Normalized Relevance Score Blending for Search Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Search engines face challenges in accurately comparing and ranking content relevance scores from different content corpora, as these scores have varying ranges and meanings across different types of content, leading to inconsistent representation of relevance in search results.
Innovation Solution
The technique involves normalizing relevance scores by generating a normalized range based on the ranges of content relevance scores from different corpora, allowing for a unified comparison and ranking of content across different categories, including social connections and public content, to provide a more relevant and blended output.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If relevance scores from different content corpora are used directly for ranking, then content from various sources can be included in search results, but the scores have varying ranges and meanings leading to inconsistent representation of relevance
Solution Approach 1:
The patent transforms relevance scores from different corpora by applying parameter changes through normalization. Each corpus-specific relevance score is converted to a standardized scale using learned parameters that account for the unique distribution and meaning of scores in each corpus, enabling consistent comparison across diverse content sources while preserving the original scoring characteristics of each corpus
2Quantity of substance
If content from multiple corpora is blended in search results, then users receive more comprehensive and diverse results, but the varying score ranges make accurate comparison and ranking difficult
Solution Approach 1:
The patent introduces an intermediary normalization layer that mediates between diverse corpus-specific relevance scores and the unified ranking requirement. This intermediary component learns the relationship between different score distributions and transforms them into a common scale, enabling accurate ranking of blended content from multiple corpora without losing the distinctive relevance characteristics of each source
Data Source
AI summary
Techniques include obtaining ranges of content relevance scores for different collections of content; generating a normalized range based on the ranges of content relevance scores; and normalizing a particular range of a particular collection of content including: generating a distribution of content relevance scores for the collection of content; identifying portions in the distribution; and generating a mapping of portions from the distribution to portions in the normalized range.