Lexicon Generation for Accurate Electronic Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic search systems fail to retrieve relevant documents due to the lack of a standardized lexicon, leading to missed results when users input slightly different terms or phrases, as they do not effectively identify and match semantic equivalents.
Innovation Solution
A method and system for generating a lexicon by determining corpus terms, creating candidate terms, selecting normalized terms, and linking them to their equivalents, enabling more accurate and efficient electronic searches by correlating various forms of terms with their normalized forms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a standardized lexicon is implemented to match semantic equivalents, then search accuracy improves, but device complexity increases
Solution Approach 1:
The system performs preliminary actions by pre-generating candidate terms from corpus terms and pre-selecting normalized terms before the actual search operation. The lexicon generation process creates term mappings in advance, so when a user performs a search, the system can quickly retrieve pre-computed results rather than performing complex semantic analysis in real-time.
Solution Approach 2:
The patent introduces an intermediary lexicon structure that mediates between user search queries and document content. The lexicon acts as a mapping layer that connects corpus terms, candidate terms, and normalized terms, allowing the system to match semantic equivalents without requiring complex real-time analysis. This intermediary structure simplifies the search process by pre-establishing term relationships.
2Device complexity
If exact term matching is used in search, then system complexity remains low, but search accuracy deteriorates due to inability to retrieve documents with variant terms
Solution Approach 1:
The system generates candidate terms and selects normalized terms in advance, creating a comprehensive term mapping before search operations. This preliminary processing allows the system to handle term variations efficiently during actual searches without adding significant complexity to the search algorithm itself.
Solution Approach 2:
The patent creates copies of terms in different forms (corpus terms, candidate terms, normalized terms) and establishes mappings between them. Instead of requiring complex semantic analysis during search, the system uses these pre-created term copies and their established relationships to match queries with document content, significantly simplifying the search process while improving accuracy.
Data Source
AI summary
Disclosed herein are embodiments for lexicon generation. More specifically, at least one embodiment of a method includes determining a corpus term from a plurality of documents, generating a candidate term from the corpus term, and selecting a normalized term from the candidate term and the corpus term. Some embodiments include linking the normalized term with the candidate term and providing an electronic search capability for locating a first document, where the electronic search capability receives the candidate term as a search term and utilizes the normalized term to locate the first document.