Search Context Clustering for Query Recommendation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search services fail to provide personalized and varied query recommendations based on user intent and query history, often suggesting the same queries for repeated search inputs, which limits user experience and convenience.
Innovation Solution
A method and system that generate candidate queries using a search context, including query history and search results, employing clustering and multiple recommendation logics such as ambiguous, attribute, related, and hot topic queries to provide unique and relevant recommendations, while removing overlapping queries to enhance user experience.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional search services provide query recommendations, then users can get search suggestions, but the recommendations are repetitive and do not reflect user intent, reducing recommendation quality
Solution Approach 1:
The system performs preliminary analysis of query history and user intent before generating recommendations. By pre-processing search context and clustering similar queries in advance, the system prepares personalized recommendation data structures that enable accurate, non-repetitive suggestions without compromising response time.
Solution Approach 2:
The system incorporates user query history and interaction patterns as feedback to continuously refine recommendation quality. By analyzing past search behavior and adjusting recommendation strategies based on user responses, the system adapts to individual user intent and reduces repetitive recommendations over time.
2Adaptability or versatility
If the system generates multiple candidate queries using multiple recommendation logics, then recommendation variety increases, but system complexity increases
Solution Approach 1:
The recommendation system is segmented into multiple independent logic modules (ambiguous query recommendation, attribute type query recommendation, related query recommendation, hot topic query recommendation). Each module handles a specific aspect of query generation, allowing the system to provide diverse recommendations while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The system employs multiple recommendation logics that serve different functions but work together within a unified framework. Each logic type (ambiguous, attribute, related, hot topic) performs a specific function, yet all contribute to the overall recommendation system, achieving versatility without proportionally increasing complexity.
3Measurement precision
If the system clusters candidate queries based on similarity, then recommendation accuracy improves, but processing time increases
Solution Approach 1:
The system performs clustering on a selected subset of candidate queries rather than all possible queries. By applying clustering only to relevant candidates identified through preliminary filtering, the system achieves sufficient recommendation accuracy while avoiding the excessive processing time that would result from comprehensive clustering of all potential queries.
Data Source
AI summary
Provided is a query recommendation method and system using a search context. A recommendation query providing method may include generating candidate queries to be recommended to a user based on a search context including a query history associated with the user and a search result provided for each query; performing clustering of the candidate query based on a similarity between candidate queries; and providing an electronic device of the user with a final candidate query selected from each cluster of candidate queries as a recommendation query.