Ontology-Based Search System for Content Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for searching and analyzing online content, such as earned media, rely on brute force keyword searches, leading to numerous false positives and false negatives, making the process time-consuming and error-prone due to lexical complexity and homonym issues.
Innovation Solution
A model-based approach is employed to analyze content, creating or updating an ontology that represents entities and their relationships, allowing for knowledge-based searching to improve the relevance and accuracy of search results by expanding search entities and tailoring searches based on contextual knowledge.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If brute force keyword search is used, then search coverage is high, but search accuracy deteriorates due to false positives and false negatives
Solution Approach 1:
The patent introduces an ontology as an intermediary layer between the search query and the content database. This ontology contains structured knowledge about entities, their attributes, and relationships, allowing the system to translate simple keyword searches into sophisticated semantic queries that distinguish between different meanings of words like 'apple' (fruit vs. company), thereby improving search accuracy without requiring complete redesign of the search infrastructure
Solution Approach 2:
The system performs preliminary action by pre-building the ontology from available content before the actual search occurs. This ontology construction process organizes entities, attributes, and relationships in advance, so that when a search is executed, the system can immediately leverage this pre-processed knowledge structure to filter and rank results accurately without performing complex analysis during the search itself
2Reliability
If keyword search is used, then processing speed is high, but search quality deteriorates due to lexical complexity and homonyms
Solution Approach 1:
The ontology is constructed in advance from the content database, pre-organizing entities, their attributes, and interrelationships. This preliminary action ensures that when a search is performed, the system can quickly query the structured ontology rather than analyzing unstructured text in real-time, maintaining fast processing speeds while significantly improving result reliability by correctly disambiguating homonyms and handling lexical complexity
Solution Approach 2:
The system changes the parameter of search from simple keyword matching to multi-dimensional entity attribute matching. By transforming queries to operate on ontology parameters (entity types, attributes, relationships) rather than raw text, the system achieves both high reliability in distinguishing between different meanings of words and efficient processing through structured query evaluation
3Measurement precision
If more content is analyzed, then model accuracy improves, but computational resources increase
Solution Approach 1:
The ontology structure enables local quality by organizing content analysis at the entity level rather than processing all content uniformly. The system can analyze and build ontology structures for specific domains or entity types as needed, allowing high model accuracy for targeted searches while avoiding the computational expense of analyzing entire content databases for every query
Solution Approach 2:
The system performs preliminary content analysis to build the ontology structure before searches are executed. This one-time computational investment creates a reusable knowledge base that can serve multiple searches efficiently, improving model accuracy through comprehensive content analysis while amortizing the computational cost across many subsequent queries rather than repeating full analysis for each search
Data Source
AI summary
Embodiments of a search system that provides knowledge based searching of content based on a knowledge model created from the content being searched are disclosed. Embodiments of such search systems may build a model of entitles and relationships representing the collective knowledge contained in a set of content analyzed. When a search is performed on content the model may be leveraged to improve the accuracy, relevance and recall of the search.