Search Re-scoring via Query-specific Impact Measures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search systems face challenges in predicting the relative importance of features for specific queries and resources, leading to difficulties in accurately ranking search results, as the influence of features can be emergent and vary significantly across different queries and resources.
Innovation Solution
The system adjusts search operations by determining impact measures for each feature on the ranking of resources and re-scoring them based on these measures, allowing for flexible adjustments to search parameter values on a per-query basis, thereby re-ranking resources to better match user informational needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a robust search algorithm is used to rank resources, then the search system performs well over a wide variety of resources, but the system cannot adapt to the varying importance of features for specific queries and resources
Solution Approach 1:
The search system dynamically adjusts feature weights based on query characteristics and resource properties. Instead of using fixed feature importance values, the system computes adjusted weights by multiplying base weights with query-specific and resource-specific modifiers, enabling the algorithm to adapt to varying feature importance across different queries and resources while maintaining the robustness of the original search algorithm.
Solution Approach 2:
The system changes the parameters of the search algorithm by introducing query-specific and resource-specific weight adjustments. The feature weights are modified based on query templates, resource types, and other contextual factors, allowing the same search algorithm to perform optimally across diverse queries and resources by dynamically altering its weighting parameters.
2Measurement precision
If feature weights are adjusted to improve ranking accuracy for specific queries, then search result relevance improves, but the complexity of the search operation increases
Solution Approach 1:
The system performs preliminary computations of query-specific and resource-specific weight adjustments before the actual search ranking. By pre-computing these modifiers based on query templates and resource types, the system reduces the computational burden during the search execution phase, as the weight adjustments are already determined and can be applied directly to the base feature weights.
Solution Approach 2:
The feature weighting process is segmented into multiple independent components: base feature weights, query-specific modifiers, and resource-specific modifiers. Each component can be computed and adjusted independently, allowing for modular optimization and reducing the overall complexity of the search operation by breaking down the weight adjustment process into manageable segments.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for receiving queries, and for each received query: receiving data indicating resources identified by a search operation as being responsive to the query, wherein the search operation scores each of the resources based, in part, on features of the resource and the query, selecting a subset of the resources, determining, from the subset of resources and for each of the features of the resources and the query, an impact measure that measures the impact of the feature on the ranking of the resources that belong to the subset, adjusting the search operation based on the respective impact measures, and initiating the search operation to re-score the resources in the subset of resources based, in part, on the adjustment and to rank the subset of resources according to a second order that is different from the first order.