Semantic Search Query Expansion for Vague User Requests
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search technologies struggle to effectively handle vague or imprecise search queries, limiting users to specific and precise terms, and fail to provide a comprehensive range of relevant results due to corpus limitations in vector and semantic searches.
Innovation Solution
Implement a vagueness-exploratory search mechanism that utilizes a word thesaurus and confident terms data store to identify and construct multiple search queries with semantically similar terms, incorporating natural language processing to parse and tag confident and vague terms, and execute these queries to enhance result relevance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional search technologies use specific and precise terms, then search query accuracy is improved, but the range of relevant results is limited
Solution Approach 1:
The system changes the parameter of search terms by generating multiple variants with different semantic meanings. It parses the input query to identify confident terms, then generates alternative interpretations by replacing these terms with semantically different but related terms from a thesaurus, thereby expanding the search scope while maintaining query accuracy
Solution Approach 2:
The search query is segmented into confident terms and vague terms. The system identifies and extracts confident terms from the input query, separates them from vague terms, and then selectively generates alternatives only for the confident terms. This segmentation allows precise control over which parts of the query to expand, balancing accuracy and versatility
2Adaptability or versatility
If vector and semantic searches are used, then broader result ranges are provided, but corpus limitations reduce result relevance
Solution Approach 1:
The system introduces an intermediary component - a thesaurus with curated semantic relationships - between the user query and the search corpus. This intermediary provides structured semantic alternatives that are more reliable than generic vector embeddings, ensuring that expanded search terms maintain relevance to the original intent while still broadening the search scope
3Productivity
If multiple search queries with semantically similar terms are constructed, then result comprehensiveness is improved, but processing complexity increases
Solution Approach 1:
The system applies partial action by selectively generating semantic alternatives only for confident terms in the query, not for all terms. It also limits the number of alternatives generated per term and uses filtering to eliminate redundant results. This partial approach maintains productivity benefits while controlling processing complexity
Data Source
AI summary
An apparatus comprises a processing device configured to receive and parse a search request related to a given event to identify confident terms, and to determine a first subset of the confident terms which indirectly reference the given event and a second subset of the confident terms which are not in the first subset. The processing device is also configured to identify, for a given confident term in the second subset, terms having at least a threshold level of semantic similarity with the given confident term, and to construct a primary search query comprising the first and second subsets of the confident terms and additional search queries comprising the first subset of the confident terms and the identified one or more terms having at least the threshold level of semantic similarity with the given confident term. The processing device is further configured to execute the primary and additional search queries.