Intelligent financial cue word generation and optimization system based on knowledge graph

Abstract

The invention relates to the technical field of financial science and technology and natural language processing, in particular to an intelligent financial cue word generation and optimization system based on a knowledge graph, which comprises a financial field knowledge base, a cue word element extraction module, a field specialization template library, a cue word generation engine, an evaluation and optimization module and a manual feedback and labeling system, the efficiency is improved, through template generation and parameterization configuration, the prompt word writing time is shortened, and the labor cost is remarkably reduced. And quality guarantee: the accuracy of professional terms is improved, the structural integrity and compliance coverage are remarkably improved, and the output deviation of a large model is reduced. And scene adaptation: the fields of security investment, financial analysis and the like are supported, and the cue word complexity and professional depth are automatically adjusted according to user roles (investors / analysts). And dynamic evolution: through real-time updating of the knowledge graph and artificial feedback learning, the system can adapt to financial environment changes (such as new supervision policies and emerging analysis methods), and continuously optimizes and generates strategies.

Description

Technical Field

[0001] The present invention relates to the fields of financial technology and natural language processing technology, and specifically to an intelligent financial prompt word generation and optimization system based on knowledge graphs. Background Art

[0002] With the widespread application of large language models (LLMs) in the financial field, building high-quality, professional financial prompts has become the key to improving the performance of AI systems. However, existing technologies have the following shortcomings:

[0003] High professional threshold: The financial field has numerous terms (such as "DCF valuation" and "Porter's Five Forces Model") and complex analysis frameworks (such as SWOT analysis and financial report indicator calculations), making it difficult for ordinary users to construct effective prompts.

[0004] High labor costs: Professionals need to manually write prompt words, which is time-consuming and labor-intensive, and it is difficult to ensure compliance (such as risk warnings, data source specifications) and logical rigor.

[0005] Unstable quality: The prompt words generated by traditional methods often have problems such as incorrect terminology, missing structure, or insufficient risk coverage, causing the output results of large models to deviate from expectations. Summary of the Invention

[0006] (1) Technical problems solved

[0007] In response to the deficiencies of the existing technology, the present invention provides an intelligent financial prompt word generation and optimization system based on knowledge graph.

[0008] (2) Technical solution

[0009] To achieve the above objectives, the present invention provides the following technical solution: a system for generating and optimizing intelligent financial prompt words based on a knowledge graph, comprising:

[0010] The financial knowledge base integrates knowledge from multiple fields such as finance, economics, and accounting to build a knowledge graph that includes entities (companies, indicators, policies), relationships (such as "indicator calculation relationships" and "industry affiliation"), and time-series data;

[0011] The prompt word element extraction module automatically parses user input to extract key elements such as professional terms, data specifications (time range, accuracy), analysis frameworks (such as DCF, SWOT), compliance factors (risk warnings, disclaimers), and user roles (investor, analyst);

[0012] A domain-specific template library, designed for specific fields such as securities investment, financial analysis, and risk management, includes structured frameworks such as goal statements, analysis steps, and output formats, and supports parameterized configuration.

[0013] Prompt word generation engine: Based on a multi-level generation strategy (skeleton generation → element filling → style adjustment), it combines knowledge graph data and template parameters to generate prompt words that meet user needs;

[0014] The evaluation and optimization module evaluates the quality of prompt words through multi-dimensional indicators (professional accuracy, structural integrity, and instruction clarity), combines large-scale model feedback with manual annotation, and dynamically optimizes the generation rules;

[0015] System integration interface, providing RESTful API, GraphQL interface, WebSocket real-time interface and multi-language SDK, supporting integration with third-party platforms such as financial analysis software and risk management systems;

[0016] The manual feedback and annotation system allows professionals to score and modify the generated prompt words, forming structured feedback data to drive the system's iterative learning.

[0017] Preferably, the financial knowledge base is constructed by the following method:

[0018] Integrate financial dictionaries, financial report data, and regulatory documents to build a knowledge graph containing company-indicator-policy entity relationships;

[0019] Embeds financial indicator calculation rules and analysis framework logic to support trend analysis of time series data.

[0020] Preferably, the domain-specific template library supports parameterized configuration, including:

[0021] Required parameter sets and optional parameter sets;

[0022] Dynamic adjustment mechanism of template parameters based on user roles.

[0023] Preferably, the mandatory parameter set includes data time range and compliance elements.

[0024] Preferably, the optional parameter set includes output format and analysis depth.

[0025] Preferably, the user roles include investors and analysts.

[0026] Preferably, the evaluation and optimization module includes:

[0027] Multi-dimensional evaluation indicators: professional accuracy, structural integrity, instruction clarity, data standardization, and risk coverage;

[0028] Closed-loop optimization process: Train the optimization engine through manually labeled data, update the knowledge graph and generate rules.

[0029] (3) Beneficial effects

[0030] Compared with the existing technology, the present invention provides an intelligent financial prompt word generation and optimization system based on knowledge graph, which has the following beneficial effects:

[0031] This knowledge graph-based intelligent financial prompt word generation and optimization system improves efficiency: through template generation and parameterized configuration, it shortens the prompt word writing time and significantly reduces labor costs.

[0032] Quality Assurance: Improved accuracy of professional terminology, significantly enhanced structural integrity and compliance coverage, and reduced deviation in large model outputs.

[0033] Scenario adaptation: Supports areas such as securities investment and financial analysis, and automatically adjusts the complexity and professional depth of prompt words based on user roles (investor / analyst).

[0034] Dynamic evolution: Through real-time updates of the knowledge graph and human feedback learning, the system can adapt to changes in the financial environment (such as new regulatory policies and emerging analytical methods) and continuously optimize generation strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of the structural system module of the present invention. DETAILED DESCRIPTION

[0036] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0037] See also Figure 1 , an intelligent financial prompt word generation and optimization system based on knowledge graph, including:

[0038] The financial knowledge base integrates knowledge from multiple fields such as finance, economics, and accounting to build a knowledge graph that includes entities (companies, indicators, policies), relationships (such as "indicator calculation relationships" and "industry affiliation"), and time-series data;

[0039] The prompt word element extraction module automatically parses user input to extract key elements such as professional terms, data specifications (time range, accuracy), analysis frameworks (such as DCF, SWOT), compliance factors (risk warnings, disclaimers), and user roles (investor, analyst);

[0040] A domain-specific template library, designed for specific fields such as securities investment, financial analysis, and risk management, includes structured frameworks such as goal statements, analysis steps, and output formats, and supports parameterized configuration.

[0041] Prompt word generation engine: Based on a multi-level generation strategy (skeleton generation → element filling → style adjustment), it combines knowledge graph data and template parameters to generate prompt words that meet user needs;

[0042] The evaluation and optimization module evaluates the quality of prompt words through multi-dimensional indicators (professional accuracy, structural integrity, and instruction clarity), combines large-scale model feedback with manual annotation, and dynamically optimizes the generation rules;

[0043] System integration interface, providing RESTful API, GraphQL interface, WebSocket real-time interface and multi-language SDK, supporting integration with third-party platforms such as financial analysis software and risk management systems;

[0044] The manual feedback and annotation system allows professionals to score and modify the generated prompt words, forming structured feedback data to drive the system's iterative learning.

[0045] Preferably, the financial knowledge base is constructed by the following method:

[0046] Integrate financial dictionaries, financial report data, and regulatory documents to build a knowledge graph containing company-indicator-policy entity relationships;

[0047] Embeds financial indicator calculation rules and analysis framework logic to support trend analysis of time series data.

[0048] Preferably, the domain-specific template library supports parameterized configuration, including:

[0049] Required parameter sets and optional parameter sets;

[0050] Dynamic adjustment mechanism of template parameters based on user roles.

[0051] Preferably, the mandatory parameter set includes data time range and compliance elements.

[0052] Preferably, the optional parameter set includes output format and analysis depth.

[0053] Preferably, the user roles include investors and analysts.

[0054] Preferably, the evaluation and optimization module includes:

[0055] Multi-dimensional evaluation indicators: professional accuracy, structural integrity, instruction clarity, data standardization, and risk coverage;

[0056] Closed-loop optimization process: Train the optimization engine through manually labeled data, update the knowledge graph and generate rules.

[0057] Financial term importance scoring algorithm:

[0058]

[0059] Prompt word template matching algorithm

[0060]

[0061] Comprehensive scoring algorithm for prompt word quality

[0062]

[0063]

[0064] Knowledge graph construction:

[0065] Multi-source data fusion: Integrate professional dictionaries, financial reports, and regulatory documents, and standardize financial indicators (such as PE, PB) and analysis frameworks.

[0066] Time series and reasoning support: Strengthens the time attribute processing of financial data, embeds financial indicator calculation rules (such as "net profit = operating income - cost"), and supports trend analysis and logical reasoning.

[0067] Template parameterized design:

[0068] Hierarchical template structure: divided into target layer (such as "generate stock investment recommendations"), strategy layer (such as "use DCF model valuation"), and execution layer (such as "require data source to be Wind database"), supporting parameter dependency modeling.

[0069] Dynamic rendering technology: Define template variable slots through a domain-specific language (DSL) and dynamically fill in expertise and compliance elements based on user roles and task types.

[0070] Two-way optimization mechanism:

[0071] Cascading calls of large models: Decompose complex tasks into steps such as "data analysis → framework construction → risk assessment", and improve efficiency through context management and intermediate result caching.

[0072] Human-machine collaborative iteration: After structured processing, manually labeled data is input into the optimization engine to adjust the generation rules, forming a "generation-evaluation-optimization" closed loop.

[0073] References to "embodiments" herein mean that the specific features, structures, or characteristics described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the word "embodiment" in various places in the specification does not necessarily refer to the same embodiment, nor does it particularly limit its independence or relevance to other embodiments. In principle, in this application, as long as there are no technical contradictions or conflicts, the various technical features mentioned in the embodiments can be combined in any manner to form a corresponding implementable technical solution.

[0074] Unless otherwise defined, the technical terms used herein have the same meanings as those generally understood by those skilled in the art to which this application belongs; the use of relevant terms herein is only for describing specific embodiments and is not intended to limit this application.

[0075] While the embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that various changes, modifications, substitutions, and alterations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. The intelligent financial prompt word generation and optimization system based on knowledge graph is characterized by: include: Financial knowledge base, used to store financial entities, relationships, and analysis frameworks, supporting time series data processing and logical reasoning; The prompt word element extraction module is used to parse user input and extract key elements such as professional terms, data specifications, analysis frameworks, and user roles; The domain-specific template library is divided into fields such as securities investment and financial analysis, and includes a three-level template structure: target layer, strategy layer, and execution layer; The prompt word generation engine generates prompt word skeletons based on multi-level strategies and populates knowledge graph data and template parameters; The evaluation and optimization module evaluates the quality of prompt words through multi-dimensional indicators and dynamically optimizes the generation rules by combining large-scale model feedback and manual annotation; System integration interface, providing software integration interface; The manual feedback and annotation system allows professionals to score and modify the generated prompt words, breaking down manual feedback into dimensions such as professional accuracy, logical completeness, and instruction clarity. The system calls a large model to analyze manual annotation data, extract improvement patterns, record the evolution history of prompt words, and support backtracking and comparison.

2. The intelligent financial prompt word generation and optimization system based on knowledge graph according to claim 1 is characterized in that: The financial knowledge base is constructed by the following method: Integrate financial dictionaries, financial report data, and regulatory documents to build a knowledge graph containing company-indicator-policy entity relationships; Embeds financial indicator calculation rules and analysis framework logic to support trend analysis of time series data.

3. The intelligent financial prompt word generation and optimization system based on knowledge graph according to claim 2 is characterized in that: The domain-specific template library supports parameterized configuration, including: Required parameter set and optional parameter set; Dynamic adjustment mechanism of template parameters based on user roles.

4. The system for generating and optimizing intelligent financial prompt words based on knowledge graph according to claim 3 is characterized in that: The mandatory parameter set includes data time range and compliance elements.

5. The system for generating and optimizing intelligent financial prompt words based on knowledge graph according to claim 4 is characterized in that: The optional parameter set includes output format and analysis depth.

6. The system for generating and optimizing intelligent financial prompt words based on knowledge graph according to claim 5, characterized in that: The user roles include investors and analysts.

7. The system for generating and optimizing intelligent financial prompt words based on knowledge graph according to claim 6, characterized in that: The evaluation and optimization module includes: Multi-dimensional evaluation indicators: professional accuracy, structural integrity, instruction clarity, data standardization, and risk coverage; Closed-loop optimization process: Train the optimization engine through manually labeled data, update the knowledge graph and generate rules.

Images