Query Suggestion System Using User Group Associations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search engine query suggestion methods fail to effectively target a user's known interests by relying solely on past queries from data logs, limiting the scope and relevance of suggested queries and ignoring other factors.
Innovation Solution
Suggesting queries based on user associations with groups formed by past queries, website links, and preferences, as well as using structured expressions to derive new query suggestions that are dynamically determined as the user types, allowing for more relevant and diverse query options.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If query suggestions are determined based only on past queries from data logs, then the suggestions are grounded in actual user behavior, but the scope and relevance of suggested queries is severely limited and fails to target known user interests
Solution Approach 1:
The patent merges multiple data sources including past queries from data logs, user profiles containing known interests, and currently entered query context to generate comprehensive query suggestions. This combination allows the system to maintain reliability from actual user behavior while expanding adaptability through diverse information sources.
Solution Approach 2:
The system implements a multi-functional suggestion mechanism that can provide different types of queries based on different sources: queries from data logs represent actual user behavior, queries from user profiles represent known interests, and queries derived from current input represent contextual relevance. This universal approach handles multiple functions within a single suggestion system.
2Ease of operation
If query suggestions are determined based only on the currently entered query, then the suggestions are contextually relevant, but many other factors that can be used to provide relevant query suggestions are ignored
Solution Approach 1:
The system performs preliminary actions by pre-processing and storing user profile information including known interests and demographic data before query suggestion is needed. This allows the system to quickly retrieve and apply relevant user-specific information when generating suggestions, rather than computing it from scratch each time.
Solution Approach 2:
The system incorporates feedback loops where user interactions with suggested queries are monitored and used to refine future suggestions. The system also uses feedback from user profiles and behavioral data to continuously improve the relevance of suggestions by adjusting weights and priorities of different information sources.
3Reliability
If query suggestions are limited to queries already submitted to a search engine, then the suggestions are based on proven search behavior, but the scope of suggested queries is severely limited
Solution Approach 1:
The patent adds new dimensions to query suggestion by incorporating user profile data and demographic information as additional layers beyond traditional query logs. This dimensional expansion allows the system to suggest queries based on user interests even when those specific queries haven't been previously submitted, thereby expanding scope while maintaining reliability through multi-factor validation.
Data Source
AI summary
Methods and computer-readable media are provided for determining suggested queries. A user enters a search website, and the user is identified based on a user identification. Suggested queries are determined based on a group associated with the user. This association is created by extracting queries from data logs, categorizing the queries into groups based on their respective subject matter, associating the user with one or more groups, and determining suggested queries for each group. The suggested queries are communicated for display.