Text Function Tagging for Search Relevance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional search systems often recommend irrelevant documents when users select the 'Similar pages' link, due to insufficient semantic context modeling and inability to transport context across taxonomies, leading to low relevance in search results.
Innovation Solution
A search system and method that uses lightweight text function tagging to analyze text excerpts and identify relevant documents by applying text function tags to both reference and secondary documents, allowing for quick and cost-effective retrieval of relevant documents without increasing latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional keyword-index search is used to identify similar documents, then search speed is maintained, but document relevance is low
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing text function tags for all documents in the corpus during an offline processing phase. This allows the search system to have semantic context information ready in advance, so that when a user queries for similar documents, the system can quickly retrieve and compare pre-tagged documents without performing complex real-time semantic analysis, thus maintaining search speed while improving relevance
Solution Approach 2:
The patent introduces text function tags as an intermediary representation between raw text and semantic meaning. These tags serve as a mediator that captures the functional role of text excerpts (such as title, summary, body, conclusion) without requiring full semantic understanding. This intermediary layer enables efficient comparison and matching of documents based on their structural and functional characteristics rather than just keyword overlap
2Reliability
If full semantic analysis is performed on all documents to improve relevance, then document relevance improves, but system complexity and processing time increase
Solution Approach 1:
The patent applies the extraction principle by isolating and analyzing only the most salient portions of documents - specifically text excerpts such as titles, summaries, and key sentences - rather than performing full semantic analysis on entire documents. By extracting and tagging only these critical portions, the system achieves improved document relevance while avoiding the complexity and computational burden of analyzing complete document contents
Solution Approach 2:
The patent applies partial action by performing semantic tagging on a subset of documents or a subset of text features rather than comprehensive analysis of all documents. The system tags text excerpts with functional categories to capture essential meaning without exhaustive semantic processing, achieving sufficient relevance improvement with reduced system complexity
3Reliability
If text function tagging is applied to all documents, then document relevance improves, but processing cost increases
Solution Approach 1:
The patent applies preliminary action by performing text function tagging during an offline pre-processing phase rather than during online query processing. All documents are tagged with their text function categories in advance, and these tags are stored for rapid retrieval. This shifts the computational cost from high-frequency online operations to a one-time offline operation, dramatically reducing the processing cost per query while maintaining improved document relevance
Solution Approach 2:
The patent applies partial action by selectively tagging only text excerpts (titles, summaries, key sentences) rather than analyzing every word and sentence in full documents. This partial tagging approach captures the essential functional structure of documents sufficient for relevance determination, while significantly reducing the computational resources and processing cost required compared to comprehensive document analysis
4Reliability
If semantic context modeling is enhanced to transport context across taxonomies, then document relevance improves, but latency in generating search results increases
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing text function tags that capture semantic context during an offline phase. These pre-computed tags encode the functional meaning and contextual role of text excerpts, enabling the system to quickly retrieve and match documents with similar semantic contexts without performing complex real-time semantic modeling. This maintains low latency while improving the ability to transport context across different document taxonomies
Solution Approach 2:
The patent introduces text function tags as an intermediary representation that bridges different document taxonomies and semantic contexts. These tags provide a standardized functional vocabulary that allows meaningful comparison and matching of documents from different categories or domains, enabling context transport across taxonomies without requiring complex real-time semantic reasoning that would increase latency
Data Source
AI summary
A search system with a user interface module that generates a search engine results page including a listing of reference documents, each with a hyperlink to at least one secondary document relevant to the reference document. The search system also includes a text function tagger that parses text excerpts of the documents, and applies text function tags thereto. A secondary processing module is provided which identifies those secondary documents that have a text function tag that matches a text function tag of a reference document, so that selection of the hyperlink generates a display of secondary documents that are determined to at least have a text function tag that matches a text function tag of the reference document for which corresponding hyperlink was selected. A method and computer readable medium are also provided.