Information processing method and device, storage medium and electronic equipment
By integrating large language models and knowledge graphs, the problem of low accuracy in direct debit and settlement services of traditional customer service systems has been solved, enabling efficient and accurate solution generation and verification, and improving user trust.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- INDUSTRIAL AND COMMERCIAL BANK OF CHINA
- Filing Date
- 2025-12-31
- Publication Date
- 2026-04-21
AI Technical Summary
Traditional customer service systems often exhibit low accuracy in handling complex, frequent, and sensitive deduction and settlement business scenarios, frequently resulting in logical judgment errors.
The system employs a pre-defined large language model to integrate and analyze text information, log files, and business data files. It uses a knowledge graph to identify problem types and generate solutions, and verifies the feasibility of the solutions in a sandbox environment. The model and knowledge graph are then optimized to improve accuracy.
It improves the accuracy and efficiency of problem handling, enhances the system's logical reasoning ability, provides clear explanations of business logic, and increases user trust.
Smart Images
Figure CN121901484A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of financial technology, and more specifically, to an information processing method and apparatus, a storage medium and an electronic device. Background Technology
[0002] In the financial sector, particularly in the direct debit and clearing operations of banks and payment institutions, current customer service systems face a series of technical challenges that significantly limit their effectiveness in today's complex financial environment. Traditional customer service systems rely on static rules and frequently asked questions (FAQ) libraries, which perform adequately when handling basic operational issues. However, they exhibit significant shortcomings when facing highly complex, frequent, and sensitive direct debit and clearing scenarios. They struggle with issues involving logical judgments or complex dependencies, often exhibiting biases in understanding and judging problems, resulting in low accuracy in problem handling.
[0003] The use of pre-defined FAQ libraries to handle customer service issues in related technologies has resulted in low accuracy in problem resolution, and no effective solution has yet been proposed. Summary of the Invention
[0004] The main objective of this application is to provide an information processing method, apparatus, storage medium, and electronic device to address the problem that the accuracy of problem handling is low when using a preset FAQ library to process customer service issues in related technologies.
[0005] To achieve the above objectives, according to one aspect of this application, an information processing method is provided. The method includes: receiving text information input by a target object and a target file associated with the text information, wherein the text information describes a problem in a target financial business; performing a fusion analysis on the text information and the target file using a preset large language model to determine the problem type of the text information; determining a solution corresponding to the text information based on a preset knowledge graph, the text information, and the problem type; verifying the solution; and returning the verified solution to the target object.
[0006] Furthermore, when the target files include log files and business data files, a pre-defined large language model is used to perform fusion analysis on the text information and the target files to determine the problem type of the text information. This includes: extracting features from the text information, log files, and business data files separately using the pre-defined large language model to obtain the text feature vector corresponding to the text information, the log feature vector corresponding to the log files, and the data feature vector corresponding to the business data files; performing weighted fusion processing on the text feature vector, log feature vector, and data feature vector to obtain the fused feature vector; and using a classifier to classify the fused feature vector to obtain the problem type of the text information.
[0007] Furthermore, feature extraction is performed on text information, log files, and business data files using a pre-defined large language model to obtain text feature vectors for text information, log feature vectors for log files, and data feature vectors for business data files. This includes: segmenting text information using the natural language processing module of the pre-defined large language model to obtain a segmentation set; performing dependency parsing on the segmentation set to obtain dependency parsing information; and vectorizing the dependency parsing information to obtain text feature vectors. Additionally, matching log files using the log parsing module of the pre-defined large language model based on pre-defined regular expressions to obtain target event information; and vectorizing the target event information to obtain log feature vectors. Finally, data verification is performed on business data files using the business data processing module of the pre-defined large language model to determine the data quality information; and vectorizing the data quality information to obtain data feature vectors.
[0008] Furthermore, based on the preset knowledge graph, text information, and question type, determining the solution corresponding to the text information includes: converting the text information into a graph query statement; determining the set of problem causes corresponding to the text information based on the preset knowledge graph, graph query statement, and question type; and determining the solution corresponding to the text information based on the set of problem causes.
[0009] Furthermore, based on the preset knowledge graph, graph query statement, and question type, the set of question causes corresponding to the text information is determined by: executing the graph query statement to retrieve entities, attributes, and relationships associated with the question type from the preset knowledge graph and obtaining the retrieval results; and determining the set of question causes corresponding to the text information based on the multi-hop reasoning algorithm and the retrieval results.
[0010] Further, the verification of the solution includes: simulating the business scenario of the target financial business in a sandbox environment, executing the solution, and recording the execution results; comparing the execution results with the expected goals, and if the execution results are the same as the expected goals, the solution is deemed to have passed verification.
[0011] Furthermore, after returning the validated solution to the target object, the method also includes: obtaining feedback information from the target object regarding the validated solution; and optimizing the preset large language model and preset knowledge graph based on the feedback information to obtain the optimized large language model and optimized knowledge graph.
[0012] To achieve the above objectives, according to another aspect of this application, an information processing apparatus is provided. The apparatus includes: a receiving unit for receiving text information input by a target object and a target file associated with the text information, wherein the text information describes a problem related to a target financial business; a first processing unit for performing a fusion analysis on the text information and the target file using a preset large language model to determine the problem type of the text information; a first determining unit for determining a solution corresponding to the text information based on a preset knowledge graph, the text information, and the problem type; and a second processing unit for verifying the solution and returning the verified solution to the target object.
[0013] Furthermore, when the target file includes log files and business data files, the first processing unit includes: a first processing subunit, used to extract features from the text information, log files, and business data files respectively using a preset large language model to obtain text feature vectors corresponding to the text information, log feature vectors corresponding to the log files, and data feature vectors corresponding to the business data files; a second processing subunit, used to perform weighted fusion processing on the text feature vectors, log feature vectors, and data feature vectors to obtain a fused feature vector; and a third processing subunit, used to classify the fused feature vector using a classifier to obtain the question type of the text information.
[0014] Furthermore, the first processing subunit includes: a first processing module, used to perform word segmentation on text information through the natural language processing module of the preset large language model to obtain a word segmentation set, and to perform dependency parsing based on the word segmentation set to obtain dependency parsing information, and to perform vectorization processing on the dependency parsing information to obtain a text feature vector; a second processing module, used to match log files according to preset regular expressions through the log parsing module of the preset large language model to obtain target event information, and to perform vectorization processing on the target event information to obtain a log feature vector; and a third processing module, used to perform data verification on business data files through the business data processing module of the preset large language model to determine the data quality information of the business data files, and to perform vectorization processing on the data quality information to obtain a data feature vector.
[0015] Furthermore, the first determining unit includes: a first determining subunit, used to convert text information into a graph query statement, and determine the set of problem causes corresponding to the text information based on a preset knowledge graph, the graph query statement, and the problem type; and a second determining subunit, used to determine the solution corresponding to the text information based on the set of problem causes.
[0016] Furthermore, the first determining subunit includes: a first determining module, used to execute a graph query statement to retrieve entities, attributes, and relationships associated with the question type from a preset knowledge graph and obtain the retrieval results; and a second determining module, used to determine the set of question causes corresponding to the text information based on a multi-hop reasoning algorithm and the retrieval results.
[0017] Furthermore, the second processing unit includes: a fourth processing subunit, used to simulate the business scenario of the target financial business in the sandbox environment, execute the solution, and record the execution results; and a fifth processing subunit, used to compare the execution results with the expected target, and if the execution results are the same as the expected target, then the solution is determined to have passed the verification.
[0018] Furthermore, the device also includes: an acquisition unit, used to acquire feedback information from the target object regarding the verified solution after returning the verified solution to the target object; and a third processing unit, used to optimize the preset large language model and preset knowledge graph based on the feedback information to obtain the optimized large language model and optimized knowledge graph.
[0019] According to another aspect of the present invention, an electronic device is also provided, comprising: a memory storing an executable program; and a processor for running the program, wherein the program executes the information processing method described above during runtime.
[0020] According to another aspect of the present invention, a computer-readable storage medium is also provided, wherein the storage medium stores a program, wherein the program controls the device where the storage medium is located to execute any of the above-described information processing methods during runtime.
[0021] In this embodiment, the following steps are employed: receiving text information input by the target object and a target file associated with the text information, wherein the text information describes the problem information of the target financial business; performing fusion analysis on the text information and the target file using a preset large language model to determine the problem type of the text information; determining the corresponding solution based on a preset knowledge graph, the text information, and the problem type; verifying the solution, and returning the verified solution to the target object. This solves the technical problem of low accuracy in problem handling when using a preset frequently asked questions library to process customer service issues in related technologies.
[0022] In this solution, by integrating multimodal information processing, we can more comprehensively understand the problem context and professional semantics, accurately locate the root cause of the problem, and generate logically sound and operable solutions. The pre-verification is completed through an automated mechanism, which improves the efficiency and accuracy of problem handling. The introduction of knowledge graph enhances the system's logical reasoning ability, making the solution not only targeted but also providing a clear explanation of business logic, thereby increasing user trust. Attached Figure Description
[0023] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:
[0024] Figure 1 A hardware structure block diagram of a computer terminal for implementing an information processing method is shown.
[0025] Figure 2 This is a flowchart of an information processing method provided according to an embodiment of this application;
[0026] Figure 3 This is a schematic diagram of the workflow of a preset large language model provided according to an embodiment of this application;
[0027] Figure 4 This is a schematic diagram of the feature extraction process provided according to an embodiment of this application;
[0028] Figure 5 This is a schematic diagram of the inference layer workflow provided according to an embodiment of this application;
[0029] Figure 6 This is a schematic diagram of an information processing apparatus provided according to an embodiment of this application;
[0030] Figure 7 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation
[0031] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.
[0032] It should be noted that the terms "first," "second," etc., 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 orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0033] It should be noted that the information collected in this application (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) are information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of this data all comply with relevant laws, regulations, and standards, necessary confidentiality measures have been taken, and they do not violate public order and good morals. Corresponding access points are provided for users to choose to authorize or refuse. For example, interfaces are set up between this system and relevant users or organizations, providing users with corresponding access points to choose to agree to or refuse automated decision-making results; if the user chooses to refuse, the process proceeds to the expert decision-making stage.
[0034] Example 1
[0035] According to an embodiment of this application, an embodiment of an information processing method is also provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.
[0036] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing an information processing method is shown. Figure 1As shown, the computer terminal 10 (or mobile device) may include one or more processors 102 (shown as 102a, 102b, ..., 102n in the figure) 102 (processor 102 may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.
[0037] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10 (or mobile device). As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).
[0038] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the information processing method in this embodiment. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the above-mentioned information processing method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.
[0039] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.
[0040] The display may be a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10 (or mobile device).
[0041] Under the aforementioned operating environment, this application provides the following: Figure 2 The information processing method shown. Figure 2 This is a flowchart of an information processing method according to Embodiment 1 of this application. The information processing method includes:
[0042] Step S201: Receive text information input by the target object and the target file associated with the text information, wherein the text information is used to describe the problem information of the target financial business.
[0043] Optionally, the information processing system can serve as an intelligent customer service system for target financial transactions (such as direct debit and clearing services). The system adopts a layered architecture design, achieving intelligent processing throughout the entire process from user access to problem resolution through the collaborative work of the interaction layer, understanding layer, reasoning layer, and verification layer. The interaction layer is the direct interface between the user and the system, supporting multiple methods such as text input, file upload, and application programming interface (API) calls. The web version provides an intelligent dialogue window, supporting natural language input and log file attachment uploads, while the mobile version offers voice-to-text functionality to simplify the operation process.
[0044] Optionally, the system can accept various input methods such as natural language descriptions and technical log files, and achieve unified understanding and association of cross-modal information through fusion modeling. For example, a user might ask the intelligent customer service system a question (i.e., the text information input by the target user) through an interaction layer (such as an intelligent dialogue window on a webpage): "Recently, a batch of cross-border direct debit transactions showed that the clearing results failed, but the specific reason is unknown." The user also submits the relevant transaction log files (i.e., the target file associated with the text information). The system can comprehensively identify business logic and technical parameters to pinpoint potential root causes.
[0045] Step S202: The text information and target file are fused and analyzed using a preset large language model to determine the problem type of the text information.
[0046] Optionally, the preset large language model is a pre-trained model fine-tuned based on financial corpus, which can accurately identify financial terminology.
[0047] Optionally, in the information processing method provided in this application embodiment, when the target file includes a log file and a business data file, the text information and the target file are fused and analyzed using a preset large language model to determine the problem type of the text information. This includes: extracting features from the text information, log file, and business data file respectively using the preset large language model to obtain the text feature vector corresponding to the text information, the log feature vector corresponding to the log file, and the data feature vector corresponding to the business data file; performing weighted fusion processing on the text feature vector, log feature vector, and data feature vector to obtain a fused feature vector; and using a classifier to classify the fused feature vector to obtain the problem type of the text information.
[0048] In an optional embodiment, the system understanding layer is used for multimodal information processing. Figure 3 This is a schematic diagram of the workflow of the preset large language model provided in the embodiments of this application, such as... Figure 3 As shown, firstly, features are extracted from text information, log files, and business data files using a pre-defined large language model to obtain text feature vectors corresponding to text information, log feature vectors corresponding to log files, and data feature vectors corresponding to business data files. Then, the text feature vectors, log feature vectors, and data feature vectors are weighted and fused to obtain fused feature vectors. Finally, a classifier is used to classify the fused feature vectors to obtain the question type of the text information.
[0049] Optionally, for multimodal feature fusion, a hierarchical processing and attention mechanism strategy is adopted, and the feature extraction formula is expressed as follows:
[0050]
[0051]
[0052]
[0053] in, E t For text features, E l For log features, E d For data characteristics. LM () represents the natural language processing module. CNN () indicates the log parsing module. f () indicates the business data processing module.
[0054] The formula for calculating attention weights is as follows:
[0055]
[0056] in, i The value is t , l , d . softmax express softmax The function, tanh represents the activation function, W∈ k ,U∈ k×d W and U are weight matrices.
[0057] The feature fusion formula is as follows:
[0058]
[0059] in, E fused This represents the fused feature vector.
[0060] The problem classification formula is as follows:
[0061]
[0062] in, Wc This represents the weight matrix of the classification layer. b c This represents the bias vector.
[0063] By fusing and analyzing text information and target files using a pre-defined large language model, the problem type can be accurately determined.
[0064] Optionally, in the information processing method provided in this application embodiment, feature extraction is performed on text information, log files, and business data files respectively using a preset large language model to obtain text feature vectors corresponding to text information, log feature vectors corresponding to log files, and data feature vectors corresponding to business data files. This includes: performing word segmentation on text information using the natural language processing module of the preset large language model to obtain a word segmentation set, performing dependency parsing based on the word segmentation set to obtain dependency parsing information, and vectorizing the dependency parsing information to obtain text feature vectors; matching log files using the log parsing module of the preset large language model based on preset regular expressions to obtain target event information, and vectorizing the target event information to obtain log feature vectors; and performing data verification on business data files using the business data processing module of the preset large language model to determine the data quality information of the business data files, and vectorizing the data quality information to obtain data feature vectors.
[0065] In an optional embodiment, the natural language processing (NLP) module can accurately identify financial terminology. By segmenting text information, it can obtain a segmentation set. For example, the NLP module might identify key information related to cross-border payments or clearing failures. Through dependency parsing, it can understand the problem's logic. For instance, it can parse "Why did the transaction happen, but the query interface still returns no data found?" into a query for "The query interface returned an exception," thus obtaining dependency parsing information. This dependency parsing information is then vectorized to obtain a text feature vector. The log parsing module can match log files using preset regular expressions to obtain target event information. For example, it can extract key information (i.e., target event information) such as error codes, transaction times, and transaction IDs from uploaded log files, and then vectorize this information to obtain a log feature vector. The business data processing module can perform data verification on business data files using data validation rules (such as loan balance checks and field integrity checks) to determine the data quality information of the business data files. For example, it can generate a data quality report, which is then vectorized to obtain a data feature vector.
[0066] In an alternative embodiment, Figure 4 This is a schematic diagram of the feature extraction process provided according to the embodiments of this application, such as... Figure 4 As shown, for natural language descriptive text, the natural language processing module performs steps such as word segmentation, named entity recognition, dependency parsing, and technical terminology recognition to accurately capture key points of the problem, such as the transaction type, error code, and business status involved. For log data, the log parsing module performs steps such as format recognition, regular expression matching, key information extraction, and anomaly location to determine log characteristics; for business data, the business data processing module performs steps such as data verification, anomaly detection, and quality report generation to determine data characteristics.
[0067] By using a pre-defined large language model to extract features from text information, log files, and business data files, the accurate determination of text feature vectors, log feature vectors, and data feature vectors was achieved, ensuring the comprehensiveness and depth of information processing.
[0068] Step S203: Based on the preset knowledge graph, text information, and question type, determine the solution corresponding to the text information.
[0069] Optionally, the inference layer integrates a knowledge graph and an inference engine to achieve intelligent decision-making. The knowledge graph is based on triples of entities, relations, and attributes. Entities include error codes (such as 400 authentication failure), interface types (such as quick payment interfaces), clearing rules (such as cross-border settlement rules), and institutional information (such as a bank's settlement cycle being T+1 days); relations include "caused" (such as an error code causing transaction failure), "dependent" (such as cross-border settlement rules depending on fee calculation rules), and "belongs to" (such as an error code belonging to an interface exception category).
[0070] Optionally, in the information processing method provided in this application embodiment, determining the solution corresponding to the text information based on the preset knowledge graph, text information, and question type includes: converting the text information into a graph query statement; determining the set of problem causes corresponding to the text information based on the preset knowledge graph, graph query statement, and question type; and determining the solution corresponding to the text information based on the set of problem causes.
[0071] In an optional embodiment, the inference engine converts text information into graph query statements. For example, "revenue sharing failed and the log shows a 500 error" is converted into "find entities related to revenue sharing transactions with a failed status and associated error code 500". Then, based on a preset knowledge graph, graph query statements, and question types, the engine determines the set of causes corresponding to the text information, and then determines the corresponding solution based on the set of causes. For example, by traversing the knowledge graph through multi-hop inference, possible causes of the problem are found and associated, such as interface status, rule matching results, account balance, etc., forming a set of causes. Combined with the semantic generation capabilities of a large language model, the technical graph query results are transformed into solutions that users can understand. For example, the solution includes operational steps (such as checking whether the authentication token used when calling the interface has expired) and business logic explanations (such as an expired token will cause the server to refuse access, and a valid token needs to be obtained again), increasing user trust in the solution.
[0072] In an alternative embodiment, Figure 5 This is a schematic diagram of the inference layer workflow provided according to an embodiment of this application, such as... Figure 5 As shown, the text information is first converted into a graph query statement. By traversing the knowledge graph and performing multi-hop reasoning, candidate causes (i.e., a set of problem causes) are generated. Then, semantic generation is performed to transform the technical graph query results into solutions that users can understand.
[0073] Through graph querying and multi-hop reasoning, the system can identify the complex causal relationships behind a problem, thereby accurately determining the solution.
[0074] Optionally, in the information processing method provided in this application embodiment, determining the set of problem causes corresponding to text information based on a preset knowledge graph, a graph query statement, and a problem type includes: executing a graph query statement to retrieve entities, attributes, and relationships associated with the problem type from the preset knowledge graph to obtain retrieval results; and determining the set of problem causes corresponding to text information based on a multi-hop reasoning algorithm and the retrieval results.
[0075] In an optional embodiment, a graph query statement is executed to retrieve entities, attributes, and relationships associated with the question type from a preset knowledge graph, obtaining the retrieval results. Then, based on a multi-hop inference algorithm and the retrieval results, the set of causes corresponding to the text information is determined. For example, through multi-hop inference, it is found that error code 500 is related to interface timeout, further linking it to the stability issue of the cross-border payment interface. Simultaneously, the knowledge graph shows that the bank's international settlement system underwent upgrades and maintenance during these transactions, potentially affecting the normal deduction and clearing process. The large language model generation solution involves checking the connection settings of the cross-border payment interface to confirm whether timeouts were caused by configuration errors during upgrades and maintenance; simultaneously, querying the maintenance logs of the international settlement system to verify whether service interruptions led to transaction failures.
[0076] By using knowledge graph queries and reasoning, we were able to deeply explore the causes of problems, accurately identify the multiple sources of problems, and improve the comprehensiveness and accuracy of problem diagnosis.
[0077] Step S204: Verify the solution and return the verified solution to the target object.
[0078] Optionally, the verification layer ensures the feasibility of the solution through a sandbox environment and returns the verified feasible solution to the user.
[0079] Optionally, in the information processing method provided in this application embodiment, verifying the solution includes: simulating the business scenario of the target financial business in a sandbox environment, executing the solution, and recording the execution results; comparing the execution results with the expected target, and if the execution results are the same as the expected target, then the solution is determined to have passed verification.
[0080] In an optional embodiment, a sandbox environment is built based on containerization technology to achieve secure isolation from the production environment. Each test task is assigned an independent container, resource usage is limited, and test data is anonymized to avoid impacting real business operations. The sandbox environment supports simulation of typical business scenarios, including normal clearing, abnormal transaction rollback, and cross-border settlement delays. Fault conditions such as network latency and interface timeouts can be customized to comprehensively verify the robustness of the solution. After the solution is executed, the expected results are compared with the actual output. For example, it verifies whether the revenue sharing amount calculation complies with the rules and whether the interface call returns the correct status code, generating a detailed report including pass rate and anomaly indicators to ensure the reliability of the proposed solution in the production environment.
[0081] By validating the solution, we ensure that its feasibility and stability have been fully tested and verified before it is delivered to the target audience. This greatly reduces the risk of failure when implementing the solution in a production environment and improves the efficiency of problem solving and user satisfaction.
[0082] In summary, by integrating multimodal information processing, we can gain a more comprehensive understanding of the problem context and professional semantics, accurately pinpoint the root cause of the problem, and generate logically sound and actionable solutions. The automated mechanism completes the pre-verification, improving the efficiency and accuracy of problem handling. The introduction of knowledge graphs enhances the system's logical reasoning ability, making the solutions not only targeted but also providing clear explanations of business logic, thereby increasing user trust.
[0083] Optionally, in the information processing method provided in the embodiments of this application, after returning the verified solution to the target object, the method further includes: obtaining feedback information from the target object regarding the verified solution; and optimizing the preset large language model and the preset knowledge graph based on the feedback information to obtain the optimized large language model and the optimized knowledge graph.
[0084] In an optional embodiment, the system feeds back the validated solution to the user, while collecting user feedback and system operation data for incremental updates of the knowledge graph and continuous training of the model, forming an automated learning and optimization loop to continuously improve the professional response capabilities and processing efficiency of the intelligent customer service system.
[0085] The information processing method provided in this application, by integrating multimodal information processing, can more comprehensively understand the problem context and professional semantics, accurately locate the root cause of the problem, and thus generate a logically reasonable and operable solution. The pre-verification is completed through an automated mechanism, which improves the efficiency and accuracy of problem processing. The introduction of knowledge graph enhances the logical reasoning ability of the system, making the solution not only targeted, but also providing a clear explanation of business logic, thereby increasing user trust.
[0086] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.
[0087] Example 2
[0088] This application also provides an information processing apparatus. It should be noted that the information processing apparatus of this application can be used to execute the information processing method provided in this application. The information processing apparatus provided in this application will be described below.
[0089] According to embodiments of this application, an information processing apparatus for implementing the above-described information processing method is also provided, such as... Figure 6 As shown, the device includes: a receiving unit 601, a first processing unit 602, a first determining unit 603, and a second processing unit 604.
[0090] The receiving unit 601 is used to receive text information input by the target object and the target file associated with the text information, wherein the text information is used to describe the problem information of the target financial business;
[0091] The first processing unit 602 is used to perform fusion analysis on text information and target files through a preset large language model to determine the problem type of the text information;
[0092] The first determining unit 603 is used to determine the solution corresponding to the text information based on the preset knowledge graph, text information and question type;
[0093] The second processing unit 604 is used to verify the solution and return the verified solution to the target object.
[0094] The information processing apparatus provided in this application embodiment receives text information input by a target object and a target file associated with the text information through a receiving unit 601. The text information describes a problem in a target financial business. A first processing unit 602 performs fusion analysis on the text information and the target file using a preset large language model to determine the problem type of the text information. A first determining unit 603 determines the corresponding solution based on a preset knowledge graph, the text information, and the problem type. A second processing unit 604 verifies the solution and returns the verified solution to the target object. By integrating multimodal information processing, the system can more comprehensively understand the problem context and professional semantics, accurately locate the root cause of the problem, and generate logically sound and operable solutions. The automated pre-verification mechanism improves the efficiency and accuracy of problem processing. The introduction of the knowledge graph enhances the system's logical reasoning ability, making the solution not only targeted but also providing a clear explanation of business logic, thus increasing user trust.
[0095] Optionally, in the information processing apparatus provided in this application embodiment, when the target file includes a log file and a business data file, the first processing unit includes: a first processing subunit, used to extract features from the text information, log file, and business data file respectively using a preset large language model to obtain a text feature vector corresponding to the text information, a log feature vector corresponding to the log file, and a data feature vector corresponding to the business data file; a second processing subunit, used to perform weighted fusion processing on the text feature vector, log feature vector, and data feature vector to obtain a fused feature vector; and a third processing subunit, used to classify the fused feature vector using a classifier to obtain the question type of the text information.
[0096] Optionally, in the information processing apparatus provided in this application embodiment, the first processing subunit includes: a first processing module, used to perform word segmentation processing on text information through a natural language processing module of a preset large language model to obtain a word segmentation set, and perform dependency parsing analysis based on the word segmentation set to obtain dependency parsing information, and perform vectorization processing on the dependency parsing information to obtain a text feature vector; a second processing module, used to perform matching on log files based on a preset regular expression through a log parsing module of a preset large language model to obtain target event information, and perform vectorization processing on the target event information to obtain a log feature vector; and a third processing module, used to perform data verification on business data files through a business data processing module of a preset large language model to determine the data quality information of the business data files, and perform vectorization processing on the data quality information to obtain a data feature vector.
[0097] Optionally, in the information processing apparatus provided in this application embodiment, the first determining unit includes: a first determining subunit, used to convert text information into a graph query statement, and determine the set of problem causes corresponding to the text information based on a preset knowledge graph, the graph query statement and the problem type; and a second determining subunit, used to determine the solution corresponding to the text information based on the set of problem causes.
[0098] Optionally, in the information processing apparatus provided in this application embodiment, the first determining subunit includes: a first determining module, used to execute a graph query statement to retrieve entities, attributes and relationships associated with the question type from a preset knowledge graph and obtain retrieval results; and a second determining module, used to determine the set of question causes corresponding to the text information based on a multi-hop reasoning algorithm and the retrieval results.
[0099] Optionally, in the information processing apparatus provided in this application embodiment, the second processing unit includes: a fourth processing subunit, used to simulate the business scenario of the target financial business in a sandbox environment, execute the solution, and record the execution result; and a fifth processing subunit, used to compare the execution result with the expected target, and if the execution result is the same as the expected target, then the solution is determined to have passed verification.
[0100] Optionally, in the information processing apparatus provided in the embodiments of this application, the apparatus further includes: an acquisition unit, configured to acquire feedback information from the target object regarding the verified solution after returning the verified solution to the target object; and a third processing unit, configured to optimize the preset large language model and the preset knowledge graph based on the feedback information to obtain the optimized large language model and the optimized knowledge graph.
[0101] It should be noted that the receiving unit 601, the first processing unit 602, the first determining unit 603, and the second processing unit 604 mentioned above correspond to steps S201 to S204 in Embodiment 1. The four units and the corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in Embodiment 1. It should be noted that the above modules or units can be hardware or software components stored in memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n). The above units can also be part of a device and run in the computer terminal 10 provided in Embodiment 1.
[0102] Example 3
[0103] Embodiments of this application may provide an electronic device. Figure 7 This is a structural block diagram of an electronic device according to an embodiment of this application. Figure 7 As shown, the electronic device may include: one or more ( Figure 7 (Only one is shown) processor 702, memory 704, memory controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.
[0104] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the methods and apparatus in the embodiments of this application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby implementing the above-described methods. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.
[0105] The processor can invoke information and applications stored in the memory through the transmission device to perform the following steps: receiving text information input by the target object and the target file associated with the text information, wherein the text information is used to describe the problem information of the target financial business; performing fusion analysis on the text information and the target file through a preset large language model to determine the problem type of the text information; determining the solution corresponding to the text information based on the preset knowledge graph, the text information and the problem type; verifying the solution and returning the verified solution to the target object.
[0106] The processor can access the information and application programs stored in the memory via the transmission device to perform the following steps: extracting features from text information, log files, and business data files using a preset large language model to obtain text feature vectors corresponding to the text information, log feature vectors corresponding to the log files, and data feature vectors corresponding to the business data files; performing weighted fusion processing on the text feature vectors, log feature vectors, and data feature vectors to obtain fused feature vectors; and using a classifier to classify the fused feature vectors to obtain the question type of the text information.
[0107] The processor can access information and applications stored in the memory via a transmission device to execute the following steps: First, the text information is segmented using the natural language processing module of the preset large language model to obtain a segmentation set. Then, dependency parsing is performed on the segmentation set to obtain dependency parsing information. This dependency parsing information is then vectorized to obtain a text feature vector. Second, the log parsing module of the preset large language model matches log files using preset regular expressions to obtain target event information. This target event information is then vectorized to obtain a log feature vector. Third, the business data processing module of the preset large language model performs data verification on business data files to determine the data quality information of the business data files. This data quality information is then vectorized to obtain a data feature vector.
[0108] The processor can access information and applications stored in the memory via a transmission device to perform the following steps: converting text information into a graph query statement; determining the set of causes corresponding to the text information based on a preset knowledge graph, the graph query statement, and the question type; and determining the solution corresponding to the text information based on the set of causes.
[0109] The processor can access the information and application programs stored in the memory via the transmission device to perform the following steps: execute a graph query statement to retrieve entities, attributes, and relationships associated with the question type from a preset knowledge graph and obtain the retrieval results; and determine the set of question causes corresponding to the text information based on the multi-hop reasoning algorithm and the retrieval results.
[0110] The processor can access information and applications stored in memory via a transmission device to perform the following steps: simulate the business scenario of the target financial business in a sandbox environment, execute the solution, and record the execution results; compare the execution results with the expected goals, and if the execution results are the same as the expected goals, then the solution is confirmed to have passed verification.
[0111] The processor can access the information and application stored in the memory via the transmission device to perform the following steps: after returning the verified solution to the target object, obtain the target object's feedback information on the verified solution; optimize the preset large language model and preset knowledge graph based on the feedback information to obtain the optimized large language model and optimized knowledge graph.
[0112] Those skilled in the art will understand that Figure 7 The structure shown is for illustrative purposes only. Electronic devices can also be smartphones, tablets, handheld computers, mobile internet devices (MIDs), PADs, and other terminal devices. Figure 7 This does not limit the structure of the aforementioned electronic device. For example, electronic devices may also include components that are more... Figure 7 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 7 The different configurations shown.
[0113] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.
[0114] Example 4
[0115] Embodiments of this application also provide a computer-readable storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the information processing method provided in Embodiment 1.
[0116] Optionally, in this embodiment, the storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.
[0117] This application also provides a computer program product that, when executed on a data processing device, is suitable for performing steps of an information processing method.
[0118] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.
[0119] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.
[0120] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.
[0121] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0122] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0123] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.
[0124] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.
Claims
1. An information processing method, characterized in that, include: Receive text information input by the target object and the target file associated with the text information, wherein the text information is used to describe the problem information of the target financial business; By fusing and analyzing the text information and the target file using a pre-defined large language model, the problem type of the text information is determined. Based on the preset knowledge graph, the text information, and the question type, the solution corresponding to the text information is determined; The solution is verified, and the verified solution is returned to the target object.
2. The method according to claim 1, characterized in that, When the target file includes log files and business data files, the text information and the target file are fused and analyzed using a preset large language model to determine the problem types of the text information, including: The preset large language model is used to extract features from the text information, the log file, and the business data file respectively, to obtain the text feature vector corresponding to the text information, the log feature vector corresponding to the log file, and the data feature vector corresponding to the business data file. The text feature vector, the log feature vector, and the data feature vector are weighted and fused to obtain a fused feature vector. A classifier is used to classify the fused feature vector to obtain the question type of the text information.
3. The method according to claim 2, characterized in that, The preset large language model is used to extract features from the text information, the log file, and the business data file respectively, to obtain the text feature vector corresponding to the text information, the log feature vector corresponding to the log file, and the data feature vector corresponding to the business data file, including: The text information is segmented by the natural language processing module of the preset large language model to obtain a segmentation set. Dependency parsing is then performed on the segmentation set to obtain dependency parsing information. The dependency parsing information is then vectorized to obtain the text feature vector. The log parsing module of the preset large language model matches the log file according to the preset regular expression to obtain the target event information, and then performs vectorization processing on the target event information to obtain the log feature vector. The business data file is validated by the business data processing module of the preset large language model to determine the data quality information of the business data file, and the data quality information is vectorized to obtain the data feature vector.
4. The method according to claim 1, characterized in that, Based on a preset knowledge graph, the text information, and the question type, the solution corresponding to the text information includes: The text information is converted into a graph query statement, and the set of problem causes corresponding to the text information is determined based on the preset knowledge graph, the graph query statement, and the problem type. Based on the set of causes of the problem, determine the solution corresponding to the text information.
5. The method according to claim 4, characterized in that, Based on the preset knowledge graph, the graph query statement, and the question type, the set of question reasons corresponding to the text information includes: Execute the graph query statement to retrieve entities, attributes, and relationships associated with the question type from the preset knowledge graph, and obtain the search results; Based on the multi-hop reasoning algorithm and the retrieval results, the set of problem causes corresponding to the text information is determined.
6. The method according to claim 1, characterized in that, Validating the solution includes: Simulate the business scenario of the target financial business in a sandbox environment, execute the solution, and record the execution results; The execution result is compared with the expected goal. If the execution result is the same as the expected goal, the solution is determined to have passed the verification.
7. The method according to claim 1, characterized in that, After returning the validated solution to the target object, the method further includes: Obtain feedback information from the target object regarding the verified solution; Based on the feedback information, the preset large language model and the preset knowledge graph are optimized to obtain the optimized large language model and the optimized knowledge graph.
8. An information processing device, characterized in that, include: A receiving unit is configured to receive text information input by a target object and a target file associated with the text information, wherein the text information is used to describe the problem information of the target financial business; The first processing unit is used to perform fusion analysis on the text information and the target file through a preset large language model to determine the problem type of the text information; The first determining unit is used to determine the solution corresponding to the text information based on the preset knowledge graph, the text information, and the question type; The second processing unit is used to verify the solution and return the verified solution to the target object.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device on which the computer-readable storage medium is located to perform the information processing method according to any one of claims 1 to 7.
10. An electronic device, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the information processing method according to any one of claims 1 to 7.