Inquiry method, device, electronic equipment, storage medium and program product
By using a large-scale inquiry model to determine the type of case and generate targeted inquiry suggestions, the problem of low efficiency and poor completeness of existing inquiry methods is solved, and efficient and logically rigorous collection of case information is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 360 MALL OF BEIJING STAR WORLD TECH CO LTD
- Filing Date
- 2026-02-03
- Publication Date
- 2026-05-29
Smart Images
Figure CN122114157A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence technology, and more specifically, to an inquiry method, apparatus, electronic device, computer-readable storage medium, and computer program product. Background Technology
[0002] With the continuous development of Internet technology, the number of Internet-related cases remains high, putting enormous pressure on law enforcement personnel. Faced with a large number of cases, law enforcement personnel spend a lot of time and energy handling cases, and it is difficult to conduct in-depth and detailed investigations and evidence collection for each case. Under the high-intensity work pace, the completeness of information collection is affected, resulting in the frequent absence of key elements in case records.
[0003] Existing inquiry methods rely on case inquiry templates. However, selecting templates and choosing appropriate questions from them is inefficient. Limited templates are insufficient to handle the ever-changing nature of cases, resulting in incomplete information and potential omissions during the inquiry process. Summary of the Invention
[0004] This application provides an inquiry method, apparatus, electronic device, computer-readable storage medium, and computer program product, aiming to solve the technical problems of low efficiency and poor integrity in existing inquiry methods.
[0005] Firstly, a query method is provided, which includes:
[0006] Obtain case description information from the person being questioned; Input the case description information into the query model, and use the query model to determine the type of case involved corresponding to the case description information; The big data model generates corresponding inquiry suggestions based on case description information and the type of case involved. Based on the inquiry suggestions, conduct inquiries with the inquiry recipients.
[0007] Optionally, the case description information is input into the query model, which determines the type of case involved, including: The case description information is input into the query model, which then identifies the key elements. The type of case is determined based on key elements.
[0008] Optionally, the large query model may include multiple query sub-models; Based on case description information and the type of case involved, the big data inquiry model generates corresponding inquiry suggestions, including: The corresponding sub-model is selected based on the type of case involved by using the large-scale inquiry model; The inquiry sub-model generates inquiry suggestions based on the case description information.
[0009] Optionally, the method also includes: Obtain case description samples for various case types; the case description samples include reference questions; Input the case description samples of each case type into the corresponding query sub-model to obtain the question output; Based on the reference question and the question output, adjust the parameters of the inquiry sub-model until the preset termination condition is met, end the training, and obtain the trained inquiry sub-models for each case type.
[0010] Optionally, based on the case description information and the type of case involved, the query model can generate corresponding query suggestions, including: Extract case reference procedures corresponding to the types of cases involved from a pre-set knowledge base; Key elements are extracted by querying a large-scale model based on case description information and case reference processes; The query model generates a case description knowledge graph based on key elements, and then generates query suggestions based on the case description knowledge graph.
[0011] Optionally, based on the case description information and the type of case involved, the query model can generate corresponding query suggestions, including: Extract case reference procedures corresponding to the types of cases involved from a pre-set knowledge base; By using a large query model, case description information and case reference procedures are matched to identify missing information; Based on missing information, query suggestions are generated using a large query model.
[0012] Optionally, the method also includes: After asking the questioner based on the inquiry suggestion, obtain the questioner's response information; Input the case description and response information into the inquiry model to obtain follow-up questions.
[0013] Secondly, an inquiry device is provided, the device comprising: The acquisition module is used to obtain case description information of the inquiry object; The type determination module is used to input case description information into the query model and determine the type of case involved corresponding to the case description information through the query model. The suggestion module is used to generate corresponding inquiry suggestions based on case description information and case type through the inquiry big model; The question module is used to ask questions to the target audience based on the suggested questions.
[0014] Thirdly, an electronic device is provided, the electronic device comprising: A memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of any of the methods in the first aspect of this application.
[0015] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the query method shown in any of the first aspects of this application.
[0016] Fifthly, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the steps of any one of the methods in the first aspect of this application.
[0017] The beneficial effects of the technical solutions provided in this application are: The inquiry method provided in this application analyzes case description information to determine the type of case involved, and then generates inquiry suggestions based on the description and the type of case involved. The determination of the type of case involved makes the inquiry suggestions more targeted, which is conducive to achieving precise inquiry. The inquiry model can understand unstructured and conversational information in the case description information, accurately capture key elements, and classify the case into the corresponding type of case involved, thereby quickly generating inquiry suggestions. Guided questioning through inquiry suggestions can systematically cover the key information nodes corresponding to the type of case involved, thereby ensuring that the case information obtained is more complete in content and more rigorous in logic, reducing the risk of omission, and significantly improving the efficiency and accuracy of overall case handling. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below.
[0019] Figure 1 This is a schematic diagram illustrating an application scenario of an inquiry method provided in an embodiment of this application; Figure 2 A flowchart illustrating an inquiry method provided in an embodiment of this application; Figure 3 A schematic diagram illustrating the training and use of a query model in an query method provided in this application embodiment; Figure 4 A flowchart illustrating an example of an inquiry method provided in this application embodiment; Figure 5 This is a schematic diagram of the structure of an inquiry device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device to which an inquiry method is applicable, as provided in an embodiment of this application. Detailed Implementation
[0020] The embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the embodiments described below with reference to the accompanying drawings are exemplary descriptions for explaining the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions of the embodiments of this application.
[0021] Those skilled in the art will understand that, unless otherwise stated, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the terms “comprising” and “including” as used in the embodiments of this application mean that the corresponding feature can be implemented as the presented feature, information, data, step, operation, element, and / or component, but do not exclude implementation as other features, information, data, step, operation, element, component, and / or combinations thereof supported by the art. It should be understood that when we say that an element is “connected” or “coupled” to another element, the one element can be directly connected or coupled to the other element, or it can mean that the one element and the other element establish a connection relationship through an intermediate element. Furthermore, “connected” or “coupled” as used herein can include wireless connection or wireless coupling. The terms “or,” “and / or,” “including at least one of the following,” etc., as used in this application, can be interpreted as inclusive, or mean any one or any combination thereof. For example, "including at least one of the following: A, B, C" means "any one of the following: A; B; C; A and B; A and C; B and C; A and B and C", and "A, B or C" or "A, B and / or C" means "any one of the following: A; B; C; A and B; A and C; B and C; A and B and C".
[0022] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.
[0023] In the specific embodiments of this application, any data related to the object, such as data involved in the use of the application, is required. When the embodiments of this application are applied to specific products or technologies, permission or consent from the object is required, and the collection, use, and processing of the relevant data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. In other words, if any of the aforementioned object-related data is involved in the embodiments of this application, this data must be obtained with the object's authorization and consent, and in accordance with the relevant laws, regulations, and standards of the country and region.
[0024] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.
[0025] Currently, the number of internet-related cases remains high. Faced with a large number of cases, investigators are often overwhelmed and find it difficult to conduct in-depth investigations and evidence collection for each case. Under such high-intensity work, the completeness of information collection is affected, often resulting in missing key elements in case records. Investigators' lack of understanding of the technical principles behind the cases is a significant reason for this information gap. Many cybercrimes heavily rely on technologies such as the internet, mobile communications, and financial payments, often employing tools such as fake base stations, phishing websites, and Trojan viruses. The process often involves complex technical steps. However, many investigators lack the necessary technical background and professional knowledge, making it difficult to accurately extract key evidence when faced with such technically complex cases. For example, in handling a typical phishing case, a lack of understanding of the phishing website's operating mechanism may lead to the neglect of collecting important electronic evidence such as website server logs and IP address traces, thus affecting the subsequent investigation of the case. The current inquiry process uses different types of case inquiry templates for execution inquiries. However, the limited templates are insufficient to cope with the ever-changing methods of crime. It is necessary to improve the professional knowledge and understanding of the various technical principles in the case among the case handlers, but this is difficult to achieve and popularize, which affects the efficiency of actual case handling.
[0026] The inquiry methods, apparatus, electronic devices, computer-readable storage media, and computer program products provided in this application are intended to solve at least one of the above-mentioned technical problems of the prior art.
[0027] In response to at least one of the aforementioned technical problems or areas requiring improvement in related technologies, this application proposes an inquiry method, apparatus, electronic device, computer-readable storage medium, and computer program product. The inquiry method provided by this solution analyzes case description information to determine the type of case involved, thereby generating inquiry suggestions based on the description and the type of case involved. Determining the type of case involved makes the inquiry suggestions more targeted, facilitating precise inquiry. The inquiry model can understand unstructured and colloquial information in the case description, accurately capture key elements, and classify the case into the corresponding type of case involved, thus quickly generating inquiry suggestions. Guided questioning based on these suggestions systematically covers the key information nodes corresponding to the type of case involved, ensuring that the obtained case information is more complete in content and more logically rigorous, reducing the risk of omissions, and significantly improving the overall efficiency and accuracy of case handling.
[0028] The technical solutions of this application and their effects are described below through several exemplary embodiments. It should be noted that the following embodiments can be referenced, borrowed from, or combined with each other. Identical terms, similar features, and similar implementation steps in different embodiments will not be repeated.
[0029] Figure 1 This is a schematic diagram of an application scenario for the query method provided in the embodiments of this application. The application environment may include a terminal 100 or a server, wherein a query system is installed on the terminal 100 or the server.
[0030] Specifically, the inquiry system in terminal 100 obtains the case description information of the inquiry object, inputs the case description information into the inquiry big model, determines the type of case involved corresponding to the case description information through the inquiry big model, generates corresponding inquiry suggestions based on the case description information and the type of case involved through the inquiry big model, and conducts inquiries on the inquiry object based on the inquiry suggestions.
[0031] In practical implementation, the above inquiry methods can be deployed in different environments to form diverse application forms, including: deploying the inquiry model and data processing logic on cloud servers or local data center servers, with inquiry personnel accessing the service through client applications (web pages, desktop software, and apps) on terminal devices (such as PCs, laptops, mobile police terminals, and tablets). The terminal is responsible for inputting case description information and displaying inquiry suggestions, while the server is responsible for performing complex model reasoning, classification analysis, and inquiry suggestion generation tasks; deploying the lightweight or distilled inquiry model directly on terminal devices, with all data processing and model reasoning completed on the local terminal, without the need for real-time communication with cloud servers, ensuring data privacy and not relying on network connections; and encapsulating the above methods into standardized application programming interfaces (APIs) or software-as-a-service platforms, providing them to third-party developers or enterprises. Other business systems (such as customer relationship management systems (CRM), case management systems, and intelligent customer service systems) can send case description information to the platform of this solution by calling the API, obtain structured inquiry suggestions, and integrate them into their own workflows, reducing the technical barriers to use and enabling intelligent inquiry capabilities to quickly empower various industries, demonstrating good scalability and value.
[0032] Furthermore, the inquiry method proposed in this application can be applied not only to the fields of public safety, financial risk control and compliance, but also to the fields of insurance, corporate investigation and customer service, and healthcare. In this case, the above-mentioned case description can be replaced with the corresponding event description or task description. The method in the embodiments of this application can deeply understand the information and logic of various fields, and improve the efficiency, completeness and logic of information collection in the case of inquiry.
[0033] The above application scenario is just an example and does not limit the application scenarios of the query method in this application.
[0034] Those skilled in the art will understand that the terminal can be a smartphone (such as an Android phone, an iOS phone, etc.), a tablet computer, a laptop computer, a digital broadcast receiver, a MID (Mobile Internet Device), a PDA (Personal Digital Assistant), a desktop computer, a smart home appliance, an in-vehicle terminal (such as an in-vehicle navigation terminal, an in-vehicle computer, etc.), a smart speaker, a smartwatch, etc. The terminal and the server can be connected directly or indirectly through wired or wireless communication, but are not limited to these.
[0035] The server may include servers installed with the ability to handle database operations. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server or server cluster providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. The embodiments of this application can be applied to various scenarios, including but not limited to cloud technology, artificial intelligence, smart transportation, and assisted driving. Specific applications can be determined based on actual application needs and are not limited here.
[0036] In some possible implementations, taking the query system as an example, embodiments of this application provide a query method, such as... Figure 2 As shown, it may include the following steps: S210, Obtain case description information of the person being questioned.
[0037] The case description information includes the oral information provided by the person being questioned and other information provided, such as screenshots of conversations, transfer records, voice messages, video content, or other supporting information.
[0038] Specifically, the initial description information uploaded by the case handlers is obtained. This initial description information may be data in different forms and formats, such as text, images, voice, or video. The initial description information is preprocessed and converted into a unified text format to obtain case description information, which can then be processed uniformly through the large-scale inquiry model.
[0039] In the specific implementation process, the preprocessing steps for the initial descriptive information may include: processing multimodal information such as images, documents, or audio / video files; allowing users to upload images (such as ID cards, receipts, contract screenshots, accident scene photos) or documents; automatically extracting the text information through optical character recognition (OCR) technology and integrating it with manually entered descriptions; for uploaded audio or video files (such as call recordings, surveillance video clips), speech recognition technology can be used to extract dialogue content, and video analysis technology can be used to identify scenes, people, and key actions to generate structured descriptive information; furthermore, source tags and confidence scores can be assigned to information from different sources (such as user text, OCR text, and speech-to-text), and deduplication, verification, and integration can be performed to ultimately form a high-quality case description that is uniform in form and comprehensive, laying a solid data foundation for subsequent intelligent analysis and the generation of inquiry suggestions.
[0040] S220: Input the case description information into the query model, and use the query model to determine the type of case involved corresponding to the case description information.
[0041] Specifically, the case description information is input into the inquiry model, which analyzes the key elements in the case description information, determines the corresponding case type based on the key elements, and then quickly generates inquiry suggestions based on the case type to improve case processing efficiency.
[0042] In the specific implementation process, such as Figure 3 As shown, the training steps for the large-scale inquiry model can include data pre-training and business scenario model training. Data pre-training includes: obtaining training samples to train the large-scale model and obtaining a preliminary large-scale inquiry model. Training samples can include high-quality, precisely labeled data or unstructured data. High-quality, precisely labeled data refers to high-quality data with fine annotations, such as laws and regulations, historical inquiry data, or expert experience. Unstructured data can include data based on network knowledge, historical judgments, official anti-fraud public account posts, or anti-fraud network public account posts. Business scenario model training includes: setting up multiple inquiry sub-models (agents) in the preliminary large-scale inquiry model, and training each inquiry sub-model using case sample data of different types of cases until the preset termination conditions are met, determining the trained inquiry sub-models, and obtaining the trained large-scale inquiry model.
[0043] In the specific implementation process, such as Figure 3As shown, the query model can be enhanced based on a preset knowledge base during actual use, thereby increasing its professionalism. After processing case description information, the query model can perform similarity searches in the preset knowledge base to determine the corresponding content, thus generating more accurate and reliable query suggestions. The preset knowledge base includes the principles of phishing techniques, the investigation and judgment ideas of various cases, the principles of various network technology tools, and common methods of committing crimes. By introducing case sample analysis knowledge bases, website sample analysis knowledge bases, and internet-related case investigation knowledge bases, it incorporates deeper levels of judgment logic and practical experience, enabling the model to not only achieve accurate analysis of case description information but also provide good suggestions in terms of querying.
[0044] In practical implementation, the basic architecture of the large-scale inquiry model can include an encoder and a decoder. The encoder is responsible for reading and understanding the input case description samples, while the decoder is responsible for generating questions word by word based on the encoder's understanding, thus obtaining inquiry suggestions. During training, prepared case description samples are input into the model's encoder, and corresponding reference questions are used as the decoder's standard answers (labels). The large-scale model generates its own questions based on the case description samples. Then, the difference between the questions generated by the model and the standard answer questions is compared. Based on this difference, all parameters inside the large-scale model are adjusted in reverse to make the questions generated next time closer to the standard answers. The training steps are repeated until the model can stably generate fluent, relevant, and in-depth questions based on various case descriptions. After training, the quality of the questions generated by the model needs to be evaluated. Evaluation criteria typically include: judging the fluency of the questions, whether the relevance is close, and whether it can generate questions from different perspectives when faced with the same description. If the evaluation is passed, the trained large-scale inquiry model is obtained.
[0045] In practice, the query model also has self-learning and optimization capabilities. As the number of cases handled increases, the model will continuously adjust its query strategy based on feedback from query suggestions, thereby making it more accurate and efficient in case handling. This dynamic optimization mechanism ensures that the technology can still maintain a high level of responsiveness when facing new types of online cases.
[0046] S230 uses a large-scale inquiry model to generate corresponding inquiry suggestions based on case description information and the type of case involved.
[0047] The inquiry suggestions can include a structured list of questions, risk warnings analyzed from uploaded information, and tips or scripts for asking questions. For example, questions can be displayed in a list with risk warnings such as: "The current person being inquired about is emotionally unstable. Pay attention to calming them down." Suggested tips include: "Maintain an objective and neutral tone and avoid being accusatory."
[0048] Specifically, the inquiry model analyzes case description information based on the case process corresponding to the type of case involved, obtains the overall case context, and then generates inquiry suggestions for the case. When handling cases, the model quickly sorts out the key features of the case based on the description of the incident by the inquiring party, and generates a detailed inquiry outline accordingly, which can help investigators quickly grasp the core of the case and conduct targeted inquiries.
[0049] For example, when analyzing a case as a money transfer case, you can focus on the transaction amount, transfer records, transfer time, the language used to guide the transfer, and the basis for building trust in the case description information. You can also pay attention to the basic information of the person being questioned and adjust the relevant language according to the individual's basic information to generate more appropriate questioning suggestions. For example, use technical terms for people with professional knowledge, and use easy-to-understand descriptions for the elderly.
[0050] S240, Inquire about the object of inquiry based on the inquiry suggestion.
[0051] The inquiry suggestions include multiple questions, along with corresponding tone, order, and relevant prompts. The inquiry can be directed at either the suspect or the victim.
[0052] Specifically, the query suggestions are displayed on the terminal's visual interface, so that the query recipient can be questioned based on the questions and related information in the displayed query suggestions. Alternatively, the questions can be directly displayed to the query recipient for questioning.
[0053] In practice, the inquiry suggestions can pre-determine the priority of questions and display them in order. After the inquirer answers, they can click "next" to proceed with the next question. Alternatively, all questions can be set as selection or judgment questions to avoid the inquirer answering a lot of irrelevant content, thereby improving the efficiency and accuracy of information acquisition. Guided questions can also be generated to help the inquirer recall information. By using diverse methods for inquiry and personalizing the matching of inquirers, the user experience can be improved.
[0054] In some possible implementations, the above steps involve inputting case description information into a query model, and using the query model to determine the type of case involving the case description information, including: The case description information is input into the query model, which then identifies the key elements. The type of case is determined based on key elements.
[0055] The types of cases involved include investment and wealth management, identity theft, fraudulent order rebates, impersonation of institutions, and impersonation of official entities.
[0056] Specifically, the case description information is input into the query model, which identifies the case description information, extracts key elements, and matches the key elements based on preset type matching rules to determine the corresponding case type.
[0057] In the specific implementation process, the case descriptions corresponding to each type of case in the preset knowledge base can be converted into vectors, and the obtained case description information can also be converted into target vectors. The target vectors are compared with the corresponding vectors in the preset knowledge base, and the similarity is calculated. This determines the type of case corresponding to the vector with the highest similarity to the target vector. This can identify colloquial content, has strong anti-interference ability, and is more tolerant of colloquial expressions. For example, the victim's description: "Someone called me and said that I had a lost package and that they would give me double compensation. Then they asked me to open a website and follow their instructions." This description does not contain keywords such as phishing or crime, but the corresponding vector will be highly similar to the vector of cases in the knowledge base that impersonate customer service for refunds, which belongs to the category of impersonating institutions.
[0058] In practice, a case may involve multiple methods. The big model outputs the confidence level for each type of case based on the logical relationship in the case description information. Based on the confidence level and a preset threshold, it determines which type of case it belongs to. For example, if the other party first asks the inquirer to open a website, it belongs to the category of counterfeit organization. However, if the inquirer's website is blocked by security and cannot be opened, and then the other party guides the inquirer to participate in a cashback activity and the inquirer successfully participates, then the type of case involved is the cashback activity.
[0059] In some possible implementations, the above steps involve generating corresponding inquiry suggestions based on case description information and the type of case using a large inquiry model, including: The corresponding sub-model is selected based on the type of case involved by using the large-scale inquiry model; The inquiry sub-model generates inquiry suggestions based on the case description information.
[0060] The large inquiry model includes multiple sub-inquiry models, each with its corresponding case type.
[0061] Specifically, by using the large inquiry model, corresponding inquiry sub-models are determined based on the type of case involved. Case description information is input into the corresponding inquiry sub-model to obtain corresponding inquiry suggestions. By conducting targeted training on inquiry sub-models for different types of internet-related cases, the corresponding inquiry sub-models can identify the key elements of the corresponding type of case and generate corresponding inquiry suggestions based on specific case description information.
[0062] In practice, when the case description involves multiple types of related content, the case description can be input into multiple inquiry sub-models corresponding to the relevant types of cases, each outputting its own inquiry suggestions. Then, the outputs of multiple inquiry sub-models are merged to obtain the final inquiry suggestion. For example, if the initial description of the inquiry object is that there was an investment loss and it is an investment dispute, but the large model analysis finds that the official investment platform is likely fake, and the case is likely to be an impersonation of an official entity, then inquiry suggestions can be generated based on the output of the impersonation of an official entity inquiry sub-model. This enhances the logic of the inquiry content analysis and makes the inquiry suggestions more accurate.
[0063] In some possible implementations, the above method further includes: Obtain case description samples for various case types; Input the case description samples of each case type into the corresponding query sub-model to obtain the question output; Based on the reference question and the question output, adjust the parameters of the inquiry sub-model until the preset termination condition is met, end the training, and obtain the trained inquiry sub-models for each case type.
[0064] The case description samples include reference questions, which are pre-prepared and appropriate questions.
[0065] Specifically, case description samples of various case types are obtained, and each case description sample is input into the corresponding inquiry sub-model to obtain question outputs. Based on the reference questions and question outputs, the parameters of the inquiry sub-models are adjusted until the preset termination conditions are met, thus ending the training and obtaining trained inquiry sub-models for each case type. Inquiry sub-models corresponding to different types of cases are trained separately. Each sub-model can accurately grasp the professional terminology, modus operandi, psychological manipulation techniques, and relevant laws and regulations of that type of case, thereby generating incision suggestions that are incisive, logically rigorous, and highly authoritative, improving the depth and professionalism of the inquiry.
[0066] In the specific implementation process, the reference questions corresponding to the case description samples can be questions raised by experts in the relevant fields. There can be multiple questions, and these questions are pre-labeled on the corresponding case description samples. Then, an existing large language model can be selected as the initial large model. These models have language understanding capabilities. The case description samples are input into the initial large model to obtain the question output. The question output is compared with the reference questions to update the parameters of the corresponding initial large model until the preset training termination conditions are met, and the trained query sub-model is obtained.
[0067] For example, a sample case description could be: A victim receives a phone call from someone claiming to be an official. The caller accurately states the victim's name and ID number and claims that the victim is involved in a serious case, demanding that all funds under their name be transferred to a designated safe account for investigation. The suggested reference questions could be: How did the caller contact you? Was it a mobile phone or a landline? Did you verify the caller's identity before transferring the money? Did you check or call the official number? Based on such case description samples and reference questions, the inquiry sub-model is trained, making the inquiry suggestions output by the sub-model more closely match the description and more logically clear.
[0068] In some possible implementations, the above steps involve generating corresponding inquiry suggestions based on case description information and the type of case using a large inquiry model, including: Extract case reference procedures corresponding to the types of cases involved from a pre-set knowledge base; Key elements are extracted by querying a large-scale model based on case description information and case reference processes; The query model generates a case description knowledge graph based on key elements, and then generates query suggestions based on the case description knowledge graph.
[0069] Key elements may include entity, behavior, timing, and emotional state.
[0070] Specifically, the process involves determining the case reference process for the corresponding case type from a pre-set knowledge base, identifying key elements in the case description information based on the case reference process, using each entity in the key elements as a node, and emotional state as an attribute. The process also involves determining the relationships between nodes in the case description information based on behavior, time sequence, etc., connecting the nodes based on these relationships to generate a case description knowledge graph, identifying problematic or missing nodes based on the case description knowledge graph, and generating corresponding inquiry suggestions based on these nodes.
[0071] In the specific implementation process, after obtaining the case description knowledge graph, missing elements in the case description knowledge graph can be marked based on the case reference process. The confidence level of each node in the case description knowledge graph and its relationship with other nodes can also be evaluated, and nodes with confidence levels lower than the preset minimum threshold or nodes with logical conflicts can be marked. High-risk nodes can also be predicted, such as nodes where the object's emotions are extreme or the victim is too confident that the other party may commit the crime again. When displaying inquiry suggestions, the case description knowledge graph and the corresponding markings can be displayed at the same time to assist the case handlers in making decisions.
[0072] In some possible implementations, the above steps involve generating corresponding inquiry suggestions based on case description information and the type of case using a large inquiry model, including: Extract case reference procedures corresponding to the types of cases involved from a pre-set knowledge base; By using a large query model, case description information and case reference procedures are matched to identify missing information; Based on missing information, query suggestions are generated using a large query model.
[0073] Specifically, the system extracts case reference processes corresponding to the types of cases involved from a pre-set knowledge base. The query model compares the structure of the case reference process with the case description information. If steps, objects, or behaviors are missing in the case description information, corresponding query suggestions can be generated based on the missing information.
[0074] In practice, the large-scale inquiry model can learn the correlation between elements during training. Based on this correlation, it can identify missing elements and generate corresponding inquiry suggestions. For example, in typical cases, emotional communication, mentors, and fraudulent applications often appear together with investment platforms. When the large-scale inquiry model identifies that the case description mentions an investment platform, it will focus on identifying frequently occurring content. If it does not appear in the description, it can generate questions to determine whether there was emotional communication, mentors, or fraudulent applications when the case occurred, thus obtaining more comprehensive case information.
[0075] In practice, two nodes that may appear independent in the description may be related in the actual course of a case. However, this connection may be missing in the case description information of the person being questioned. In such cases, corresponding inquiry suggestions can be generated based on the missing connection. For example, if a victim transfers money to account A and then transfers money to account B after a period of time, these two accounts may appear independent in the case description information. However, in reality, accounts A and B may both be aggregated into account C. In this case, the inquiry model can predict that there is some connection between accounts A and B, and inquiry is needed to complete the relationship.
[0076] In some possible implementations, the above method further includes: After asking the questioner based on the inquiry suggestion, obtain the questioner's response information; Input the case description and response information into the inquiry model to obtain follow-up questions.
[0077] Specifically, after conducting inquiries based on the inquiry suggestions, investigators can continue to input the responses to the questions into the inquiry model, adjust the previous inquiry suggestions based on the responses, and obtain follow-up inquiry suggestions.
[0078] In practice, the responses from the respondents can be not only text, but also images, documents, audio and video, or other multimodal information. In the actual inquiry process, a complete case description is input for the first time to generate the first batch of questions. After obtaining the responses to the first batch of questions, the inquiry model generates more in-depth follow-up questions based on the original case description information and the respondents' responses. After obtaining the corresponding responses, the original case description, responses, and historical question-and-answer dialogues are combined and used as input to the inquiry model to obtain suggestions for the next follow-up questions.
[0079] In the specific implementation process, when preparing training data samples, the above method not only requires a single description and question, but also requires the construction of complete dialogue samples. By training the large model on conversational data samples, the inquiry model can learn how to ask progressively deeper questions based on new information, thereby uncovering the key details of the case.
[0080] For example, in the initial input of the query model: the victim clicked on a link in a text message and entered their salary card information and verification code, subsequently discovering that funds had been stolen from their account. The query model outputs: "What is the content of the link in the text message?" and "Was there anything unusual on the page when entering the verification code?" After conducting the queries, the model receives a response from the victim: "The victim clicked on the link in the text message, and the page looked exactly like the official website." The query model then generates follow-up questions: "Did you verify that the domain name of the link is the official domain?" and "Besides your salary card information, were you asked to enter any other personal information?" By asking follow-up questions in this way, more details can be obtained, which is beneficial to the smooth investigation of the case.
[0081] In the above embodiments, by analyzing the case description information, the type of case is determined, and inquiry suggestions are generated based on the description and the type of case. The determination of the type of case makes the inquiry suggestions more targeted, which is conducive to achieving precise inquiry. The inquiry model can understand the unstructured and colloquial information in the case description information, accurately capture key elements, and classify the case into the corresponding type of case, thereby quickly generating inquiry suggestions. Guided questioning through inquiry suggestions can systematically cover the key information nodes corresponding to the type of case, thereby ensuring that the case information obtained is more complete in content and more rigorous in logic, reducing the risk of omission, and significantly improving the efficiency and accuracy of the overall case handling.
[0082] Furthermore, a knowledge graph can be generated based on the case description information. This knowledge graph can be used to quickly identify abnormal or missing content, thereby generating more accurate inquiry suggestions. When providing inquiry suggestions, the knowledge graph can be displayed to the case handlers to help them quickly understand the overall structure of the case and improve case handling efficiency.
[0083] In one example, the query method of this application, such as Figure 4 As shown, it may include: S410, Obtain case description information of the person being questioned.
[0084] S420: Input the case description information into the query model, and use the query model to determine the type of case involved corresponding to the case description information.
[0085] The large inquiry model includes multiple inquiry sub-models, with each inquiry sub-model corresponding to a type of case.
[0086] S430, an inquiry sub-model in the large inquiry model based on the type of case involved.
[0087] S440 generates inquiry suggestions based on case description information and case reference process through the inquiry sub-model.
[0088] S450, inquires about the object of inquiry based on the inquiry suggestion.
[0089] S460, obtain the response information from the person being queried.
[0090] S470: Input the case description information and response information into the inquiry model to obtain follow-up questions and conduct follow-up inquiries.
[0091] Specifically, the conditions for stopping the questioning of the subject include instructions from the case handler to stop, or stopping after a certain number of rounds. After stopping, the entire questioning dialogue can be compiled into a complete questioning report, and a knowledge graph corresponding to the case can be generated and displayed based on the questioning report.
[0092] The aforementioned inquiry method analyzes case description information to determine the type of case involved, and then generates inquiry suggestions based on the description and the type of case involved. The determination of the type of case involved makes the inquiry suggestions more targeted, which is conducive to achieving precise inquiry. The inquiry model can understand unstructured and conversational information in the case description information, accurately capture key elements, and classify the case into the corresponding type of case involved, thereby quickly generating inquiry suggestions. Guided questioning through inquiry suggestions can systematically cover the key information nodes corresponding to the type of case involved, thereby ensuring that the case information obtained is more complete in content and more rigorous in logic, reducing the risk of omissions, and significantly improving the efficiency and accuracy of overall case handling.
[0093] Furthermore, a knowledge graph can be generated based on the case description information. This knowledge graph can be used to quickly identify abnormal or missing content, thereby generating more accurate inquiry suggestions. When providing inquiry suggestions, the knowledge graph can be displayed to the case handlers to help them quickly understand the overall structure of the case and improve case handling efficiency.
[0094] This application provides an inquiry device, such as... Figure 5 As shown, the inquiry device 50 may include: an acquisition module 510, a type determination module 520, a suggestion module 530, and an inquiry module 540, wherein, Module 510 is used to obtain case description information of the inquiry object; The type determination module 520 is used to input case description information into the query big model and determine the type of case involved corresponding to the case description information through the query big model. Module 530 is used to generate corresponding inquiry suggestions based on case description information and case type through the inquiry big model; The question module 540 is used to ask questions to the object of the question based on the question suggestions.
[0095] As an optional embodiment, in this device, the type determination module 520 is specifically used for: The case description information is input into the query model, which then identifies the key elements. The type of case is determined based on key elements.
[0096] As an optional embodiment, in this device, module 530 is specifically used for: The corresponding sub-model is selected based on the type of case involved by using the large-scale inquiry model; The inquiry sub-model generates inquiry suggestions based on the case description information.
[0097] As an optional embodiment, in this device, module 530 is specifically used for: Obtain case description samples for various case types; the case description samples include reference questions; Input the case description samples of each case type into the corresponding query sub-model to obtain the question output; Based on the reference question and the question output, adjust the parameters of the inquiry sub-model until the preset termination condition is met, end the training, and obtain the trained inquiry sub-models for each case type.
[0098] As an optional embodiment, in this device, module 530 is specifically used for: Extract case reference procedures corresponding to the types of cases involved from a pre-set knowledge base; Key elements are extracted by querying a large-scale model based on case description information and case reference processes; The query model generates a case description knowledge graph based on key elements, and then generates query suggestions based on the case description knowledge graph.
[0099] As an optional embodiment, in this device, module 530 is specifically used for: Extract case reference procedures corresponding to the types of cases involved from a pre-set knowledge base; By using a large query model, case description information and case reference procedures are matched to identify missing information; Based on missing information, query suggestions are generated using a large query model.
[0100] As an optional embodiment, the device also includes a follow-up questioning module, specifically used for: After asking the questioner based on the inquiry suggestion, obtain the questioner's response information; Input the case description and response information into the inquiry model to obtain follow-up questions.
[0101] The inquiry device provided in this application analyzes case description information to determine the type of case involved, and then generates inquiry suggestions based on the description and the type of case involved. The determination of the type of case involved makes the inquiry suggestions more targeted, which is conducive to achieving precise inquiry. The inquiry model can understand unstructured and colloquial information in the case description information, accurately capture key elements, and classify the case into the corresponding type of case involved, thereby quickly generating inquiry suggestions. Guided questioning through inquiry suggestions can systematically cover the key information nodes corresponding to the type of case involved, thereby ensuring that the case information obtained is more complete in content and more rigorous in logic, reducing the risk of omission, and significantly improving the efficiency and accuracy of overall case handling.
[0102] Furthermore, a knowledge graph can be generated based on the case description information. This knowledge graph can be used to quickly identify abnormal or missing content, thereby generating more accurate inquiry suggestions. When providing inquiry suggestions, the knowledge graph can be displayed to the case handlers to help them quickly understand the overall structure of the case and improve case handling efficiency.
[0103] The apparatus in this application embodiment can execute the method provided in this application embodiment, and its implementation principle is similar, and it has corresponding technical effects. The actions performed by each module in the apparatus of each embodiment of this application correspond to the steps in the method of each embodiment of this application. For a detailed functional description of each module of the apparatus, please refer to the description in the corresponding method shown above, which will not be repeated here.
[0104] This application provides an electronic device including a memory, a processor, and a computer program stored on the memory. The processor executes the computer program to implement the steps of the method provided in any optional embodiment of this application. Compared with the prior art, it can achieve the following: by analyzing case description information, the type of case is determined, and inquiry suggestions are generated based on the description and the type of case. The determination of the type of case makes the inquiry suggestions more targeted, which is conducive to achieving accurate inquiry. The inquiry model can understand unstructured and colloquial information in the case description information, accurately capture key elements, and classify the case into the corresponding type of case, thereby quickly generating inquiry suggestions. Guided questioning through inquiry suggestions can systematically cover the key information nodes corresponding to the type of case, thereby ensuring that the obtained case information is more complete in content and more rigorous in logic, reducing the risk of omission, and significantly improving the efficiency and accuracy of overall case handling.
[0105] In one alternative embodiment, an electronic device is provided, such as Figure 6 As shown, this device can be a screen or display for an electronic device, and its internal structure diagram can be as follows. Figure 6 As shown, this electronic device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. The display unit is used to create a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the electronic device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the electronic device, or external keyboards, touchpads, mice, air mice, or remote controls, etc.
[0106] Those skilled in the art will understand that Figure 6The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0107] This application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it can implement the steps and corresponding content of the aforementioned method embodiments.
[0108] It should be noted that the computer-readable storage medium described above in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.
[0109] This application also provides a computer program product, including a computer program that, when executed by a processor, can implement the steps and corresponding content of the aforementioned method embodiments.
[0110] The terms "first," "second," "third," "fourth," "1," "2," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in a sequence other than that shown in the illustrations or text descriptions.
[0111] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0112] It should be understood that although arrows indicate various operation steps in the flowcharts of this application's embodiments, the order in which these steps are implemented is not limited to the order indicated by the arrows. Unless explicitly stated herein, in some implementation scenarios of this application's embodiments, the implementation steps in each flowchart can be executed in other orders as required. Furthermore, some or all steps in each flowchart, based on the actual implementation scenario, may include multiple sub-steps or multiple stages. Some or all of these sub-steps or stages can be executed at the same time, and each sub-step or stage can also be executed at different times. In scenarios where execution times differ, the execution order of these sub-steps or stages can be flexibly configured according to requirements, and this application's embodiments do not limit this.
[0113] The above description is only an optional implementation method for some implementation scenarios of this application. It should be noted that for those skilled in the art, other similar implementation methods based on the technical concept of this application without departing from the technical concept of this application also fall within the protection scope of the embodiments of this application.
Claims
1. A query method, characterized in that, include: Obtain case description information from the person being questioned; The case description information is input into the query model, and the case type corresponding to the case description information is determined by the query model. The inquiry model generates corresponding inquiry suggestions based on the case description information and the type of case involved. The inquiry is conducted on the object of the inquiry based on the inquiry suggestions.
2. The inquiry method according to claim 1, characterized in that, The step of inputting the case description information into the query model and determining the type of case involving the case description information through the query model includes: The case description information is input into the query model, and the query model identifies the key elements of the case description information. The type of case is determined based on the aforementioned key elements.
3. The inquiry method according to claim 1, characterized in that, The large query model includes multiple sub-query models; The process of generating corresponding inquiry suggestions based on the case description information and the type of case using the inquiry big data model includes: The corresponding inquiry sub-model is selected based on the type of case involved by the large inquiry model. The inquiry sub-model generates the inquiry suggestions based on the case description information.
4. The inquiry method according to claim 3, characterized in that, The method further includes: Obtain case description samples for various case types; the case description samples include reference questions; Input the case description samples of each of the aforementioned case types into the corresponding inquiry sub-model to obtain the question output; Based on the reference question and the question output, the parameters of the inquiry sub-model are adjusted until the preset termination condition is met, the training ends, and the trained inquiry sub-models for each of the case types are obtained.
5. The inquiry method according to claim 1, characterized in that, The process of generating corresponding inquiry suggestions based on the case description information and the type of case using the inquiry big data model includes: Extract the case reference process corresponding to the type of case from the preset knowledge base; Based on the case description information and the case reference process, the query big data model extracts key elements; The query model generates a case description knowledge graph based on the key elements, and then generates query suggestions based on the case description knowledge graph.
6. The inquiry method according to claim 1, characterized in that, The method further includes: After inquiring with the object of inquiry based on the inquiry suggestion, obtain the object of inquiry's response information; The case description information and the response information are input into the query model to obtain follow-up question suggestions.
7. An inquiry device, characterized in that, include: The acquisition module is used to obtain case description information of the inquiry object; The type determination module is used to input the case description information into the query model and determine the type of case involved corresponding to the case description information through the query model. The suggestion module is used to generate corresponding inquiry suggestions based on the case description information and the type of case involved through the inquiry big model; The question module is used to ask the object of inquiry based on the inquiry suggestions.
8. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1-6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the query method according to any one of claims 1-6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-6.