Question diagnosis method and device for financial system

Abstract

The invention provides a problem diagnosis method and device for a financial system, which can be used in the financial field or other fields, and the method comprises the steps: receiving a problem diagnosis text of a target financial system transmitted from a front end by a user; performing similarity retrieval by applying a pre-constructed knowledge graph and the problem diagnosis text, determining a diagnosis result corresponding to the problem diagnosis text, and outputting the diagnosis result; wherein the nodes in the pre-constructed knowledge graph represent key entities, and the connection between the two nodes represents the relationship between the key entities. According to the method, the accuracy of financial system problem diagnosis can be improved on the basis of ensuring the efficiency of financial system problem diagnosis.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and device for diagnosing problems in a financial system. Background Art

[0002] Diagnosing problems in the financial system is a key step in ensuring the stability and reliability of financial services. Common problem diagnosis methods include:

[0003] Application stress testing tools: such as JMeter and Gatling. These tools simulate a large number of user accesses to test the performance of the system under high load. However, they usually only support structured test data and have limited processing capabilities for unstructured data. Application log analysis platforms: such as the ELK stack. These platforms can collect and analyze system logs, but they mainly provide keyword searches and simple statistical analysis, making it difficult to extract in-depth knowledge from large amounts of unstructured logs. Application performance management tools: such as skywalking and pinpoint. These tools can monitor the performance indicators of applications, but lack the ability to integrate and correlate multi-source data. Rule-based alert system: Alerts are issued for performance anomalies through preset thresholds and rules. However, fixed rules are difficult to adapt to complex and changing financial business scenarios and are prone to false positives or omissions. Summary of the Invention

[0004] In response to at least one problem in the prior art, the present application proposes a method and apparatus for diagnosing problems in a financial system, which can improve the accuracy of diagnosing problems in the financial system while ensuring the efficiency of the diagnosis.

[0005] In order to solve the above technical problems, this application provides the following technical solutions:

[0006] In a first aspect, the present application provides a method for diagnosing problems in a financial system, comprising:

[0007] Receive problem diagnosis text of the target financial system from the user's front-end;

[0008] Apply the pre-built knowledge graph and the problem diagnosis text to perform similarity search, determine the diagnosis result corresponding to the problem diagnosis text and output it;

[0009] Among them, the nodes in the pre-built knowledge graph represent key entities, and the connection between two nodes represents the relationship between key entities.

[0010] In one embodiment, the application performs similarity retrieval between the pre-built knowledge graph and the problem diagnosis text, determines and outputs the diagnosis result corresponding to the problem diagnosis text, including:

[0011] Segmenting the problem diagnosis text to obtain multiple text blocks;

[0012] Apply the embedding generation algorithm to generate text embeddings corresponding to each text block;

[0013] The pre-built knowledge graph and the text embedding corresponding to each text block are applied to perform similarity retrieval, and the diagnosis result corresponding to the problem diagnosis text is determined and output.

[0014] In one embodiment, before applying the pre-built knowledge graph and the problem diagnosis text to perform similarity retrieval, the method further includes:

[0015] Collect multi-source unstructured data from the target financial system, including transaction logs, system logs, user feedback data, external documents, and market data;

[0016] Applying the multi-source unstructured data, the pre-trained large-scale language model, and the relation extraction algorithm to obtain key entities and the relationships between the key entities;

[0017] Based on the key entities and the relationships between the key entities, the knowledge graph is constructed.

[0018] In one embodiment, applying the multi-source unstructured data, the pre-trained large-scale language model, and the relation extraction algorithm to obtain key entities and the relationships between key entities includes:

[0019] Applying the multi-source unstructured data and the pre-trained large-scale language model to obtain key entities;

[0020] The relationships between the key entities are obtained based on the multi-source unstructured data and the relationship extraction algorithm.

[0021] In a second aspect, the present application provides a problem diagnosis device for a financial system, comprising:

[0022] The receiving module is used to receive the problem diagnosis text input by the user from the front end;

[0023] A diagnosis module is used to apply a pre-built knowledge graph and the problem diagnosis text to perform similarity search, determine the diagnosis result corresponding to the problem diagnosis text, and output it;

[0024] Among them, the nodes in the pre-built knowledge graph represent key entities, and the connection between two nodes represents the relationship between key entities.

[0025] In one embodiment, the diagnostic module includes:

[0026] A segmentation unit, configured to segment the problem diagnosis text into multiple text blocks;

[0027] A text embedding unit, which is used to apply an embedding generation algorithm to generate text embeddings corresponding to each text block;

[0028] The diagnosis unit is used to apply the pre-built knowledge graph and the text embedding corresponding to each text block to perform similarity retrieval, determine the diagnosis result corresponding to the problem diagnosis text, and output it.

[0029] In one embodiment, the problem diagnosis device for a financial system further includes:

[0030] A collection module, configured to collect multi-source unstructured data from the target financial system, including transaction logs, system logs, user feedback data, external documents, and market data;

[0031] An extraction module, configured to apply the multi-source unstructured data, a pre-trained large-scale language model, and a relation extraction algorithm to obtain key entities and the relationships between them;

[0032] A construction module is used to construct the knowledge graph based on the key entities and the relationships between the key entities.

[0033] In one embodiment, the extraction module includes:

[0034] An application unit, configured to apply the multi-source unstructured data and the pre-trained large-scale language model to obtain key entities;

[0035] An extraction unit is used to obtain the relationship between the key entities based on the multi-source unstructured data and a relationship extraction algorithm.

[0036] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for diagnosing problems in a financial system when executing the program.

[0037] In a fourth aspect, the present application provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the method for diagnosing problems in a financial system.

[0038] As can be seen from the above technical solution, the present application provides a method and apparatus for diagnosing problems in a financial system. The method comprises: receiving problem diagnosis text for a target financial system from a user input from a front-end; performing a similarity search between a pre-constructed knowledge graph and the problem diagnosis text, determining and outputting a diagnosis result corresponding to the problem diagnosis text; wherein nodes in the pre-constructed knowledge graph represent key entities, and connections between two nodes represent relationships between key entities. This method can improve the accuracy of diagnosing problems in the financial system while maintaining the efficiency of the problem diagnosis. Specifically, it can enhance multi-source data integration and analysis capabilities, effectively integrating structured and unstructured multi-source heterogeneous data, comprehensively capturing various factors affecting system performance, and improving data utilization. It can also enhance unstructured data processing capabilities. By utilizing large-scale language models and natural language processing technologies, it can achieve in-depth understanding and knowledge extraction of unstructured data, thus addressing the shortcomings of existing technologies. It can also improve testing efficiency and accuracy. Through automated knowledge extraction and intelligent analysis, it reduces reliance on manual experience, increases testing efficiency by approximately 40%, and achieves an accuracy rate of approximately 92% for detecting performance bottlenecks. It can realize knowledge accumulation and sharing, build a knowledge map for financial system performance testing, systematically accumulate and manage testing experience and knowledge, avoid duplication of work, and promote knowledge sharing. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0040] Figure 1 This is a first flow chart of the problem diagnosis method for a financial system in an embodiment of the present application;

[0041] Figure 2 This is a second flow chart of the problem diagnosis method for a financial system in an embodiment of the present application;

[0042] Figure 3 This is a third flow chart of the problem diagnosis method for a financial system in an embodiment of the present application;

[0043] Figure 4 This is a schematic diagram of a knowledge graph used as an example in this application;

[0044] Figure 5 This is a first structural diagram of a problem diagnosis device for a financial system in an embodiment of the present application;

[0045] Figure 6This is a second structural diagram of the problem diagnosis device for the financial system in an embodiment of the present application;

[0046] Figure 7 It is a schematic block diagram of the system structure of the electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0047] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0048] Existing methods for diagnosing problems in the financial system lack the ability to integrate multi-source data. They lack effective integration of heterogeneous multi-source data such as transaction data, system logs, user feedback, and external market information, and are unable to fully capture the various factors that affect system performance. Analysis of unstructured data is limited, as a large amount of important information exists in unstructured data, and the ability to process unstructured data such as log text, user comments, and fault reports is limited. They rely on manual experience and have a low degree of automation. Performance testing and problem diagnosis mainly rely on the experience of testers, making it difficult to detect potential performance problems in a timely manner. The lack of intelligent analysis methods leads to highly subjective test and analysis results, making them difficult to apply on a large scale.

[0049] To address the aforementioned challenges of prior art problem diagnosis for financial systems, such as difficulties integrating multi-source data, insufficient processing capabilities for unstructured data, and reliance on manual experience and insufficient knowledge accumulation, the present invention combines multi-source data with a large language model to provide a method and apparatus for diagnosing financial system problems. These methods integrate multi-source heterogeneous data and utilize large language models and knowledge graph technology to intelligently analyze and optimize financial system performance, improving the efficiency, accuracy, and automation of performance testing. They also provide efficient graph database support, integrating high-performance graph database technology to ensure efficient storage, query, and management of knowledge graph data and support large-scale data processing. They also offer advanced natural language processing capabilities. Using large language models and natural language processing technology, they can automatically extract, understand, and analyze key knowledge from unstructured data. They enable intelligent analysis and decision support. Leveraging machine learning and reasoning techniques, the system can automatically identify performance bottlenecks, provide optimization suggestions, and enhance the intelligence of testing. Furthermore, they offer a user-friendly interface, including an interactive visualization interface and natural language interaction capabilities, enhancing the user experience and lowering the barrier to entry.

[0050] It should be noted that the method and apparatus for diagnosing problems in financial systems disclosed in this application can be used in the field of financial technology, as well as in any field outside of the financial technology field. The application fields of the method and apparatus for diagnosing problems in financial systems disclosed in this application are not limited. The acquisition, storage, use, and processing of data in the technical solutions of each embodiment of this application comply with relevant laws and regulations.

[0051] The details are described in the following embodiments.

[0052] In order to improve the accuracy of financial system problem diagnosis while ensuring the efficiency of financial system problem diagnosis, this embodiment provides a financial system problem diagnosis method in which the execution subject is a financial system problem diagnosis device, and the financial system problem diagnosis device includes but is not limited to a server, such as Figure 1 As shown, the method specifically includes the following contents:

[0053] Step 100: Receive the problem diagnosis text of the target financial system input by the user from the front end.

[0054] Specifically, the target financial system may be an existing financial transaction system; the problem diagnosis text may be questions raised by the user for diagnosing financial system problems; for example, "the reason why there is a response delay in cash withdrawal transactions during a specific period A", "why there is a response delay in cash withdrawal transactions during a specific period A".

[0055] Step 200: Apply a pre-constructed knowledge graph and the problem diagnosis text to perform similarity retrieval, determine the diagnosis result corresponding to the problem diagnosis text and output it, wherein the nodes in the pre-constructed knowledge graph represent key entities, and the connection between two nodes represents the relationship between key entities.

[0056] Specifically, key entities may include transaction types, service components, performance indicators, and abnormal events, such as "withdrawal transaction" and "server node B." Relationships between key entities may include "initiator" and "influence." Diagnostic results may reflect the causes of financial system performance issues, such as "memory leak" and "external market fluctuations."

[0057] Furthermore, the correspondence between the diagnostic results and the optimization scheme can be pre-stored. After the diagnostic results are determined, the corresponding optimization scheme can be determined based on the correspondence between the diagnostic results and the optimization scheme and the diagnostic results, so as to facilitate timely operation and maintenance of the financial system.

[0058] In order to improve the intelligence level of problem diagnosis and thus improve the reliability of problem diagnosis, such as Figure 2 As shown, in one embodiment, step 200 includes:

[0059] Step 201: Segment the problem diagnosis text to obtain multiple text blocks.

[0060] Specifically, the problem diagnosis text may be segmented by length and semantics to ensure that each text block has complete meaning.

[0061] Step 202: Apply an embedding generation algorithm to generate text embeddings corresponding to each text block.

[0062] Specifically, embedding generation algorithms (such as BERT and Doc2Vec) can be applied to generate text embeddings to capture contextual semantic information and provide support for subsequent similarity retrieval and clustering analysis.

[0063] Step 203: Apply the pre-built knowledge graph and the text embedding corresponding to each text block to perform similarity retrieval, determine the diagnosis result corresponding to the problem diagnosis text, and output it.

[0064] Specifically, when the large model detects an anomaly (such as a delay in "withdrawal transaction"), it can find possible causes based on the knowledge graph. Similarity retrieval: Analyze the similarity between the current problem and historical events, and infer the cause of the current bottleneck (such as the memory problem of "server node B"). Dynamic adjustment: Adjust the priority of the rules according to real-time data to make reasoning more flexible. For example, "Server Node B" and "GC Frequency: 30 times / minute" in the knowledge graph are the nodes closest to the text embedding corresponding to each text block. According to the preset diagnosis result relationship table, it can be determined that the node connected to "GC Frequency: 30 times / minute" and indicating the diagnosis result is a GC anomaly, then it can be determined that the diagnosis result includes: GC anomaly. The preset diagnosis result relationship table can include relationships used to reflect the diagnosis results, such as indication, trigger, cause, etc., which can be set according to actual conditions.

[0065] Specifically, you can also apply pre-built knowledge graphs to detect high-frequency bottleneck paths (such as frequent delays in "withdrawal transactions" on "server node B"). Set thresholds for performance indicators (such as triggering an alarm when the response time exceeds 300 milliseconds). Define the causal relationship between events to help locate the root cause of the bottleneck. Discover recurring patterns based on historical data (such as delays caused by peaks at the beginning of the month) to facilitate the prediction of future bottlenecks. You can also analyze historical peak patterns to provide early warning of upcoming bottlenecks. Propose optimization plans based on the reasoning results (such as increasing server resources, optimizing memory configuration), and present them to users in natural language.

[0066] In order to achieve the reliability of unstructured data processing and improve the reliability of knowledge graphs, such as Figure 3 As shown, in one embodiment, before step 200, the method further includes:

[0067] Step 001: Collect multi-source unstructured data from the target financial system, where the multi-source unstructured data includes transaction logs, system logs, user feedback data, external documents, and market data.

[0068] Specifically, the transaction log may include: transaction ID, user ID, transaction type, transaction time and transaction results, etc.; the system log may include: log time, event type (such as "error"), description (such as "the system cannot connect to the database"), impact module (such as "user authentication service"), etc.; the user feedback data may include: feedback content (such as "the application often crashes when loading"), feedback type (such as "technical support"), etc.; external documents may include: document title, publishing agency, key points (such as "the global economic growth rate is expected to be XX in XXX") and document connection, etc.; market data may include: transaction date, stock code, opening price, highest price and trading volume, etc. When constructing a knowledge graph, applying multi-source unstructured data can improve the comprehensiveness and reliability of the knowledge graph.

[0069] Step 002: Apply the multi-source unstructured data, the pre-trained large-scale language model and the relationship extraction algorithm to obtain key entities and the relationships between key entities.

[0070] Specifically, the multi-source unstructured data can be segmented to obtain multiple historical text blocks; an embedding generation algorithm can be applied to generate historical text embeddings corresponding to each historical text block; and a pre-trained large-scale language model, a relation extraction algorithm, and the historical text embeddings corresponding to each historical text block can be applied to obtain key entities and the relationships between key entities.

[0071] Step 003: Based on the key entities and the relationships between the key entities, the knowledge graph is constructed.

[0072] Specifically, performance-related structured and unstructured data, such as "withdrawal transactions" and "response latency," can be collected. Key entities (such as transaction types and system components) and relationships (such as "trigger" and "dependency") can be extracted using large models. Graph construction: These entities and relationships are imported into a graph database (such as Neo4j) to form a system knowledge graph, providing a foundation for reasoning. This can then guide the system to identify bottlenecks and predict performance issues.

[0073] To improve the reliability of obtaining key entities and relationships between key entities, in one embodiment, step 002 includes:

[0074] Step 021: Apply the multi-source unstructured data and the pre-trained large-scale language model to obtain key entities.

[0075] Step 022: Obtain the relationship between the key entities based on the multi-source unstructured data and the relationship extraction algorithm.

[0076] To further illustrate this solution, this application provides an application example of a problem diagnosis method for a financial system, which is described in detail as follows:

[0077] 1. Data preparation and knowledge extraction stage.

[0078] Step 11: Multi-source data collection. Automatically collect performance-related data about the financial system from multiple data sources, including transaction logs, system logs, user feedback, external documents, and market information, to establish a comprehensive data foundation.

[0079] Specifically, we can use application programming interface (API), log parsing and other technologies to automatically collect data from multiple data sources such as transaction logs, system logs, user feedback, external documents, market conditions, etc., to establish a comprehensive data set.

[0080] Step 12: Entity and Relationship Extraction: Use pre-trained large-scale language models (such as BERT) to process unstructured data, automatically extract key entities in the financial system (such as transaction types, service components, performance indicators, and abnormal events) and their relationships, and build structured knowledge.

[0081] Specifically, pre-trained and fine-tuned large-scale language models (such as BERT) can be used, combined with named entity recognition (NER) and relationship extraction algorithms, to automatically identify key entities such as transaction types, service components, performance indicators, abnormal events, and their relationships.

[0082] Step 13: Data Segmentation and Embedding Generation: Segment the text data based on semantics and length to ensure that each text block has complete meaning. Use embedding generation algorithms (such as BERT and Doc2Vec) to generate text embeddings to capture contextual semantic information and support similarity retrieval and cluster analysis.

[0083] 2. Knowledge graph construction stage.

[0084] Step 14: Knowledge Graph Construction. Import the extracted entities and relationships into the graph database to construct the knowledge graph for the financial system. Define node and edge types, design the graph schema, and achieve unified representation and management of multi-source heterogeneous knowledge. Step 14 specifically includes:

[0085] Step 141: Graph database application.

[0086] (1) Choose a high-performance graph database (such as Neo4j) to build a knowledge graph storage system.

[0087] (2) Define node types (such as entity types) and edge types (such as relationship types), design the schema structure of the graph, and achieve efficient storage and query of complex relationships.

[0088] Step 142: Knowledge integration and dynamic expansion.

[0089] (1) Establish a data synchronization and update mechanism to add newly extracted entities and relationships to the knowledge graph in real time.

[0090] (2) Implement conflict detection and resolution strategies, maintain the consistency and accuracy of the knowledge graph, and support the dynamic expansion of the knowledge base.

[0091] Step 143: Embedding and similarity retrieval.

[0092] (1) Use the k-nearest neighbor (k-NN) algorithm and cluster analysis to find similar performance issues and cases in the embedding space.

[0093] (2) Set similarity thresholds to filter out noise data and improve the accuracy of retrieval and analysis.

[0094] 3. Knowledge graph application stage.

[0095] Step 15: Performance Bottleneck Analysis. Use an embedding generation algorithm to generate text embeddings. Leverage similarity retrieval and cluster analysis techniques to identify and categorize historical performance issues. Based on the knowledge graph, use the inference engine to perform logical reasoning, automatically identifying potential performance bottlenecks in the current system and predicting potential performance issues.

[0096] Step 16: Optimize decision support. Provide optimization suggestions for performance testing, including recommended test cases, optimization strategies, and performance improvement measures. Support natural language interaction, allowing users to obtain analysis results and optimization suggestions through question-and-answer format, improving the user experience.

[0097] Specifically, it can use the inference engine to perform logical reasoning based on the knowledge graph, automatically identify the performance bottlenecks of the current system, and predict potential performance issues. It can also provide optimization suggestions, including recommended test cases, optimization strategies, and performance improvement measures, to assist testers in making decisions.

[0098] A natural language interface is provided. This interface integrates natural language understanding (NLU) technology, allowing users to query and interact using natural language. This interface implements intelligent question-and-answer functionality, allowing users to directly inquire about system performance, bottleneck analysis results, and optimization suggestions. API integration and system expansion are possible, providing a RESTful API interface that supports integration with third-party testing tools, monitoring systems, and operations and maintenance platforms. A modular architecture facilitates system expansion and upgrades, enhancing system flexibility and maintainability.

[0099] As can be seen from the above description, this application example provides efficient graph database support, integrating high-performance graph database technology to ensure efficient storage, query, and management of knowledge graph data and support large-scale data processing. It also offers advanced natural language processing capabilities. By leveraging large language models and natural language processing technologies, the system can automatically extract, understand, and analyze key knowledge from unstructured data. It also provides intelligent analysis and decision support. Utilizing machine learning and reasoning techniques, the system can automatically identify performance bottlenecks, provide optimization suggestions, and enhance the intelligence of testing. It also offers an interactive visual interface and natural language interaction capabilities, enhancing the user experience and lowering the barrier to entry.

[0100] To further illustrate this solution, this application uses an e-commerce butler transaction system as an example to provide an example of a method for diagnosing problems in a financial system. The specific description is as follows:

[0101] Step 21: Data Collection: We collected data from the bank’s transaction logs, system logs, user feedback, external market conditions, and other data sources over the past year, totaling approximately 15 TB, to establish a comprehensive dataset.

[0102] Step 22: Knowledge extraction: Using the fine-tuned BERT model, combined with NER and relationship extraction algorithms, we identified key entities (such as "withdrawal transaction" and "server node B") and relationships (such as "trigger" and "influence"), and extracted more than 1.5 million entity and relationship data.

[0103] Step 23: Knowledge graph construction: The extracted entities and relationships are imported into the Neo4j graph database to build a knowledge graph covering transaction types, system components, performance indicators, abnormal events, etc., realizing systematic management of knowledge. Figure 4 This is a schematic diagram of a knowledge graph for example. The nodes in the knowledge graph can represent entities, and the directed edges between nodes can represent the relationships between entities.

[0104] Step 24: Performance bottleneck analysis.

[0105] (1) The system detects that there is a response delay in the "withdrawal transaction" during a specific period of time. Through embedding generation and similarity retrieval, similar historical performance issues are located.

[0106] (2) Based on the reasoning of the knowledge graph, it is identified that the memory leak of "Server Node B" and the external market fluctuations are the main reasons.

[0107] Step 25: Optimize decision support.

[0108] (1) The system automatically recommended optimization solutions, including fixing memory leaks, increasing server resources, and optimizing trading algorithms.

[0109] (2) After optimization, transaction response time was shortened by approximately 35% and system stability was significantly improved.

[0110] From the software level, in order to improve the accuracy of financial system problem diagnosis while ensuring the efficiency of financial system problem diagnosis, this application provides an embodiment of a financial system problem diagnosis device for implementing all or part of the content of the financial system problem diagnosis method, see Figure 5 The problem diagnosis device of the financial system specifically includes the following contents:

[0111] The receiving module 10 is used to receive the problem diagnosis text input by the user from the front end.

[0112] The diagnosis module 20 is used to apply a pre-constructed knowledge graph and the problem diagnosis text to perform similarity retrieval, determine the diagnosis result corresponding to the problem diagnosis text and output it; wherein, the nodes in the pre-constructed knowledge graph represent key entities, and the connection between two nodes represents the relationship between key entities.

[0113] like Figure 6 As shown, in one embodiment, the diagnostic module 20 includes:

[0114] The segmentation unit 30 is used to segment the problem diagnosis text to obtain multiple text blocks.

[0115] The text embedding unit 40 is configured to apply an embedding generation algorithm to generate text embeddings corresponding to each text block.

[0116] The diagnosis unit 50 is used to apply the pre-built knowledge graph and the text embedding corresponding to each text block to perform similarity retrieval, determine the diagnosis result corresponding to the problem diagnosis text, and output it.

[0117] In one embodiment, the problem diagnosis device for a financial system further includes:

[0118] The acquisition module is used to collect multi-source unstructured data from the target financial system, wherein the multi-source unstructured data includes transaction logs, system logs, user feedback data, external documents, and market data.

[0119] The extraction module is used to apply the multi-source unstructured data, the pre-trained large-scale language model and the relationship extraction algorithm to obtain key entities and the relationships between key entities.

[0120] A construction module is used to construct the knowledge graph based on the key entities and the relationships between the key entities.

[0121] In one embodiment, the extraction module includes:

[0122] An application unit is used to apply the multi-source unstructured data and the pre-trained large-scale language model to obtain key entities.

[0123] An extraction unit is used to obtain the relationship between the key entities based on the multi-source unstructured data and a relationship extraction algorithm.

[0124] The embodiment of the problem diagnosis device for a financial system provided in this specification can be specifically used to execute the processing flow of the embodiment of the problem diagnosis method for a financial system. Its functions are not described in detail here, and reference can be made to the detailed description of the embodiment of the problem diagnosis method for a financial system.

[0125] Figure 7 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as Figure 7 As shown, the electronic device includes: a memory 701, a processor 702, and a computer program stored in the memory 701 and executable on the processor 702. When the processor 702 executes the computer program, the following method is implemented:

[0126] Step 100: Receive the problem diagnosis text of the target financial system input by the user from the front end.

[0127] Step 200: Apply a pre-constructed knowledge graph and the problem diagnosis text to perform similarity retrieval, determine the diagnosis result corresponding to the problem diagnosis text and output it, wherein the nodes in the pre-constructed knowledge graph represent key entities, and the connection between two nodes represents the relationship between key entities.

[0128] This embodiment discloses a computer program product, which includes a computer program. When the computer program is executed by a processor, the following method is implemented:

[0129] Step 100: Receive the problem diagnosis text of the target financial system input by the user from the front end.

[0130] Step 200: Apply a pre-constructed knowledge graph and the problem diagnosis text to perform similarity retrieval, determine the diagnosis result corresponding to the problem diagnosis text and output it, wherein the nodes in the pre-constructed knowledge graph represent key entities, and the connection between two nodes represents the relationship between key entities.

[0131] This embodiment provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the following method is implemented:

[0132] Step 100: Receive the problem diagnosis text of the target financial system input by the user from the front end.

[0133] Step 200: Apply a pre-constructed knowledge graph and the problem diagnosis text to perform similarity retrieval, determine the diagnosis result corresponding to the problem diagnosis text and output it, wherein the nodes in the pre-constructed knowledge graph represent key entities, and the connection between two nodes represents the relationship between key entities.

[0134] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0135] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0136] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0137] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0138] Throughout this specification, reference to terms such as "one embodiment," "a specific embodiment," "some embodiments," "for example," "example," "specific example," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0139] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for diagnosing problems in a financial system, characterized in that: include: Receive problem diagnosis text of the target financial system from the user's front-end; Apply the pre-built knowledge graph and the problem diagnosis text to perform similarity search, determine the diagnosis result corresponding to the problem diagnosis text and output it; Among them, the nodes in the pre-built knowledge graph represent key entities, and the connection between two nodes represents the relationship between key entities.

2. The problem diagnosis method of a financial system according to claim 1, characterized in that: The application performs similarity retrieval on the pre-built knowledge graph and the problem diagnosis text, determines and outputs the diagnosis result corresponding to the problem diagnosis text, including: Segmenting the problem diagnosis text to obtain multiple text blocks; Apply the embedding generation algorithm to generate text embeddings corresponding to each text block; The pre-built knowledge graph and the text embedding corresponding to each text block are applied to perform similarity retrieval, and the diagnosis result corresponding to the problem diagnosis text is determined and output.

3. The problem diagnosis method of a financial system according to claim 1, characterized in that: Before performing similarity retrieval on the pre-built knowledge graph and the problem diagnosis text, the method further includes: Collect multi-source unstructured data from the target financial system, including transaction logs, system logs, user feedback data, external documents, and market data; Applying the multi-source unstructured data, the pre-trained large-scale language model, and the relation extraction algorithm to obtain key entities and the relationships between the key entities; Based on the key entities and the relationships between the key entities, the knowledge graph is constructed.

4. The problem diagnosis method of a financial system according to claim 3, characterized in that: The application of the multi-source unstructured data, the pre-trained large-scale language model, and the relation extraction algorithm to obtain key entities and the relationships between key entities includes: Applying the multi-source unstructured data and the pre-trained large-scale language model to obtain key entities; The relationships between the key entities are obtained based on the multi-source unstructured data and the relationship extraction algorithm.

5. A problem diagnosis device for a financial system, characterized in that: include: The receiving module is used to receive the problem diagnosis text input by the user from the front end; A diagnosis module is used to apply a pre-built knowledge graph and the problem diagnosis text to perform similarity search, determine the diagnosis result corresponding to the problem diagnosis text, and output it; Among them, the nodes in the pre-built knowledge graph represent key entities, and the connection between two nodes represents the relationship between key entities.

6. The problem diagnosis device for a financial system according to claim 5, characterized in that: The diagnostic module includes: A segmentation unit, configured to segment the problem diagnosis text into multiple text blocks; A text embedding unit, which is used to apply an embedding generation algorithm to generate text embeddings corresponding to each text block; The diagnosis unit is used to apply the pre-built knowledge graph and the text embedding corresponding to each text block to perform similarity retrieval, determine the diagnosis result corresponding to the problem diagnosis text, and output it.

7. The problem diagnosis device for a financial system according to claim 5, characterized in that: Also includes: A collection module, configured to collect multi-source unstructured data from the target financial system, including transaction logs, system logs, user feedback data, external documents, and market data; An extraction module, configured to apply the multi-source unstructured data, a pre-trained large-scale language model, and a relation extraction algorithm to obtain key entities and the relationships between them; A construction module is used to construct the knowledge graph based on the key entities and the relationships between the key entities.

8. The problem diagnosis device for a financial system according to claim 7, characterized in that: The extraction module includes: An application unit, configured to apply the multi-source unstructured data and the pre-trained large-scale language model to obtain key entities; An extraction unit is used to obtain the relationship between the key entities based on the multi-source unstructured data and a relationship extraction algorithm.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the problem diagnosis method for a financial system according to any one of claims 1 to 4 is implemented.

10. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the instructions are executed by a processor, the problem diagnosis method for a financial system according to any one of claims 1 to 4 is implemented.

Images