Example-Based Document Search via Textual and Linkage Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing keyword-based search engines are inadequate for complex and ambiguous information retrieval tasks, such as patent and literature searches, as they require iterative and time-consuming processes to find relevant information.
Innovation Solution
A method and system for searching electronic documents using example-based search queries, which processes inputs as textual, topical, and linkage-based similarity scores to provide more relevant results by determining relevance scores based on textual similarity, topical similarity, and linkage relationships within a network of documents.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If keyword-based search is used, then search precision can be maintained for simple queries, but search productivity deteriorates due to iterative trial-and-error processes
Solution Approach 1:
The patent uses example documents as copies or templates to represent search queries. Instead of requiring users to formulate keywords, the system accepts example documents that directly represent the desired search intent. This copying approach allows the system to learn from the example document's structure, content, and relationships to retrieve similar documents, thereby maintaining precision while eliminating iterative keyword guessing and improving productivity.
2Measurement precision
If multiple similarity metrics are computed, then search result quality improves, but computational complexity increases
Solution Approach 1:
The patent segments the relevance scoring process into three distinct similarity metrics: textual similarity (comparing document content), topical similarity (comparing subject matter), and linkage relationship similarity (comparing document connections). Each metric is computed independently and then integrated to produce the final relevance score. This segmentation allows for systematic computation of multiple factors while maintaining computational tractability through modular processing.
Data Source
AI summary
A computer implemented method for searching electronic documents, and associated system and computer program product. The method includes receiving an input representing an example-based search query and processing the input. The method also includes determining, for each of the electronic documents, a relevance score between the input and the corresponding electronic document. The determination is based on, at least, textual similarity between the input and the corresponding electronic document, topical similarity between the input and the corresponding electronic document, as well as linkage relationship in a linkage network of the plurality of electronic documents. The method also includes determining, based on the determined relevance scores, a search result containing one or more of the electronic documents. The search results will be provided to the user.