Entity Feature Vectors for Role-Aware Query Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning techniques fail to consider the order and different roles of entities in atoms, limiting the ability to perform processing that accounts for these differences.
Innovation Solution
A processing device and method that includes an entity pre-processing unit to calculate entity feature vectors reflecting the correspondence between keys and entities in an associative array, and a post-processing unit to execute queries using these vectors, allowing for processing that distinguishes the roles of entities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If entity feature vectors are calculated without considering the order and roles of entities in atoms, then the processing is simpler and faster, but the accuracy of entity-specific processing deteriorates
Solution Approach 1:
The patent segments the entity representation by creating separate embedding vectors for different entity roles (subject, object, etc.) within atoms. Instead of treating all entities uniformly, the system divides entity processing into role-specific segments, allowing each entity to be represented according to its specific function in the atom, thereby improving accuracy while maintaining manageable complexity through structured organization.
Solution Approach 2:
The patent applies local quality by assigning different processing characteristics to entities based on their roles. Each entity position in an atom (subject, object, etc.) receives specialized embedding treatment tailored to its specific function. This allows the system to optimize processing for each role independently, improving overall entity-specific processing accuracy without requiring complete redesign of the entire processing pipeline.
2Measurement precision
If entity order and roles in atoms are considered, then the accuracy of entity-specific processing improves, but the processing time and computational complexity increase
Solution Approach 1:
The patent implements preliminary action by pre-computing and storing embedding vectors for entities and atoms before actual processing occurs. These pre-computed embeddings capture the structural information and role relationships in advance, allowing the system to quickly retrieve and utilize this information during entity-specific processing without performing complex calculations in real-time, thus reducing processing time while maintaining accuracy.
Solution Approach 2:
The patent changes parameters by transforming entity and atom information into fixed-dimensional embedding vectors that encode structural and role information. This parameter transformation converts complex variable-length atom structures into standardized vector representations, enabling efficient computation and reducing processing time while preserving the essential information needed for accurate entity-specific processing.
Data Source
AI summary
A processing device includes: an entity pre-processing means for receiving a set of atoms, the atom indicating a combination of a predicate and an associative array of entities that are arguments of the predicate, and calculating for each of the entities an entity feature vector for each of the entities, the vector reflecting a correspondence between keys and the entity in the associative array; and a post-processing means for receiving a query indicating contents of processing, and executing the processing indicated by the query using the entity feature vector.