Topic-Based Relevance Ranking for Web Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional web search models often fail to return the most relevant results and improperly rank links based on relevance, leading to suboptimal user experiences in web searches.
Innovation Solution
The technology employs topic-based relevance ranking, where topics from web page titles, anchor texts, URLs, and bodies are extracted and matched against user queries to compute a relevance score, using a modified edit distance algorithm to determine the matching level and rank pages accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional relevance models (BM25, language modeling, proximity modeling) are used for web search ranking, then the search system can operate with simple algorithms, but the relevance and ranking accuracy of returned results deteriorates
Solution Approach 1:
The patent introduces topics as an intermediary representation between web pages and user queries. Instead of directly comparing raw page content with queries using complex conventional models, the system extracts topics from pages and matches them with query topics, simplifying the relevance ranking process while improving accuracy.
Solution Approach 2:
The patent changes the representation parameters from traditional term-frequency-based models (BM25) to topic-based representations. By transforming page content into topic vectors and query into topic distributions, the system achieves better relevance matching with more manageable computational complexity.
2Ease of operation
If conventional relevance models are used to return search results, then the system maintains simplicity in implementation, but the quality and relevance of returned links deteriorates
Solution Approach 1:
Topics serve as a mediator that simplifies the relationship between pages and queries. The system extracts topics from page content (titles, anchor texts, URLs, bodies) and represents queries as topic distributions, enabling simpler yet more reliable relevance ranking through topic matching.
Solution Approach 2:
The patent replaces complex mechanical relevance ranking mechanisms (BM25 term matching, language modeling) with a topic-based semantic matching approach. This substitution maintains implementation simplicity while significantly improving search result reliability and relevance.
3Measurement precision
If topics are extracted from multiple page elements (title, anchor texts, URL, body) to improve relevance scoring, then the relevance measurement accuracy improves, but the topic extraction and processing time increases
Solution Approach 1:
The patent segments the topic extraction process into multiple independent components: title extraction, anchor text extraction, URL extraction, and body extraction. Each component processes specific page elements separately, allowing for optimized processing strategies and parallel execution, thereby reducing overall time loss while maintaining comprehensive relevance measurement.
Data Source
AI summary
Described is a technology by which topics corresponding to web pages are used in relevance ranking of those pages. Topics are extracted from each web page of a set of web pages that are found via a query. For example, text such as nouns may be extracted from the title, anchor texts and URL of a page, and used as the topics. The extracted topics from a page are used to compute a relevance score for that page based on an evaluation of that page's topics against the query. The pages are then ranked relative to one another based at least in part on the relevance score computed for each page, such as by determining a matching level for each page, ranking pages by each level, and ranking pages within each level. Also described is training a model to perform the relevance scoring and/or ranking.