Semantic Data Representation for Explainable Ad Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing computer systems struggle to deeply understand and process the meaning of natural language, leading to inefficiencies in handling broad and heterogeneous data sets, particularly in applications requiring extensive structured data schemas, which are impractical to build and maintain.
Innovation Solution
A machine-readable language, referred to as UL, is used to represent data in a structured and expressive format that allows for semantic nodes and links to be processed efficiently, enabling automated systems to understand and reason about natural language inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If structured data schemas are used to represent broad and heterogeneous data, then data organization and processing capability are improved, but schema design complexity and maintenance difficulty increase significantly
Solution Approach 1:
The patent introduces an intermediary layer (knowledge graph, semantic network, or structured representation system) that mediates between natural language input and structured data processing. This intermediary automatically converts unstructured natural language into structured representations, eliminating the need for manual schema design while maintaining data organization and processing capabilities.
2Adaptability or versatility
If comprehensive structured schemas are created to cover all possible data types, then data coverage and application scope are improved, but building and coding effort become impractical
Solution Approach 1:
The system employs self-service mechanisms where the structured representation system automatically generates, updates, and maintains schemas based on incoming natural language data. The system self-adapts to new data types and concepts without requiring manual schema design, thereby achieving comprehensive data coverage with minimal building effort.
Solution Approach 2:
The schema structure is made dynamic and adaptive, automatically evolving to accommodate new data types and relationships as they appear in natural language inputs. This dynamic approach allows the system to cover all possible data types without requiring预先 design of comprehensive schemas.
3Productivity
If statistical machine learning and deep learning are used to process natural language, then processing capability is improved, but explainability and human-understandable reasoning are lost
Solution Approach 1:
The patent introduces an intermediary structured representation system that serves as a bridge between statistical machine learning processing and human-understandable reasoning. This intermediary maintains structured, interpretable representations that preserve explainability while enabling sophisticated processing capabilities through automated reasoning over the structured data.
Data Source
AI summary
A computer implemented method for the automated analysis or use of data, comprising: storing in a non-transitory storage medium a structured, machine-readable representation of data that conforms to a machine-readable language that represents meaning. The data relates to a part of advertisements, news articles or other information items. The structured, machine-readable representation of data comprises semantic nodes and passages, and each semantic node represents an entity represented by an identifier; and each passage is a semantic combination of semantic nodes. Machine-readable meaning comes from the choice of semantic nodes and the way they are combined and ordered as passages; (b) representing information about a specific individual in the structured, machine-readable representation of data; (c) automatically processing the structured, machine-readable representation of data to identify advertisements relevant to the specific individual.