Image Search Term Scoring via Navigation Probabilities
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image search engines face degradation in search result quality due to uninformative or poorly labeled images, where keywords extracted from surrounding text may not accurately represent the image content.
Innovation Solution
A method and system that determine scores for images with respect to terms by calculating probabilities of navigating between images, identifying relevant terms based on these scores, and associating them with the images, thereby improving the relevance of image search results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If keywords are extracted from text surrounding the image, then the image can be indexed and retrieved, but the keywords may be uninformative or not useful for accurately representing image content
Solution Approach 1:
The patent replaces the mechanical text-extraction method with a probabilistic navigation-based scoring system. Instead of relying on surrounding text keywords, the system calculates scores based on navigation probabilities between images, effectively substituting a computational model that better captures actual image relevance and content accuracy
Solution Approach 2:
The patent changes the parameter used for keyword selection from text-based frequency or relevance to navigation probability-based scoring. By transforming the selection criterion from static text analysis to dynamic navigation behavior data, the system identifies terms that truly reflect image content and user interaction patterns
2Adaptability or versatility
If all extracted terms are associated with images, then comprehensive indexing is achieved, but search result quality degrades due to inclusion of uninformative terms
Solution Approach 1:
The patent applies local quality by assigning different scores to different terms based on their individual navigation probability metrics. Instead of uniformly including all terms, each term is evaluated and weighted according to its specific relevance score, allowing high-quality terms to be associated with images while filtering out uninformative ones
Solution Approach 2:
The patent introduces a scoring mechanism as an intermediary between term extraction and term association. This intermediary layer calculates navigation probabilities and applies selection criteria, acting as a filter that separates useful terms from uninformative ones before final association with images
Data Source
AI summary
The subject matter of this specification can be embodied in, among other things, a method that includes determining a score for an image of a plurality of images with respect to each of one or more terms, identifying one or more of the terms for each of which the score for the image with respect to the respective identified term satisfies a criterion, and associating the identified terms with the image. Determining the score for the image with respect to a respective term includes determining probabilities of navigating between images in the plurality of images and determining the score for the image with respect to the respective term based on the probabilities.