Data Object Generation Using Vector Search and Knowledge Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Creating new data objects in complex enterprise software applications is challenging due to schema complexity, requiring technical know-how and is prone to errors, and users without knowledge or skills struggle to create semantically useful objects that are resource-efficient.
Innovation Solution
A system that leverages a vector store and knowledge graph to automatically generate new data objects from user input by identifying schema elements and association paths, using large language models to ensure data integrity and relevance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If users manually create new data objects in complex enterprise software applications, then data integrity and semantic correctness can be maintained, but the process becomes extremely time-consuming and requires high technical expertise
Solution Approach 1:
The patent introduces an intermediary system comprising a language model and a vector store that mediates between user natural language input and the complex schema creation process. The vector store embeds schema elements and relationships into vector representations, enabling the language model to understand and navigate complex data relationships without users needing technical expertise, thus maintaining reliability while reducing time investment
Solution Approach 2:
The patent replaces the mechanical manual process of creating data objects with an automated AI-driven system. Instead of users manually navigating complex schemas and writing code, the system uses natural language processing and vector similarity matching to automatically generate data objects, substituting manual mechanical operations with intelligent automation
2Manufacturing precision
If users manually navigate complex schemas to create data objects, then precise control over data structure can be achieved, but the complexity and difficulty increase significantly
Solution Approach 1:
The vector store acts as an intermediary that transforms complex schema structures into simplified vector representations. By embedding schema elements, attributes, and relationships into vectors, the system maintains precise control over data structure while hiding the underlying complexity from users, who interact only through natural language
Solution Approach 2:
The patent transforms the parameter space of schema navigation by converting discrete, complex schema elements into continuous vector representations. This parameter transformation enables the language model to operate in a simplified space while maintaining the ability to generate precise data structures through vector similarity matching and embedding comparisons
3Productivity
If automated methods are used to generate data objects, then time consumption is reduced, but understanding and maintaining data relationships becomes more difficult
Solution Approach 1:
The language model serves as an intermediary that bridges automated generation and relationship understanding. It processes user intent, queries the vector store for relevant schema elements and relationships, and generates data objects while preserving relationship semantics. This intermediary layer maintains relationship understanding even as automation increases productivity
Solution Approach 2:
The system incorporates feedback loops where the language model continuously refines its understanding of data relationships based on user input and vector store responses. This feedback mechanism ensures that automated generation maintains relationship understanding by iteratively adjusting the generated data objects to match user intent and schema constraints
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
The present disclosure relates to techniques for automatically generating new data objects from user input. The system receives user input comprising a plurality of words and executes a first query on a vector store to identify schema elements similar to keywords in the user input. The vector store provides a response with similarity scores for identified elements. A second query is executed on a knowledge graph to identify association paths between data objects that include the identified elements. The knowledge graph response includes association information linking source and target data objects through selected elements. Full association paths are constructed from this information, and a command is generated to instantiate a new data object with elements corresponding to the user input. This approach leverages the strengths of large language models, vector stores, and knowledge graphs to efficiently and accurately create new data objects, ensuring data integrity and relevance.