Knowledge Graph Entity Filtering for Natural Language Query Significance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning systems for natural language queries are limited in identifying significant events beyond the keywords present in the query, failing to leverage deeper relationships and entities within a corpus, leading to incomplete search results.
Innovation Solution
A method that builds a knowledge graph based on a corpus to identify related entities and events, using mutual information to calculate significance scores, allowing for deeper analysis and extraction of significant events beyond the initial query keywords.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If the system limits keywords to those occurring within the actual query, then the simplicity of the search mechanism is maintained, but the completeness and depth of search results deteriorate
Solution Approach 1:
The system performs preliminary actions by pre-processing the corpus to build a knowledge graph that captures entity relationships, event structures, and semantic connections before the actual search query is executed. This pre-computed knowledge structure enables the system to expand query keywords and identify significant events without adding complexity to the real-time search mechanism.
Solution Approach 2:
The knowledge graph serves as an intermediary between the simple keyword query and the complex task of identifying significant events. It mediates by providing pre-computed relationships and contextual information that bridge the gap between basic keyword matching and comprehensive event analysis, allowing the system to maintain simplicity while achieving completeness.
2Measurement precision
If the system uses only keywords from the query, then the processing speed is maintained, but the precision of event identification deteriorates
Solution Approach 1:
The system pre-computes and stores event significance scores, entity relationships, and contextual information in a knowledge graph during an offline phase. When a query is received, the system quickly retrieves and utilizes this pre-computed information rather than performing complex analyses in real-time, thus maintaining fast processing speed while achieving high precision in event identification.
Solution Approach 2:
The knowledge graph creates a copied or simplified representation of the complex corpus data, storing pre-extracted entities, events, and their relationships in a structured format. This copied knowledge structure allows the system to perform precise event identification by querying the pre-processed data rather than re-analyzing the entire corpus for each search.
3Loss of information
If the system performs deep analysis to uncover relationships beyond query keywords, then the relevance of search results is improved, but the computational resources required increase
Solution Approach 1:
The system performs the computationally intensive deep analysis work in advance by building the knowledge graph offline, extracting entities, events, and relationships from the corpus before search queries are executed. This shifts the computational burden from online query processing to offline pre-processing, reducing real-time energy consumption while maintaining high relevance of search results through the use of pre-computed deep relationships.
Solution Approach 2:
The knowledge graph creates a compact copied representation of the corpus that encodes deep relationships and semantic information in a space-efficient manner. This copied structure allows the system to retrieve relevant information without re-performing expensive computational analyses, thereby reducing computational resource usage while preserving the relevance benefits of deep analysis.
Data Source
AI summary
A knowledge graph is built based on a corpus stored in the computer system. The corpus includes a set of searchable events and each event includes a respective set of entities. A set of entities is identified in a first set of significant events returned by natural language query (NLQ). The knowledge graph determines which ones of the set of entities are related to the entities in the NLQ to produce a filtered set of entities. The filtered set of entities is used to identify a second set of significant events in the selected corpus. Members of the first and second set of significant events are presented to a user as a search result.