Synonym Generation from Query Logs for Search Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users face challenges in accessing network-accessible content due to variations in how they query information items, as existing techniques are not reliable in matching non-trivial variations of canonical forms, and manual annotation of synonyms is time-consuming and incomplete.
Innovation Solution
A three-phase approach using query log data to generate synonyms for canonical information items, including an expansion phase to identify initial candidates, a clean-up phase to remove noise, and a reduction phase to rank and threshold the candidates, thereby providing a set of selected synonyms that reflect how users prefer to refer to information items.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual annotation of synonyms is performed, then the system can capture known variants of information items, but it requires significant user time and effort and may fail to capture all variants
Solution Approach 1:
The system automatically generates synonym lists by analyzing query log data without requiring manual user input. The computer identifies variants of information items autonomously by processing historical query patterns, eliminating the need for users to manually annotate synonyms while still capturing comprehensive variants that users actually employ
Solution Approach 2:
The system uses query log data as feedback to continuously improve synonym generation. By analyzing actual user queries that successfully retrieve information items, the system learns and adapts to capture accurate synonyms dynamically, improving reliability without additional manual effort
2Productivity
If simple textual matching techniques are used, then the system can quickly process queries, but it fails to match non-trivial variations such as nicknames or abbreviated forms
Solution Approach 1:
The system performs preliminary analysis of query log data to pre-identify and store synonym relationships before actual query processing. By preparing synonym lists in advance based on historical patterns, the system can quickly match queries against these pre-computed synonyms during runtime, maintaining high speed while improving accuracy for non-trivial variations
Solution Approach 2:
The system introduces an intermediary synonym matching layer between the user query and the canonical information item. This intermediary layer translates various query forms (nicknames, abbreviations) into canonical forms using pre-learned synonym relationships, enabling accurate matching without requiring complex processing of each query variant
3Device complexity
If the system maintains only canonical forms of information items, then the data structure remains simple, but users cannot access content when using non-canonical query forms
Solution Approach 1:
The system creates synthetic synonym copies of canonical information items based on query log analysis. Instead of storing multiple actual variants, it generates virtual synonym representations that map various query forms to the canonical form, maintaining simple data storage while providing flexible query matching through these synthesized copies
Data Source
AI summary
An approach is described for generating synonyms to supplement at least one information item, such as, in one case, a set of related items. The approach can involve an expansion phase, a clean-up phase, and a reduction phase. In the expansion phase, the approach identifies, for each related item, a set of initial synonym candidates. In the clean-up phase, the approach removes noise from the set of initial synonym candidates (if such noise exists), to provide a set of filtered synonym candidate items. In the reduction phase, the approach ranks and applies a threshold (or thresholds) to the set of filtered synonym candidate items, to generate, for each information item, a set of selected synonyms. The approach uses query log data at various points in its operation. The selected synonyms can be used to improve the effectiveness of user searches.