Context-Aware Query Expansion via Statistical Machine Translation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search query expansion methods often select inappropriate synonyms, leading to irrelevant search results due to the ambiguity of words with multiple meanings, as seen in queries like 'How to ship a box' where 'ship' can be synonymous with 'boat' or 'send', resulting in irrelevant results related to fishing trawlers.
Innovation Solution
The method employs statistical machine translation to identify context-specific synonyms by analyzing recorded search queries, using bi-directional phrase alignments and translation models trained on question-answer pairs, query-snippet pairs, and paraphrase pairs to expand search queries, ensuring that only relevant synonyms are used based on the context of the original query.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If synonyms are identified from expert-specified thesaurus or lexical ontology, then the query expansion covers more potential meanings, but irrelevant search results increase due to word ambiguity
Solution Approach 1:
The patent applies local quality by making synonym selection context-dependent rather than uniform. It extracts contextual features (surrounding words, query structure) and uses them to select appropriate synonyms from multiple candidates. This resolves the contradiction by ensuring high relevance (reliability) while maintaining comprehensive coverage (adaptability) through localized, context-aware synonym selection.
Solution Approach 2:
The patent changes the parameter of synonym selection from static (fixed thesaurus entries) to dynamic (context-dependent selection). By introducing contextual parameters and using machine learning models to weigh different synonym candidates based on query context, it achieves both broad coverage and high relevance simultaneously.
2Adaptability or versatility
If multiple potential synonyms are considered for query expansion, then the search coverage increases, but the difficulty of selecting the appropriate synonym increases due to context ambiguity
Solution Approach 1:
The patent uses feedback mechanisms by training machine learning models on query-expansion data and using the results to improve future synonym selections. The system learns from contextual patterns and adjusts synonym selection probabilities based on feedback from successful query expansions, reducing selection difficulty while maintaining comprehensive coverage.
Solution Approach 2:
The system performs self-service by automatically selecting appropriate synonyms without manual intervention. The machine learning models autonomously evaluate multiple synonym candidates and select the most appropriate ones based on contextual analysis, resolving the difficulty of manual synonym selection while maintaining broad search coverage.
3Reliability
If context-aware synonym selection is implemented, then search result relevance improves, but the complexity of the query expansion system increases
Solution Approach 1:
The patent applies segmentation by breaking down the complex synonym selection task into manageable components: contextual feature extraction, synonym candidate generation, and context-based filtering. This modular approach maintains high relevance while reducing system complexity through divided, specialized processing stages.
Solution Approach 2:
The patent introduces contextual features as an intermediary between the query and synonym selection process. This intermediary layer analyzes the query context and guides synonym selection, simplifying the overall system by providing a structured intermediate representation that bridges the gap between raw queries and synonym choices.
Data Source
AI summary
Methods, systems and apparatus, including computer program products, for expanding search queries. One method includes receiving a search query, selecting a synonym of a term in the search query based on a context of occurrence of the term in the received search query, the synonym having been derived from statistical machine translation of the term, and expanding the received search query with the synonym and using the expanded search query to search a collection of documents. Alternatively, another method includes receiving a request to search a corpus of documents, the request specifying a search query, using statistical machine translation to translate the specified search query into an expanded search query, the specified search query and the expanded search query being in the same natural language, and in response to the request, using the expanded search query to search a collection of documents.