Hierarchical Metadata Generator for Granular Document Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search technologies face challenges in efficiently locating specific information within documents due to the overinclusive nature of full text searching and the inadequacy of metadata-based searching in capturing finer-grained details, leading to irrelevant or missing results.
Innovation Solution
A hierarchical metadata generator system that splits documents into fragments based on logical components or semantic content, associates metadata with these fragments, and uses a query engine to locate documents based on topic-specific queries, allowing for precise retrieval of information by comparing query terms with metadata and fragment order.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If full text searching is used, then all documents containing search terms are located, but too many irrelevant results are returned
Solution Approach 1:
The patent divides documents into multiple fragments based on markup tags (paragraphs, sections, chapters) and creates separate metadata records for each fragment. This segmentation allows the search engine to index and search individual fragments independently, enabling more precise matching of search queries to specific document portions rather than returning entire documents based on any keyword match.
2Reliability
If metadata-based searching is used, then fewer and more relevant results are obtained, but finer-grained details of documents are not captured
Solution Approach 1:
The patent segments documents into fragments and generates metadata for each fragment independently, preserving fine-grained information that would be lost in traditional document-level metadata. This allows search queries to match specific fragments containing detailed information while maintaining search relevance.
Solution Approach 2:
The patent adds a new dimension to metadata organization by creating a hierarchical structure where metadata exists at both the document level and the fragment level. This multi-level metadata approach enables searches to operate on different granularities, from overall document topics to specific fragment details, without losing information at any level.
3Measurement precision
If documents are divided into fragments, then finer-grained search capability is achieved, but system complexity increases
Solution Approach 1:
The patent divides documents into fragments based on existing markup tags, leveraging the document's inherent structure rather than creating arbitrary divisions. This approach achieves fine-grained search capability while reducing complexity by using already-defined logical boundaries in the document structure.
Solution Approach 2:
The patent creates a universal fragment-based metadata system that can be applied to any structured document regardless of type or content. The same fragmentation and metadata generation process works across diverse document formats and structures, reducing the need for document-specific processing logic and thereby lowering overall system complexity.
Data Source
AI summary
A computer-implemented method of locating information in a database of electronic documents includes defining fragments of the documents, associating the fragments with the document from which the fragments originated, and associating metadata with the fragments, where the metadata associated with a fragment includes metadata related to one or more topics of the fragment. A query for one or more documents containing information about a topic is received, and a document is located from the database based on a comparison of the query with the metadata associated with a fragment of the document.