Query modification based on non-textual resource context
Topics: Image Search, Ranking, Scoring, Search Query Processing, Semantic Search, Video Search
This Google patent describes a system designed to improve search accuracy by looking at the non-textual content a user is currently viewing, such as an image or a video, when they submit a search query. By identifying the visual context from the user’s device, the system can take a simple or ambiguous search term and modify it into a more specific query. This allows the search engine to provide highly relevant results that match what the user is seeing in real-time without requiring them to type a long, detailed description.
