Document Object Search via Formatting Attribute Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing file formats do not effectively capture and make searchable the semantic content implied by authors, as this information is not explicitly encoded, leading to difficulties in searching and retrieving specific information within documents.
Innovation Solution
A method and system that identify objects within documents, determine their formatting attributes, and use inference rules to generate metadata that estimates author-defined semantic characteristics, making the content searchable by comparing these attributes and embedding the metadata within the document for search engines to identify the objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If semantic content is not explicitly encoded in file formats, then file format simplicity is maintained, but searchability and information retrieval capability deteriorate
Solution Approach 1:
The patent segments the document into multiple objects (text, images, tables, etc.) and applies different metadata schemas to each object type. This allows semantic information to be captured at the object level without complicating the overall file format structure, resolving the contradiction between simplicity and information richness.
Solution Approach 2:
The patent introduces metadata as an intermediary layer between the raw document content and the search functionality. This metadata layer captures semantic characteristics (author intent, object type, formatting attributes) without modifying the underlying file format, enabling searchability while maintaining format simplicity.
2Quantity of substance
If author-defined semantic characteristics are not captured, then document storage efficiency is maintained, but search accuracy and information retrieval quality deteriorate
Solution Approach 1:
The patent changes the parameter representation by inferring semantic characteristics from formatting attributes (font size, bold, italic, color, position) and encoding these as searchable metadata parameters. This allows search accuracy to improve without significantly increasing storage requirements, as the metadata is derived from existing document properties rather than requiring separate storage of semantic information.
3Device complexity
If formatting attributes are used to infer semantic characteristics, then implementation complexity is reduced, but inference accuracy may deteriorate compared to explicit author tagging
Solution Approach 1:
The patent implements a feedback mechanism where multiple inference rules are applied to formatting attributes and the results are combined to determine the most likely semantic characteristic. The system continuously refines its inferences by comparing multiple attributes and rules, improving accuracy while maintaining relatively simple implementation through rule-based processing.
Data Source
AI summary
A method for generating a searchable file from a document. The method includes identifying an object within the document and a formatting attribute of the object, determining an inferred semantic characteristic of the object by comparing the formatting attribute with a plurality of inference rules, wherein the inferred semantic characteristic comprises an estimation of author-defined semantic characteristic of the object, generating metadata based at least on the inferred semantic characteristic and the formatting attribute of the object, wherein the metadata comprises text data that is searchable by a search application for the searchable file to identify the object within the searchable file, and generating, based on the document, the searchable file comprising the metadata.