Query Intent-Based Multilingual Search Result Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current search engines face processing delays and overheads when routinely issuing queries to multiple languages and regions to improve search result relevance, especially for content-poor languages and multilingual regions.
Innovation Solution
A method for multilingual and cross-region search result merging, where query intent features are used to select and merge search results from different language and region pairs, allowing for the translation of queries while preserving intent features, and presenting all relevant results to the user without manual translation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If search engines routinely issue queries to multiple languages and regions to improve search result relevance, then search result relevance is improved, but processing delays and computational overheads increase significantly
Solution Approach 1:
The system performs multilingual and cross-region search only for a selected subset of queries based on query analysis, rather than routinely for all queries. This partial action approach improves relevance for content-poor languages and multilingual regions while avoiding the processing delays and computational overheads that would result from applying the technique to every query.
2Reliability
If search engines issue queries to multiple languages and regions, then search result relevance is improved, but computational overheads increase significantly
Solution Approach 1:
The system applies multilingual and cross-region search processing only to a selected subset of queries that benefit most from this approach, rather than processing all queries through multiple language-region pairs. This reduces computational overheads while maintaining improved relevance for queries where it matters most.
Solution Approach 2:
The system segments the query processing workflow into distinct stages: initial query analysis to identify candidates for multilingual processing, selective issuance of queries to multiple language-region pairs only for selected queries, and merging of results. This segmentation allows the system to avoid the computational overheads of routine multilingual processing while maintaining relevance improvements where needed.
3Productivity
If search engines only return results with the same language as the query and from the same region, then processing is faster and simpler, but search result relevance deteriorates for content-poor languages and multilingual regions
Solution Approach 1:
The system maintains fast and simple processing for the majority of queries by using only the query's original language and region, while selectively applying multilingual and cross-region search to a subset of queries where it improves relevance. This partial application preserves processing speed while improving relevance for content-poor languages and multilingual regions.
Solution Approach 2:
The system applies different processing strategies to different queries based on their characteristics. For most queries, it uses simple same-language same-region processing. For specific queries identified through analysis (particularly in content-poor languages or multilingual regions), it applies the more complex multilingual and cross-region search approach. This local quality differentiation optimizes both processing speed and relevance.
Data Source
AI summary
Multilingual/cross-region search result merging is described. In various examples, a user enters a query to an information retrieval system and receives search results appropriate to different combinations of human language and geographical region, and where the combinations are selected according to features describing intent of the query. In examples, query intent features are computed using a query understanding component (e.g. a trained query intent classifier) and used, optionally with other query features, to identify combinations of human language and geographical region from a stored profile. In examples the stored profile is computed using data about documents available to an information retrieval system and data about past search behavior. In examples, different combinations of language and region are used to retrieve search results which are merged. For example, a query is translated into another language while preserving calculated intent features of the query.