Vertical Search Ranking via Data Quality and Popularity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Vertical search engines face limitations in providing comprehensive search results as they rely solely on feed data, missing out on relevant information from web crawls, and struggle to effectively combine and rank items from both feed and crawl sources, leading to incomplete and unreliable results.
Innovation Solution
A method for calculating a global ranking score for items based on both data quality and user popularity, allowing vertical search engines to order search results consistently across all data sources, regardless of whether the data comes from feed or crawl sources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If vertical search engines rely solely on feed data, then data reliability is improved, but information completeness deteriorates
Solution Approach 1:
The patent combines data from two distinct sources: feed data (structured, reliable) and crawl data (unstructured, comprehensive). The system merges these data types into a unified index, allowing the search engine to benefit from both the reliability of feed data and the completeness of crawl data simultaneously.
Solution Approach 2:
The patent creates a composite data structure that integrates feed data and crawl data with different characteristics. By treating data sources as composite materials with different properties (reliability vs. completeness), the system optimizes the combination to achieve both reliability and completeness in search results.
2Reliability
If items are ordered based on data source reliability, then data quality is improved, but user interest coverage deteriorates
Solution Approach 1:
The patent changes the ranking parameter from simple source-based ordering to a composite scoring system that incorporates multiple factors including user behavior signals (clicks, views, conversions). This parameter transformation allows items to be ranked by overall relevance rather than just data source quality, improving user interest coverage.
Solution Approach 2:
The ranking system is dynamic and adapts based on user interactions. As users click on and interact with items, the system updates popularity scores and adjusts rankings in real-time, allowing the ordering to evolve from static source-based ranking to dynamic user-interest-based ranking.
3Loss of information
If vertical search engines use only feed sources, then data completeness is improved, but data source versatility deteriorates
Solution Approach 1:
The patent creates a universal search system that can handle multiple data source types (feeds, crawls, structured data, unstructured data) through a common indexing and ranking framework. This multi-functional approach allows the system to process and rank items from diverse sources uniformly, achieving both completeness and versatility.
Data Source
AI summary
A vertical search engine may rank items based on both the quality of data associated with each item and the popularity of each item. The vertical search engine may access data associated with items from a variety of different sources, including feed sources and crawl sources. Data quality inputs are determined for each item based on the quality of the data associated with each respective item. In addition, popularity inputs are determined for each item based on user interest in each respective item. A global rankings score is then calculated for each item based on the data quality inputs and popularity inputs for each respective item. The global ranking score may be used to order search results for search queries in such a way that items from feed data and items from crawl data may be displayed in a unified manner, rather than being segregated by data source.