Generative Model Answers Using Query-Specific User Data Structures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current generative models, such as chatbots powered by large language models, struggle to provide satisfactory answers to context-specific queries that require user-specific information, as they often rely solely on general knowledge bases and fail to integrate user-specific data effectively.
Innovation Solution
A method and system that involves creating a computer-readable data structure specific to a context-specific query, incorporating user-specific features and values, and applying a generative model to generate a context-specific answer by integrating both general and user-specific information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a generative model relies solely on general knowledge bases to answer queries, then the model can provide general information (e.g., IRS rules), but it cannot provide satisfactory context-specific answers (e.g., which filing status saves more money for a specific user)
Solution Approach 1:
The patent implements nesting by embedding a user-specific data structure within the generative model's context. The system creates a nested architecture where user-specific features (income, filing status, deductions) are nested within the general knowledge base structure, allowing the model to access both general IRS rules and specific user data simultaneously through a hierarchical data organization approach
Solution Approach 2:
The patent introduces an intermediary data structure that mediates between the general knowledge base and user-specific information. This intermediary structure processes and integrates user-specific features with general tax rules, enabling the generative model to receive unified context that combines both general and specific information without direct complex integration between disparate systems
2Loss of information
If a chatbot provides general information about tax rules, then it can answer general questions, but it fails to answer specific questions requiring user-specific data
Solution Approach 1:
The patent applies segmentation by dividing the information processing into distinct segments: general knowledge base segmentation (tax rules, regulations) and user-specific data segmentation (income, filing status, deductions). Each segment is processed independently through dedicated data structures, then integrated by the generative model to provide comprehensive answers without overwhelming system complexity
Solution Approach 2:
The patent implements local quality by creating specialized data structures tailored to specific user contexts. Instead of a uniform data structure for all queries, the system dynamically creates local data structures that contain only the relevant user-specific features needed for each particular query type, optimizing information integration while minimizing unnecessary complexity
3Reliability
If user-specific data is integrated into the query processing, then context-specific answers can be generated, but the data structure complexity increases
Solution Approach 1:
The patent implements universality by designing a multi-functional data structure that can handle both general knowledge base queries and user-specific context queries through the same processing framework. The computer-readable data structure is engineered to universally accommodate different types of information (general rules and specific user data) with a unified schema, reducing the need for separate complex processing paths
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
A method including receiving a context-specific query, specific to the user, from a user device of a user. The method also includes creating a computer-readable data structure for storing data. The computer-readable data structure is specific to the context-specific query. The method also includes determining a feature related to the context-specific query. The method also includes retrieving a user value for the feature. The user value is retrieved from a data repository storing user-specific data that includes the user value. The method also includes modifying the computer-readable data structure to generate a modified computer-readable data structure by adding the feature and the user value for the feature to the computer-readable data structure. The method also includes applying a generative model to the modified computer-readable data structure to generate an output. The output includes a context-specific answer to the context-specific query. The method also includes returning the output.