Query Suggestion Templates Using Entity Categories
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Information retrieval systems, such as search engines, face challenges in providing effective query suggestions as they often rely on past user queries, which may not accurately reflect user intent or context, leading to irrelevant or less useful suggestions.
Innovation Solution
The method involves identifying query templates from a database of entities, such as a knowledge graph, where queries conforming to a template include terms that match entity categories and aliases, and ranking these templates based on occurrence frequency and similarity, to determine their effectiveness in generating relevant query suggestions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If query suggestions are generated using past user queries, then the system can provide suggestions based on real user behavior, but the suggestions may not accurately reflect user intent or context leading to irrelevant results
Solution Approach 1:
The patent introduces query templates as an intermediary structure between raw past queries and suggestion generation. These templates capture contextual patterns and entity relationships, serving as a mediator that preserves contextual information while enabling reliable suggestion generation. The templates act as a structured representation that bridges the gap between historical data and future predictions.
Solution Approach 2:
The system transforms raw query data into structured templates with specific parameters including entity categories, aliases, and occurrence frequencies. By changing the parameter representation from simple query strings to structured templates with multiple attributes, the system maintains contextual information while improving the reliability of suggestions through ranked template matching.
2Reliability
If the system uses entity categories and aliases from knowledge graphs, then query suggestions become more contextually appropriate, but the system complexity increases
Solution Approach 1:
The system performs preliminary action by pre-processing queries into structured templates with entity categories and aliases before suggestion generation. This advance structuring captures contextual relationships upfront, allowing the suggestion system to simply match against pre-defined templates rather than analyzing raw queries in real-time, thus managing complexity through preparation.
Solution Approach 2:
The patent segments queries into distinct components: template structure, entity categories, aliases, and occurrence frequencies. This segmentation breaks down the complex task of contextual understanding into manageable structured elements, where each component serves a specific function in the suggestion generation process, reducing overall system complexity through modular organization.
3Productivity
If the system ranks query templates based on occurrence frequency, then it can identify effective templates quickly, but it may miss less frequent but more relevant query patterns
Solution Approach 1:
The system changes the ranking parameter from simple occurrence frequency to a composite metric that incorporates entity category coherence and template structure quality. This parameter transformation allows the system to maintain fast identification through quantitative scoring while improving relevance by weighting factors that indicate meaningful query patterns, balancing speed and accuracy.
Data Source
AI summary
Methods and apparatus related to identifying a query template that includes one or more terms and a member of an entity category and ranking the query template. Query suggestions may be determined based on the query template. The ranking of the query template may be utilized to determine if the query template is a valid query template. The ranking of the template may be based on at least a count of occurrences of one or more of the entities of the entity category in past queries that conform to the query template.