Query Ranking via Clustering and Categorization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search query ranking systems suffer from redundancy and inefficiency in presenting top rising and top volume queries, as they often list multiple variations of the same topic and fail to include queries with significant interest due to individual search volume thresholds, leading to a lack of diversity and accuracy in reflecting current trends and interests.
Innovation Solution
Implement a query ranking system that groups similar search queries into clusters, selects a representative query for each cluster based on popularity and quality indicators, and ranks clusters according to their constituent queries and associated categories, reducing redundancy and enhancing diversity in query listings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If search queries are ranked by individual search volume thresholds, then queries with high search volume are prominently displayed, but redundant query variations of the same topic are listed separately, reducing diversity
Solution Approach 1:
The patent merges similar search queries into clusters based on their result documents, URLs, or search terms. Queries that target the same topic are grouped together, and only one representative query is displayed per cluster. This combining approach reduces redundancy while maintaining the quantity of displayed queries through selective representation.
Solution Approach 2:
The patent segments the query list into distinct clusters based on topic similarity. Each cluster is then represented by a selected representative query. This segmentation allows the system to handle multiple queries efficiently by processing them in grouped units rather than individually, reducing overall complexity.
2Measurement precision
If multiple query variations are listed separately, then individual query performance can be evaluated, but the listing becomes redundant and less efficient
Solution Approach 1:
The patent combines multiple query variations into single representative queries per cluster. The representative query is selected based on metrics such as search volume, increase in search volume, or quality indicators like page rank and click-through rate. This merging maintains evaluation accuracy while significantly improving listing efficiency by reducing the number of individual queries to process and display.
3Adaptability or versatility
If queries are grouped into clusters with representative queries, then redundancy is reduced and diversity is improved, but the system complexity increases
Solution Approach 1:
The patent employs self-service mechanisms where the clustering system automatically groups queries and selects representative queries without manual intervention. The system uses automated algorithms to evaluate queries based on multiple metrics and make decisions about which queries to cluster and which to represent, reducing the need for complex manual processing while maintaining high diversity.
4Quantity of substance
If individual search volume thresholds are used for ranking, then queries meeting the threshold are included, but queries with significant interest but below threshold are excluded, reducing accuracy
Solution Approach 1:
The patent changes the ranking parameters from simple search volume thresholds to a multi-parameter evaluation system. Queries are evaluated based on search volume, increase in search volume, and quality indicators such as page rank and click-through rate. This parameter change allows the system to include queries with significant interest that may not meet individual search volume thresholds, improving trend reflection accuracy.
Data Source
AI summary
Methods, systems, and apparatus, including computer program products, for query ranking based on query clustering and categorization, are disclosed. In one aspect, search queries are selected and grouped into one or more clusters. A representative query is selected for each cluster. Each cluster is associated with a respective representative category. A rank is assigned to each cluster based on a cluster popularity score of the cluster and a category popularity score of the cluster's representative category. The selected representative queries are presented in order according to the ranks of their respective clusters.