Relationship Graph Search for Cross-Entity Structured Records
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search and analytics engines in large-scale data systems struggle with retrieving meaningful, cross-entity insights due to the lack of native join capabilities across schema-defined entities, leading to complexity and inefficiency, especially in non-relational data platforms like the Open Subsurface Data Universe (OSDU).
Innovation Solution
A system that integrates schema definitions, relationship graphs, and structured query generation using a language model to construct relationship graphs, plan traversal paths, and produce outputs, incorporating Retrieval-Augmented Generation (RAG) techniques to manage large schema contexts and adapt to query complexity without relying on native relational features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If native relational join capabilities are used, then cross-entity retrieval is efficient, but schema-based data systems lack this capability and require complex workarounds
Solution Approach 1:
The patent introduces a relationship graph as an intermediary structure that bridges schema-defined entities without requiring native relational join capabilities. The relationship graph models inter-entity relationships and enables traversal between entities, serving as a mediator that provides relational-like functionality in non-relational data platforms.
Solution Approach 2:
The patent replaces the mechanical relational join operations with a language model-based system that uses natural language queries to navigate relationship graphs. Instead of relying on database-level join mechanisms, the system uses semantic understanding and graph traversal to achieve cross-entity retrieval.
2Measurement precision
If complex queries are decomposed into multiple entity-specific queries, then retrieval precision improves, but query processing complexity increases
Solution Approach 1:
The patent segments complex queries into multiple entity-specific queries by decomposing the original query into components that can be processed against individual entities in the relationship graph. This segmentation enables precise retrieval from each entity while maintaining overall query intent through the coordinated execution of multiple simpler queries.
Solution Approach 2:
The patent implements dynamic query generation where the language model adapts the query structure based on the specific entities and relationships involved. The system dynamically determines which entities to query and how to traverse the relationship graph, making the query processing flexible rather than rigid.
3Ease of manufacture
If schema definitions are used to structure data, then data organization is improved, but cross-entity relationships become difficult to query
Solution Approach 1:
The relationship graph serves as an intermediary layer between the schema definitions and the query processing. It translates schema-based data organization into traversable relationships that can be efficiently queried through natural language, bridging the gap between structured data storage and flexible query operations.
Solution Approach 2:
The language model provides universal query processing capability that works across all schema-defined entities and relationships. Instead of requiring entity-specific query handling, the single language model can interpret and execute queries for any combination of entities in the relationship graph, providing multi-functional query support.
Data Source
AI summary
A method implements language model powered search on structured records using relationship graphs. A relationship graph representing entity relationships among a set of tables in a database based on multiple schema definitions is constructed. A structured query based on the natural language query and the multiple schema definitions is generated. The structured query is deconstructed to extract a source table of the set of tables, a target table of the set of tables, a query condition, and an aggregation operator. A traversal path is determined across the relationship graph based on the source table and the target table. A set of entity-specific queries are executed using the query condition and the traversal path to retrieve records from the database. A response is generated based on the retrieved records and the aggregation operator. The response includes one or more of a textual summary and a visualization based on the output prompt.