Search Engine Relevance Scoring via Word Prominence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional search engines often return irrelevant results due to relying solely on the quantity of matching words, rather than the relevance of those words, when matching item descriptions with search phrases, leading to items with little or no actual relevance being identified.
Innovation Solution
A search engine system that assigns prominence scores to words in item descriptions, determining relevance scores based on these scores, and uses these scores to filter and rank matching items, ensuring that only items with high relevance scores are returned, with the option to set a threshold that requires the headword to appear in the search phrase.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the search engine relies solely on the quantity of matching words to identify relevant items, then the quantity of matching words is maximized, but the relevance of search results deteriorates
Solution Approach 1:
The patent changes the parameter used for matching from simple word quantity to a weighted relevance score that incorporates prominence scores of individual words. This transforms the matching criterion from a quantitative measure to a quality-based measure, resolving the contradiction by prioritizing relevance over quantity.
Solution Approach 2:
The patent assigns different prominence scores to different words within item descriptions, treating each word with its specific importance weight rather than uniform treatment. This local differentiation allows the system to identify relevant items based on the significance of matching words rather than merely their quantity.
2Quantity of substance
If the search engine returns items with the highest quantity of matching words, then the quantity of search results is maximized, but the precision of matching deteriorates
Solution Approach 1:
The patent changes the sorting parameter from word quantity to relevance score, which incorporates the prominence scores of matching words. This ensures that search results are ordered by precision rather than quantity, resolving the contradiction between result quantity and matching precision.
3Measurement precision
If the search engine uses a threshold based on headword prominence scores, then the precision of relevance filtering is improved, but the quantity of matching items decreases
Solution Approach 1:
The patent introduces a threshold mechanism based on headword prominence scores to filter results. This parameter-based filtering improves precision by ensuring that only items with sufficiently relevant headwords are returned, accepting a trade-off in quantity for gains in precision.
Data Source
AI summary
A method may include receiving, at a search engine, a search phrase. The search engine may query a database in response to receiving the search phrase. The search engine may query the database to retrieve, from the database, one or more item descriptions matching the search phrase. The one or more item descriptions may include an item description based on a relevance score of the item description. The relevance score of the item description may correspond to a prominence score of each word in the item description that also appear in the search phrase. The search engine may generate a search result that includes items associated with the item descriptions retrieved from the database. Related systems and articles of manufacture, including computer program products, are also provided.