Query Expansion via Click-Through Data for Web Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional information retrieval models face term mismatch issues due to differences in expression between user queries and document content, leading to irrelevant search results, especially in web searches where 'AND' logic is applied, causing relevant documents to be overlooked.
Innovation Solution
A system that identifies and issues multiple similar search queries based on click-through data to retrieve and re-rank search results, combining them to enhance relevance and comprehensiveness, using a re-ranking model to address term mismatch and improve search accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If query expansion is applied to web search by adding terms to a query, then the scope of search is expanded, but the search engine returns fewer documents due to AND logic
Solution Approach 1:
The patent segments the search process into multiple independent queries. Instead of expanding a single query with multiple terms connected by AND logic, the system divides the search into several separate queries, each containing a subset of terms. This allows each query to be processed independently, avoiding the restrictive AND logic that reduces result quantity while maintaining comprehensive search scope.
Solution Approach 2:
The patent transitions from a single-dimension query expansion approach (adding terms to one query) to a multi-dimensional approach by generating multiple alternative queries. Each query represents a different dimension or perspective of the search intent, allowing the system to explore diverse document sets that would otherwise be excluded by conventional AND logic.
2Measurement precision
If term matching is used in information retrieval, then documents sharing query terms are identified, but documents with relevant concepts using different terms are missed
Solution Approach 1:
The patent changes the parameter of query representation by generating multiple variations of the original query with different term combinations. This allows the search system to capture documents that use alternative terminology by matching against multiple query formulations rather than relying on a single rigid term match.
Solution Approach 2:
The patent introduces an intermediary process that generates multiple queries as mediators between the user's original search intent and the document collection. These intermediate queries serve as bridges to connect the user's terminology with the diverse terminology used in documents, reducing information loss from term mismatch.
3Reliability
If multiple similar queries are issued to retrieve search results, then diversity and relevance of results increase, but system complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-generating multiple query variations before executing the search. This preparation step creates a set of ready-to-use queries that can be executed in parallel or sequence, reducing the complexity of real-time query generation while maintaining high result relevance through diverse query formulations.
Data Source
AI summary
Systems, methods, and devices are described for retrieving query results based at least in part on a query and one or more similar queries. Upon receiving a query, one or more similar queries may be identified and/or calculated. In one embodiment, the similar queries may be determined based at least in part on click-through data corresponding to previously submitted queries. Information associated with the query and each of the similar queries may be retrieved, ranked, and or combined. The combined query results may then be re-ranked based at least in part on a responsiveness and/or relevance to the previously submitted query. The re-ranked query results may then be output to a user that submitted the original query.