, and closed-loop service characteristics, relying on the enterprise's long-term accumulated customer data and exclusive business rules (such as membership level benefits and targeted discount strategies).

[0035] However, the current mainstream approach is to directly deploy general-purpose large language models. The training corpora for these models originate from publicly available online data, lacking coverage of private business domain knowledge. This often leads to responses deviating from actual business needs and easily "forgetting" key information or logical contradictions during multi-turn interactions. Furthermore, current evaluation systems for model output responses still rely on general metrics such as fluency and semantic similarity, failing to accurately reflect business usability. These factors make it difficult for large models to meet enterprises' demands for high-precision, adaptive intelligent services.

[0036] Therefore, this disclosure proposes an information processing method.

[0037] Figure 1 This is a schematic diagram illustrating an application scenario of the information processing method according to embodiments of this disclosure. For example... Figure 1As shown, in this application scenario, a server 100 and a terminal device 200 may be included. The server 100 and the terminal device 200 can connect via a network or Bluetooth to exchange data. The server 100 can be a cloud server or a physical server, and the terminal device 200 can be a smart device such as a computer, mobile phone, or tablet. The server 100 provides the basic data required to run the information processing method, and the terminal device 200 executes the information processing method of this disclosure based on the basic data provided by the server 100.

[0038] Figure 2 This is a flowchart of an information processing method according to an embodiment of this disclosure. For example... Figure 2 As shown, this disclosure provides an information processing method M200, which proposes a business knowledge graph that records the association between user tags and business domain knowledge. By utilizing this graph, it is possible to generate response information that not only meets the user's inquiry needs but also closely aligns with relevant business domain content and language style, thereby improving the accuracy and professionalism of the responses.

[0039] Step S210: In response to receiving the query information, determine the target user tag contained in the query information.

[0040] Inquiry refers to a dialogue request initiated by a user during an interaction. It contains the user's specific needs, intentions, or questions, such as "What benefits can I enjoy as a Black Card member?" or "Can I use the newly issued coupons for the baby formula I bought last time?" This information not only carries literal meaning but also implies the user's subjective characteristics.

[0041] Target user tags are used to characterize the primary features of the user initiating the inquiry, such as identity, behavior, or preferences. These tags can be generated based on user-permitted private information and may include, for example, "user type XX" or "member level XX." Target user tags are crucial for accurate understanding of user needs and effective business responses. It's important to note that user-permitted private information is information actively provided or authorized for collection by the user, and this information, once acquired, is used solely to determine target user tags, undergoing necessary privacy processing when used internally.

[0042] It's important to note that each user's tag is not singular and fixed, but rather multi-tagged and dynamically evolving. On one hand, a user can simultaneously possess multiple target user tags, such as being both a "user of type XX" and a "member of level XX." On the other hand, user tags are continuously updated based on user behavior, business rules, or the passage of time. This multi-dimensional, real-time tagging system ensures contextual understanding and decision-making based on the latest and most comprehensive user profile, generating responses that better meet user needs, thereby providing highly personalized services tailored to current requirements.

[0043] In this implementation plan, the subject characteristics represented by the query information are used as target user tags, ensuring that the interaction process closely aligns with business domain knowledge. Furthermore, target user tags can be comprehensively determined by combining historical interaction data, account information, and other interaction background information, ensuring that the response content highly fits the business context while further enhancing its relevance to the surrounding context. Other methods for obtaining and determining target user tags will be explained in detail later and will not be elaborated upon here.

[0044] Step S220: In the business knowledge graph, determine the business domain knowledge applicable to the target user tags.

[0045] The business knowledge graph records the relationships between user tags and business domain knowledge, and further defines the priority of association between related user tags and business domain knowledge. By structurally modeling the relationships between user tags and business domain knowledge, the business knowledge graph enables a more accurate understanding of the user's identity and intent after obtaining query information, dynamically matching the most relevant business knowledge, and significantly improving the accuracy and personalization of responses.

[0046] Association priority characterizes the business relevance between user tags and business domain knowledge with association relationships, and serves as the basis for determining the content generation of response information. When there are multiple pieces of business domain knowledge that are associated with the target user tags, the business domain knowledge with higher association priority can be selected as the basis for generating response information, thereby ensuring the degree of relevance between the response information and the user's main characteristics.

[0047] Business domain knowledge can be extracted from the business data accumulated by an enterprise. Specifically, it includes: business functions supported by business rules (such as "discounts for membership levels" and "free shipping on first order for new customers"), process nodes required to implement these functions (such as orderly operation steps such as "submitting an application, reviewing, and verifying"), and response templates that are suitable for specific business scenarios and are embedded in the business data (such as compliant wording generated for "how to return goods"). These elements together constitute structured business domain knowledge, providing a foundation for generating accurate, consistent, and business-context-appropriate response information.

[0048] Step S230: Based on business domain knowledge that has an association priority higher than the priority threshold with the target user's tag, generate response information suitable for querying information.

[0049] Priority threshold is a key parameter used to determine whether the business relevance between business domain knowledge and target user tags meets the usability standard. When the association priority between business domain knowledge and target user tags is higher than the priority threshold, the business domain knowledge can be used as the basis for generating response information; if it is lower than the priority threshold, it is considered weakly relevant or inapplicable and can be filtered.

[0050] Furthermore, priority thresholds can be flexibly set based on business objectives, scenario sensitivity, or differences in user groups—for example, higher thresholds can be used in high-value customer service scenarios to ensure recommendation accuracy, while marketing and new user acquisition scenarios can be appropriately relaxed. At the same time, priority thresholds support dynamic adjustment, allowing for real-time optimization based on online performance feedback (such as click-through rate and conversion rate), thereby ensuring the quality of response information while maintaining business flexibility and adaptability.

[0051] The response information is the result of responding to user inquiries. Its content is generated based on business domain knowledge matched with target user tags, aiming to accurately meet the user's personalized needs. The response information is not limited to natural language text, such as "As a member of level XX, you can use a 20 yuan coupon"; it can also be presented in interactive media formats, such as jump links (pointing to a dedicated activity page), operation buttons (such as "one-click return application"), card-style service entry points, or embedded operation interfaces (such as order forms, benefit verification panels), etc. This multimodal output method not only meets the business closed-loop requirements of private domain scenarios but also improves user operation efficiency and experience continuity.

[0052] The complete information processing process is described below.

[0053] In some implementations, during the process of generating response information in response to user queries, it is necessary to construct a business knowledge graph based on business data of the target business domain that the query information focuses on. This business knowledge graph also records the relationships between multiple business domains, as well as the association priority between knowledge from two related business domains.

[0054] It should be noted that the target business domain refers broadly to various vertical business scenarios that the enterprise actually conducts and supports, including but not limited to real estate, finance, e-commerce, education, and healthcare. Each business domain has its own unique user tags, business rules, business processes, and response templates. For example, the real estate domain focuses on regional preferences and home purchase qualifications, while e-commerce focuses on membership benefits and promotional strategies. This solution does not presuppose a specific business domain but provides a general framework that can flexibly adapt to the knowledge structure and interaction needs of different target business domains, ensuring the universality and scalability of the method.

[0055] Figure 3 This is a flowchart illustrating the construction process of a business knowledge graph according to an embodiment of this disclosure. The following is in conjunction with... Figure 3 The process of constructing the business knowledge graph is explained.

[0056] In some implementations, a business knowledge graph is formed through steps 301 to 307. That is, the main features supported by business rules in the business data are used as user tags, and the business functions supported by the business rules, the multiple process nodes in the business rules that implement the business functions, and the response templates in the business data that are applicable to the business functions are used as business domain knowledge. Based on the business rules, the association between user tags and business domain knowledge, and the association between multiple business domain knowledge are determined to form a business knowledge graph.

[0057] Business data refers to structured or semi-structured information generated and accumulated by enterprises in their daily operations that is directly related to specific business activities and is used to support business services. Business data includes business rules (such as supported business functions and processes) and response templates, and is the core foundation for building a private knowledge system and accurately responding to information.

[0058] Business rules are the executable logic defined and followed by an enterprise in the process of providing business services. Business rules mainly include two parts: business functions and business processes. Business functions refer to the specific service capabilities that an enterprise provides to users, including business products (such as properties for sale, financial products, loan processing services, property appraisal services, etc.) and their details (such as scope of application, validity period, and usage thresholds), and business benefits (such as coupons, membership level benefits, etc.) and their details (such as scope of application, validity period, and usage thresholds). Business processes are an ordered set of multiple process nodes designed to achieve a specific business function, such as the series of steps "application → review → issuance → cancellation," used to standardize the operational sequence, ensure service compliance, and support the contextual coherence in multiple rounds of interaction.

[0059] User tags are structured abstractions of a user's identity, behavior, preferences, or status in a specific business scenario, typically generated based on private domain information authorized by the user. For example, "user of type XX" or "member of level XX" are typical user tags, serving as the core basis for personalized identification, accurate matching of business rules, and generation of context-relevant response information.

[0060] Response templates are standard reply structures that enterprises pre-set in private domain business scenarios to respond to user inquiries. They are usually derived from historical high-frequency inquiries and standard customer service scripts. Response templates include fixed scripts and dynamic placeholders. Dynamic placeholders can be, for example, "user name" or "available coupons." They can be automatically filled after matching user tags and business domain knowledge to generate personalized, compliant, and business-friendly response information. They are widely used in interactive processes such as answering inquiries, informing users of rights and benefits, and guiding users through processes.

[0061] Furthermore, response templates can be in not only text format but also interactive media format. To enhance user experience and service efficiency, response templates can also be multimodal and operable media structures, such as: button-style interactions, like quick access points for "Claim Now" and "Request a Refund"; card-style messages, integrating images, titles, descriptions, and action buttons, such as property recommendation cards and coupon cards; redirect links or mini-programs to guide users to a dedicated page, such as a "Click to view your personalized loan plan" link; forms, such as directly filling in information like scheduling a property viewing or submitting after-sales applications within the conversation; audio or video clips for complex explanations, such as property guides or financial product explanation videos; and dynamic components, such as calendar selectors or sliders for selecting loan terms and budget ranges. Response templates not only convey information but also directly facilitate the closed loop of business actions, serving as a key vehicle for realizing "service as interaction."

[0062] The business functions, process nodes, and response templates described above all belong to business domain knowledge. In the business knowledge graph, business domain knowledge and user tags together constitute the entity (node) part of the graph, while the relationships between business domain knowledge and the relationships between business domain knowledge and user tags together constitute the edge (relationship) part of the graph, thus forming a structured, reasonable semantic network that supports the intelligent system's understanding and response to private domain business logic.

[0063] User tags are associated with business domain knowledge applicable to those user tags, enabling personalized service matching. For example, the "XX type user" tag and the "dedicated customer service channel" function are associated with an "applicable" relationship; the membership level tag and "coupon a" are associated with a "triggered when membership level is XX" relationship. However, not all business domain knowledge is directly linked to user tags. Many pieces of business domain knowledge are associated with other business domain knowledge, and are only indirectly associated when the business function corresponding to the user touches upon that business domain knowledge. For example, the response template for "10% discount" is associated with the "10% discount" function, but not with "XX level member".

[0064] On the other hand, there are rich relationships among business domain knowledge, reflecting the enterprise's business logic. For example, a business function (such as "exclusive member discounts") is typically implemented collaboratively by multiple process nodes (such as "qualification verification," "coupon issuance," and "verification") in a specific order; and for this function or specific process node, corresponding response templates are configured (such as an interactive media "coupon issuance component" or a text message "You have successfully received your exclusive coupon"). Furthermore, different business functions may have mutually exclusive, overlapping, or priority relationships, and there may also be nested conditions or trigger dependencies between them. For example, there is a constraint relationship of "first order only" between "new user coupons" and the "verification process node." These relationships among business domain knowledge do not depend on user tags but are determined by the business logic itself, forming a network of relationships in the business knowledge graph that is independent of the user side.

[0065] Figure 4 This is a flowchart illustrating the construction process of another business knowledge graph according to an embodiment of this disclosure. The following is in conjunction with... Figure 4 This section explains another aspect of the construction process for business knowledge graphs.

[0066] Specifically, as shown in steps 401 to 404, user tags and business domain knowledge are determined, and it is determined whether the business domain knowledge is a process node. When the business domain knowledge is a process node that implements a business function, the dependency relationship between the process node and other process nodes in the business process is determined according to the business process recorded in the business rules. The business process includes multiple process nodes that implement the business function and the execution order between the multiple process nodes. Finally, a business knowledge graph containing the aforementioned dependency relationships is formed.

[0067] A process node is a basic execution unit in a business process, representing an independent step in a specific operation or decision. Each process node has clear business semantics and functions. For example, in a real estate transaction scenario, "qualification review," "contract signing," "loan approval," and "property transfer" are typical process nodes. These process nodes are sequentially linked according to the execution order defined by the business process, together forming an end-to-end business service loop. In intelligent interaction systems, process nodes not only standardize the logical sequence of service progression but also guide the generation of response information, thereby effectively supporting the contextual coherence, operational traceability, and service accuracy in multi-turn dialogues.

[0068] Multiple process nodes belonging to the same business process have dependencies on each other, which determine the execution order of these process nodes. Dependency is a special type of association with mandatory, directional, and execution constraints, emphasizing that the establishment or execution of one process node requires the completion or fulfillment of another process node as a prerequisite. Ordinary associations, on the other hand, usually only indicate a semantic connection between two entities (such as "applicable," "belongs to," or "describes"), without implying an execution order or necessary conditions. For example, "refund processing" depending on "warehouse confirmation of receipt" means that the former cannot be executed if the latter is not completed, demonstrating strong constraint; while the association between "high-net-worth clients" and "dedicated customer service" only indicates a matching relationship, without forcibly triggering or blocking other operations. Therefore, dependencies carry logical rigidity in business processes, while ordinary associations are more often used for knowledge organization and recommendation matching.

[0069] It's important to note that within the business knowledge graph, both user tags and business domain knowledge are configured with attribute information for refined description and dynamic control of the corresponding entities. For example, the business domain knowledge "dedicated customer service channel" can include attributes such as service hours = 9:00-22:00, response time ≤ 30 seconds, and applicable channel being a mini-program. These attributes not only enrich the semantic expression of entities but also provide crucial basis for matching business domain knowledge, upgrading the business knowledge graph from static associations to a quantifiable, configurable, and executable business intelligence engine.

[0070] In some implementations, in response to the generation of new business data, the business knowledge graph is updated using the new business data, and new identifiers are configured for the business domain knowledge generated based on the new business data. The new identifiers have a display period and disappear when the display period expires.

[0071] Specifically, when new business data is generated (such as launching new products, releasing new activities, or adjusting rules), the business knowledge graph is updated accordingly, and a new identifier is configured for the newly added business domain knowledge (such as new coupons or new service processes). This new identifier is a temporary metadata tag used to mark recently added knowledge nodes in the graph so that they can be given priority exposure or highlighted in response generation or recommendation strategies (e.g., prompting users "You have new exclusive benefits"). The display period refers to the effective time window of the new identifier (such as 7 days or 30 days). Once the set time limit is exceeded, the identifier automatically expires and is removed or hidden from the business knowledge graph to avoid long-term interference with the regular matching logic and ensure the timeliness of the graph content and the simplicity of the user experience.

[0072] In some implementations, a scheduled task mechanism can be configured to automatically trigger the update process of the business knowledge graph periodically (e.g., every morning, every Monday, or before major promotional events). This task scans newly added or changed business data (such as adjustments to membership rules, launch of new activities, and optimization of response templates) and dynamically synchronizes it to the knowledge graph, including operations such as adding entities, updating relationships, adjusting priorities, or adding "new identifiers." By triggering the task on a scheduled basis, the knowledge graph content is ensured to remain consistent with the company's latest business strategies without manual intervention, achieving continuous self-maintenance and agile response of the knowledge system.

[0073] In some implementations, a simulated information set is generated based on a business knowledge graph. This simulated information set is used to verify the response information obtained based on the business knowledge graph and can serve multiple business scenarios: simulated information sets focusing on a single business function and simulated information sets focusing on multiple business functions. The simulated information set consists of multiple simulated query messages with a dialogue sequence, and at least one background interference message is included in the multiple simulated query messages.

[0074] By constructing a simulated information set, problems in the business knowledge graph can be discovered by verifying the response information. These problems include missing business domain knowledge, unreasonable priority configuration of associations between related entities, and conflicts between dependencies between process nodes and business processes. The reliability and iteration efficiency of the business knowledge graph can be optimized without relying on real user interaction.

[0075] In some implementations, target prompts are constructed to generate a simulated information set. The target prompts include instructions on the number of business functions to focus on, instructions on generating background interference information, and a target expression method. Based on the business knowledge graph and the target prompts, an inquiry information generation model is invoked to generate a simulated information set containing background interference information in the target expression method.

[0076] Target prompts guide the query information generation model to output simulated information sets that meet business evaluation requirements. The content of target prompts includes not only basic elements, such as a description of the simulated information set generation task and learning examples, but also instructions on the amount of business functionality to focus on, instructions on generating background interference information, and the target expression method. Target prompts drive the query information generation model to generate semantically diverse, clearly intentd, and realistically user-behavior-oriented query information with evaluation relevance, providing reliable input for subsequent generation of response information based on business knowledge graphs.

[0077] The number of business functions focused on is a key differentiator in dialogue scenarios, used to control the complexity of the interaction process. For example, a dialogue can focus on a single business function (such as revolving entirely around "member coupon usage rules"), or it can involve multiple business functions (such as first inquiring about "product A's return and exchange policy," and then asking "whether product B supports installment payments"). These business functions may be independent and logically unrelated, but they appear together in a single user interaction, more closely resembling the natural behavior of users across needs and multiple intentions in real-world scenarios. By adjusting the number of business functions focused on, diverse simulated information sets ranging from simple to complex can be generated, effectively evaluating the rationality and feasibility of the business knowledge graph.

[0078] Background interference refers to content introduced into a query that requires consideration of the interaction's context (i.e., historical context) to simulate information redundancy or ambiguity in real conversations. For example, a user might ask in a new question, "Can the coupon mentioned earlier be used here?"—this statement itself doesn't specify the coupon type, applicable conditions, or business scenario, requiring a comprehensive understanding of previously issued coupons, the user's identity, and the current inquiry's business function for an accurate response. This type of interference is not noise, but rather a consideration of the feasibility of the time span of the interaction's contextual information taken into account when generating the response.

[0079] The target expression style refers to the generated query information, which should closely resemble the natural interaction patterns of real users or business domains in terms of language style and expression habits. For example, in a real estate scenario, a conversational, referential question like "Can this house be used for a housing provident fund loan?" is used, rather than a mechanical "Please check if this property supports housing provident fund loans." In financial customer service, this might manifest as hesitant, tentative, or emotional wording, such as "I don't think I received the cashback, did I miss it?" By mimicking users' expression habits or the interaction habits of the current business domain—including sentence structure, terminology usage, tone, and contextual dependence—the target expression style makes simulated queries more realistic and diverse, effectively verifying the feasibility of business knowledge graphs in non-standardized language environments.

[0080] The query information generation model can be a large language model, capable of automatically synthesizing simulated information sets based on preset target prompts. The simulated information set is a set of well-structured, semantically diverse, and context-coherent simulated query messages. Each simulated query message represents one or more rounds of questioning that a user might ask in real-world interactions. The simulated information set not only covers simple inquiries with a single intent but can also include complex dialogue flows that cross business functions, include background noise, and involve multiple progressive rounds (e.g., first asking "Are there any discounts for mothers and babies?", then following up with "Can previously issued coupons be combined?"). This comprehensively simulates actual user behavior, providing high-quality, high-coverage test input for subsequent response generation, consistency verification, and system evaluation based on business knowledge graphs.

[0081] Figure 5 This is a flowchart illustrating the process of constructing a simulation information set according to an embodiment of this disclosure. The following is in conjunction with... Figure 5 The process of constructing the simulated information set is explained.

[0082] In some implementations, steps 501 to 506 are proposed: constructing target prompts for generating a simulated information set, wherein the target prompts include the attention quantity instruction for business functions, the generation instruction for background interference information, and the target expression method; based on the business knowledge graph and the target prompts, calling the query information generation model to generate a simulated information set containing background interference information in the target expression method. Furthermore, the validity of the simulated information set is verified, namely: determining whether the simulated information set contains business domain knowledge in the business knowledge graph; when the simulated information set contains business domain knowledge, determining whether the simulated information set contains business domain knowledge with a new addition identifier in the business knowledge graph; when the simulated information set contains business domain knowledge with a new addition identifier, determining whether each simulated query message in the simulated information set is the target expression method; and when each simulated query message is the target expression method, determining that the simulated information set is valid.

[0083] When the simulated information set does not contain business domain knowledge, it indicates a low degree of fit with the business knowledge graph, making it difficult to effectively drive or verify the response generation logic based on the business knowledge graph, thus constituting an invalid test sample. Similarly, if the simulated information set lacks business domain knowledge marked with new additions (such as newly launched benefits or processes), it is impossible to examine the model's ability to perceive, invoke, and integrate new knowledge, and it also lacks value for iterative optimization. Furthermore, if the simulated information does not adopt the target expression method (i.e., it does not imitate the language habits of real users or the interaction habits of the business domain), although it may obtain a "correct" response in form, it cannot truly reflect the robustness and practicality in complex, ambiguous, or colloquial interaction scenarios, leading to overly optimistic and distorted evaluation results.

[0084] If the simulated information set is invalid, it can be discarded, and the target prompt words can be updated based on the reason for invalidity, so as to call the query information generation model to regenerate the simulated information set.

[0085] To ensure the quality and effectiveness of the simulated information set, a manual sampling verification mechanism can be introduced: samples are randomly selected from the automatically generated simulated information set according to a certain proportion, and the relevant business personnel review the content. This will not be elaborated on here.

[0086] Figure 6 This is a flowchart illustrating the process of analyzing response information according to an embodiment of this disclosure. The following is in conjunction with… Figure 6 It details the process of analyzing response information.

[0087] In step 601, response information suitable for simulated query information is generated based on the business knowledge graph; in step 602, the response information of simulated query information is verified to determine its validity; in step 603, it is determined whether the validity is valid. If the validity of the response information is valid, no processing is performed; otherwise, step 604 is executed to generate alarm information.

[0088] It should be noted that the process of generating response information for simulated queries is as follows: In response to receiving simulated queries, the target user tags contained in the simulated queries are determined; furthermore, the interaction background information of the simulated queries is determined, and the main features contained in the interaction background information are used as the target user tags. The interaction background information includes the historical queries submitted by the user in the first cycle; in the business knowledge graph, business domain knowledge applicable to the target user tags is determined. The business knowledge graph records the association relationship between user tags and business domain knowledge, as well as the association priority between user tags and business domain knowledge with association relationships; based on the business domain knowledge whose association priority with the target user tags is higher than the priority threshold, response information applicable to the simulated queries is generated.

[0089] Verifying response information essentially involves assessing whether it meets pre-defined expected results. This includes evaluating whether it accurately aligns with relevant business domain knowledge, whether it promptly incorporates new business domain knowledge (such as newly launched offers or services), whether it strictly adheres to the logical order and constraints of the business process (e.g., not skipping necessary steps), and whether it adequately fits the context (e.g., correctly interpreting the specific offer referred to by "previously issued coupons"). Only when all these dimensions are met is the response information considered valid, reliable, and possessing genuine business value.

[0090] Alarm messages are feedback prompts generated during the verification process of response information when an unmet expectation is detected. They identify specific problems in the response information. Alarm messages typically include: error type, such as "knowledge coverage is less than the coverage threshold," "process node error," "response information does not contain business domain knowledge with a new identifier," and "background relevance rate is less than the relevance rate threshold." They can also include error location and expected content. Structured alarm messages allow for rapid identification of model defects, guiding business knowledge graph optimization, prompt word adjustment, or response generation strategy iteration.

[0091] In some implementations, after generating alarm information regarding the response information of the simulated query information, at least one of the following is included: adjusting the business knowledge graph based on the alarm information; adjusting the first period corresponding to the interaction background information of the simulated query information based on the alarm information, wherein the interaction background information is historical query information in the first period.

[0092] The first period is the time window during which the interaction background information is traced back when the query information is generated. The longer the time span of the first period, the more interaction background information is considered, and the better the correlation between the query information and the interaction background information can be improved.

[0093] Specifically, the process of adjusting the business knowledge graph based on alarm information can be referred to steps 605 to 610, including: when the knowledge coverage rate of the response information of the alarm information is less than the coverage threshold, increasing the association priority between the target user tag corresponding to the simulated query information and the business domain knowledge that is related to the target user tag, and supplementing the attribute information of the business domain knowledge that is related to the target user tag, wherein the attribute information is used to describe the business domain knowledge; when the process node of the response information of the alarm information is incorrect, correcting the dependency relationship of the business domain knowledge corresponding to the erroneous process node; and when the response information of the alarm information is not containing business domain knowledge with a new identifier, monitoring whether there is new business data, and updating the business knowledge graph when there is new business data.

[0094] In some implementations, the process of adjusting the first period corresponding to the interactive background information of the simulated inquiry information based on the alarm information can refer to steps 611 to 612, including: when the background correlation rate of the alarm information is the response information of the simulated inquiry information is less than the correlation rate threshold, the time span of the first period is increased.

[0095] Background relevance rate refers to the proportion of characters in the response that correctly reference or relate to the context of the interaction. For example, if the simulated question is "Can I still use the coupon I received before?", and the response includes the specific discount referred to by "the coupon I received before" and records the availability of "the coupon I received before", then the background relevance rate is considered high. The higher this metric, the stronger the coherence and accuracy of the response with the user's actual interaction context. The relevance rate threshold is a standard for measuring the degree of relevance between the interaction's background information and the response information; it is set according to the actual situation and is not subject to any restrictions.

[0096] In some implementations, the generation of response information and the generation of response information based on simulated information sets can be implemented by a first model. This model is typically a large language model capable of automatically generating diverse simulated queries that conform to private domain business scenarios based on preset target prompts. The first model understands the business context and user behavior patterns, outputting natural language questions that closely resemble real interactions, which are then used to generate response information based on a business knowledge graph. Furthermore, the model is simultaneously optimized during the optimization of the business knowledge graph and the first cycle using simulated information sets.

[0097] In some implementations, the analysis of response information generated from the simulated information set can be achieved through a second model, which is another large language model specifically designed to evaluate whether the response content conforms to the logical requirements of the business knowledge graph. Based on business rules, user tags, and newly added identifiers, it can determine the performance of the response information in terms of knowledge accuracy, process compliance, contextual coherence, and new knowledge coverage, and output structured evaluation results or alarm information, thereby supporting the validation and continuous optimization of the business knowledge graph and the first cycle's effectiveness.

[0098] In some implementations, based on the analysis of response information from all simulated information sets, a comprehensive multi-round evaluation and analysis report can be generated. This report summarizes interaction data and self-inspection results from the perspectives of scenario, business module, and time, quantifying performance differences based on the business knowledge graph across scenarios involving different business functions, different business functions, and changes in the knowledge update cycle. The results of the multi-round evaluation and analysis report are then synchronized with relevant business personnel to facilitate their comprehensive understanding and optimization of the business knowledge graph.

[0099] Figure 7 This is a flowchart of another information processing method for determining response information according to an embodiment of this disclosure. Figure 7 As shown, another information processing method is proposed.

[0100] In this implementation, the method includes: step 701, in response to receiving query information, determining the target user tag contained in the query information; and step 702, determining the interaction background information of the query information, and using the user's main characteristics contained in the interaction background information as the target user tag, the interaction background information including historical query content raised by the user in the first period; step 703, determining business domain knowledge applicable to the target user tag in a business knowledge graph, the business knowledge graph recording the association relationship between user tags and business domain knowledge, and the association priority between user tags and business domain knowledge with association relationships; and step 704, generating response information applicable to the query information based on business domain knowledge whose association priority with the target user tag is higher than the priority threshold.

[0101] By integrating interactive contextual information into the response generation process, the system considers not only the literal intent of the query but also the user's identity characteristics and historical interaction information to determine the business process within which the query falls. It can also determine the referential content within the query based on the aforementioned factors for accurate inference. This mechanism significantly improves the personalization, business compliance, and contextual coherence of responses, effectively avoiding common problems in general models such as "forgetting history," "answering the wrong question," or "rule misuse." Furthermore, through multi-round simulation evaluation and feedback loops based on contextual information, the system can continuously optimize the knowledge graph structure and model response strategies, ensuring high availability and strong adaptability in real-world private domain scenarios.

[0102] Figure 8 This is a schematic block diagram of the structure of an information processing apparatus according to embodiments of the present disclosure. Figure 8 As shown, this disclosure provides an information processing apparatus 800, including: a tag determination module 810, configured to determine a target user tag contained in the query information in response to receiving query information; a business knowledge graph application module 820, configured to determine business domain knowledge applicable to the target user tag in a business knowledge graph, wherein the business knowledge graph records the association relationship between user tags and business domain knowledge, and the association priority between user tags and business domain knowledge with association relationships; and a response module 830, configured to generate response information applicable to the query information based on business domain knowledge whose association priority with the target user tag is higher than a priority threshold.

[0103] The information processing apparatus 800 disclosed herein may be in the form of computer software, and each module of the information processing apparatus 800 may be in the form of computer software modules.

[0104] The various modules of the information processing apparatus 800 disclosed herein are set up to implement the various steps of the information processing method. Their execution principles and steps can be referred to the preceding text and will not be repeated here.

[0105] Figure 9 This is a schematic block diagram of an electronic device according to one embodiment of the present disclosure. Figure 9 As shown, this disclosure also provides an electronic device 1000, including: a processor 1200 and a memory 1300, the memory 1300 storing execution instructions; the processor 1200 executes the execution instructions stored in the memory 1300, causing the processor 1200 to execute an information processing method.

[0106] The hardware architecture of the electronic device 1000 can be implemented using a bus architecture. The bus architecture can include any number of interconnect buses and bridges, depending on the specific application of the hardware and overall design constraints. Bus 1100 connects various circuits, including one or more processors 1200, memory 1300, and / or hardware modules. Bus 1100 can also connect various other circuits 1400, such as peripheral devices, voltage regulators, power management circuits, external antennas, etc.

[0107] Bus 1100 can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, this diagram uses only one connection line, but this does not imply that there is only one bus or one type of bus.

[0108] This disclosure also provides a readable storage medium storing a computer program that, when executed by a processor, is used to implement the methods described above. A "readable storage medium" can be any means capable of containing, storing, communicating, propagating, or transmitting a program for use by or in conjunction with an instruction execution system, apparatus, or device. More specific examples of a readable storage medium include: an electrical connection with one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and portable read-only memory (CDROM), etc.

[0109] This disclosure also provides a computer program product, the methods of which can be implemented wholly or partially through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented wholly or partially as a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed, all or part of the processes or functions of this disclosure are performed.

[0110] Computer programs or instructions can be stored in a readable storage medium or transferred from one readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The readable storage medium can be any available medium capable of access, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; an optical medium, such as a digital video optical disc; or a semiconductor medium, such as a solid-state drive. The computer-readable storage medium can be a volatile or non-volatile storage medium, or it can include both volatile and non-volatile types of storage media.

[0111] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, electronic devices, readable storage media, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0112] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0113] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0114] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0115] In the description of this specification, the references to terms such as "one embodiment / mode," "some embodiments / modes," "example," "specific example," or "some examples," etc., refer to specific features, structures, or characteristics described in connection with that embodiment / mode or example, which are included in at least one embodiment / mode or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment / mode or example. Moreover, the specific features, structures, or characteristics described may be combined in any suitable manner in one or more embodiments / modes or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments / modes or examples described in this specification, as well as the features of different embodiments / modes or examples.

[0116] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this disclosure, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0117] Those skilled in the art should understand that the above embodiments are merely for illustrating the present disclosure and are not intended to limit the scope of the disclosure. Those skilled in the art can make other changes or modifications based on the above disclosure, and these changes or modifications still fall within the scope of the present disclosure.

[0118] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0119] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0120] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0121] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0122] At the same time, it is understood that the data involved in this disclosed technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

Claims

1. An information processing method, characterized in that, include: In response to receiving an inquiry message, the system determines the target user tag contained in the inquiry message, wherein the target user tag represents the main characteristics of the user who initiated the inquiry message; In the business knowledge graph, business domain knowledge applicable to the target user tag is determined. The business knowledge graph records the association between user tags and business domain knowledge, as well as the association priority between user tags and business domain knowledge with association relationships. as well as Based on business domain knowledge that has an association priority higher than a priority threshold with the target user's tag, response information suitable for the query information is generated.

2. The information processing method according to claim 1, characterized in that, Before responding to receiving the query information, the following is included: Based on the business data of the target business domain that the query information focuses on, the business knowledge graph is constructed. The business knowledge graph also records the relationships between multiple business domain knowledge and the association priority between two related business domain knowledge.

3. The information processing method according to claim 2, characterized in that, Based on business data related to the target business area identified in the query information, the business knowledge graph is constructed, including: The user tags are defined by the main features supported by the business rules in the business data, and the business domain knowledge is defined by the business functions supported by the business rules, the multiple process nodes in the business rules that implement the business functions, and the response templates in the business data that are applicable to the business functions. Based on the business rules, the association between the user tag and the business domain knowledge, as well as the association between multiple pieces of business domain knowledge, are determined to form the business knowledge graph.

4. The information processing method according to claim 2, characterized in that, Constructing the business knowledge graph includes: When the business domain knowledge is a process node for implementing a business function, the dependency relationship between the process node and other process nodes in the business process is determined according to the business process recorded in the business rules. The business process includes multiple process nodes for implementing the business function and the execution order among the multiple process nodes.

5. The information processing method according to claim 2, characterized in that, After constructing the business knowledge graph, the following is included: In response to the generation of new business data, the business knowledge graph is updated using the new business data, and a new identifier is configured for the business domain knowledge generated based on the new business data. The new identifier has a display period and disappears when the display period expires.

6. The information processing method according to claim 2, characterized in that, After constructing the business knowledge graph, the following is also included: Based on the business knowledge graph, a simulated information set is generated. The simulated information set is used to verify the response information obtained based on the business knowledge graph. The simulated information set is either a simulated information set focusing on a single business function or a simulated information set focusing on multiple business functions. The simulated information set consists of multiple simulated query messages with a dialogue sequence, and at least one background interference message is present in the multiple simulated query messages.

7. The information processing method according to claim 6, characterized in that, Based on the aforementioned business knowledge graph, a simulated information set is generated, including: Construct target prompt words for generating the simulated information set, wherein the target prompt words include the attention quantity instruction of the business function, the generation instruction of background interference information, and the target expression method; Based on the business knowledge graph and the target prompt words, the query information generation model is invoked to generate a simulated information set containing the background interference information in the target expression manner.

8. The information processing method according to claim 7, characterized in that, After generating a simulated information set containing the background interference information in the manner described in the target representation, the process includes: Verify the validity of the simulated information set.

9. An electronic device, characterized in that, include: The memory stores execution instructions; as well as A processor that executes execution instructions stored in the memory, causing the processor to perform the information processing method according to any one of claims 1 to 8.

10. A readable storage medium, characterized in that, The readable storage medium stores execution instructions, which, when executed by a processor, are used to implement the information processing method of any embodiment of this disclosure.

Images