Section-Scoped Search for Cognitive Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional information retrieval systems face challenges in efficiently searching and reviewing literature due to non-normalized documents with unique section layouts and headings, making it cumbersome for users to find relevant content without reading entire documents or using general text searches that return excessive results.
Innovation Solution
A cognitive analytics and search system that uses natural language processing and machine learning to identify and normalize document sections, allowing users to define and search specific sections within documents, thereby focusing efforts on relevant content and improving discovery efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If general text search is used to search entire documents, then comprehensive content coverage is achieved, but search results become excessive and review efficiency decreases
Solution Approach 1:
The patent segments documents into standardized sections (abstract, introduction, methods, results, discussion, conclusion) and enables users to search within specific sections rather than entire documents. This segmentation reduces the volume of search results while maintaining comprehensive content coverage by allowing targeted searches in relevant sections only.
Solution Approach 2:
The patent applies local quality by treating different document sections with different searchability properties. Each section can be independently searched, filtered, or excluded based on user needs, allowing high-quality relevant results from specific sections without the noise from other sections, thus improving review efficiency.
2Adaptability or versatility
If documents with unique section layouts and headings are searched without normalization, then document diversity is preserved, but search difficulty increases
Solution Approach 1:
The patent performs preliminary normalization of document sections before searching. By pre-processing documents to identify and standardize section headings and layouts, the system prepares the data in advance, making subsequent searches easier and more efficient while preserving the original document diversity and unique characteristics.
Solution Approach 2:
The patent introduces an intermediary normalization layer between the diverse document formats and the search function. This intermediary process standardizes section identification across different document types without altering the original documents, thereby maintaining document diversity while reducing search difficulty through unified section recognition.
3Loss of information
If entire documents are reviewed to find relevant content, then complete information is obtained, but time consumption increases
Solution Approach 1:
The patent divides documents into searchable sections and allows users to target specific sections relevant to their queries. This segmentation enables obtaining complete information from relevant sections without the time cost of reviewing entire documents, as users can directly search and access only the sections containing needed content.
Solution Approach 2:
The patent applies partial action by enabling users to search and review only the necessary portions (specific sections) of documents rather than entire documents. This partial approach maintains information completeness for relevant content while significantly reducing time consumption by excluding irrelevant sections from the review process.
Data Source
AI summary
A computer system includes a natural language processing (NLP) unit, a storage unit, a user interface and a search engine. The NLP unit analyzes a content source to identify one or more sections containing searchable content and generate section metadata respective to each identified section included in the content source. The storage unit stores the section metadata and the user interface receives a section-scoped query aimed at searching an identified section corresponding to the at least one first section metadata stored in the storage unit without searching an identified section corresponding to at least one second section metadata stored in the storage unit. Based on the section-scoped query, the search engine analyzes the at least one first section metadata stored in the storage unit without analyzing the at least one second section metadata.