Image Browsing via Mined Hyperlinked Text Snippets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current search engines lack the ability to provide users with semantically relevant and interesting textual information for images, making it difficult for users to efficiently locate images that match their interests or needs.
Innovation Solution
An image browsing framework that extracts semantically relevant snippets of text from webpages, detects keyterms, and generates relevance and interestingness scores for sentences, allowing users to browse images through a graphical user interface with hyperlinked keyterms and snippets, optimizing the image search process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional search engines are used to search for images, then users can access a vast array of online information resources, but users cannot efficiently locate images that match their interests or needs due to lack of semantically relevant textual information
Solution Approach 1:
The patent segments the webpage content into multiple components: image elements, text snippets, keyterms, and hyperlinks. Each component is processed and stored separately, allowing users to access specific semantic information (text snippets and keyterms) that is directly relevant to the image being searched, thereby improving search efficiency without losing important contextual information.
Solution Approach 2:
The patent introduces an intermediary processing layer that extracts semantically relevant text snippets and keyterms from webpages, and creates hyperlinked connections between these snippets and related images. This intermediary layer acts as a bridge between traditional search engines and users, providing enriched semantic information that helps users efficiently locate relevant images.
2Ease of operation
If text snippets and keyterms are extracted and hyperlinked for each image, then users are provided with semantically meaningful information, but the complexity of the image browsing system increases
Solution Approach 1:
The patent performs preliminary actions by pre-extracting text snippets, identifying keyterms, and creating hyperlinked structures during the image crawling and indexing phase. This preliminary processing is done automatically in the background, so when users perform image searches, they immediately benefit from the pre-organized semantic information without experiencing the complexity of the processing operations.
Solution Approach 2:
The system performs self-service by automatically extracting semantically relevant text snippets, identifying keyterms, and creating hyperlinked connections without requiring manual intervention. The framework autonomously processes webpages, extracts information, and organizes it in a user-friendly format, reducing the operational complexity for users while improving browsing ease.
3Productivity
If relevance and interestingness scores are generated for sentences, then users can browse images more effectively, but the processing time and computational resources required increase
Solution Approach 1:
The patent changes the parameters of text evaluation by generating multiple scoring dimensions (relevance score, interestingness score) for each sentence. These parameter changes allow the system to rank and filter text snippets more effectively, presenting users with the most relevant and interesting information first, thereby improving search effectiveness while managing processing time through prioritized presentation.
Solution Approach 2:
The patent applies local quality by generating relevance and interestingness scores specifically for sentences that are locally associated with particular images and their contextual regions. Rather than uniformly processing all text, the system focuses computational resources on generating scores for sentences that are semantically and spatially relevant to specific images, improving effectiveness while reducing overall processing time.
Data Source
AI summary
Images stored in an information repository are prepared for browsing. For each image in the repository, text in the repository is mined to extract snippets of text about the image which are semantically relevant to the image, and for each of these snippets of text, keyterms are detected in the snippet of text which represent either concepts that are related to the image or entities that are related to the image, and the snippet of text and keyterms are associated with the image. Each keyterm that is associated with each image in the repository is hyperlinked to each other image in the repository that has this keyterm associated therewith. A graphical user interface allows a user to browse the images in the repository by using their associated snippets of text and hyperlinked keyterms.