Iterative Image Search Algorithm Using Continuous Human Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image search engines rely solely on user input and fail to guide users to their current object of interest, leading to inefficient searches, excessive network bandwidth usage, and frustration, as they do not allow for iterative discovery based on user feedback and emotional responses.
Innovation Solution
A computer-implemented search algorithm that displays one image at a time, continuously receives user feedback, and processes tags associated with images to iteratively present the next image, allowing for quick identification of the user's desired object by dynamically adjusting search criteria based on positive or negative preferences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional image search engines display many images at once, then the user has more options to choose from, but the user must sift through hundreds or thousands of images which increases time consumption and frustrates the user
Solution Approach 1:
The patent segments the image search results into a sequence of individual images presented one at a time rather than displaying all images simultaneously. This segmentation allows the system to maintain versatility in search options while reducing the cognitive load and time required for users to evaluate results, as they only need to focus on one image at a time with clear navigation controls.
Solution Approach 2:
The patent implements dynamic interaction where the system adapts to user responses in real-time. The search sequence dynamically adjusts based on user feedback (such as liking or disliking an image), allowing the system to learn and refine results during the interaction. This dynamic approach maintains adaptability while reducing time loss by eliminating the need to manually sift through static batches of images.
2Measurement precision
If conventional image search engines require users to re-enter search terms when initial results are not relevant, then the search can be refined, but previously downloaded images are discarded and network bandwidth is consumed repeatedly
Solution Approach 1:
The patent implements continuous feedback loops where user responses to each image (such as liking, disliking, or skipping) are immediately processed to adjust the search sequence. This feedback mechanism allows the system to refine search relevance incrementally without requiring complete re-searches, thereby maintaining measurement precision while significantly reducing network bandwidth consumption by building upon previously downloaded images.
Solution Approach 2:
The patent performs preliminary actions by downloading and caching images in advance as a searchable sequence before the user needs them. When users provide feedback, the system refines results from this pre-downloaded set rather than triggering new downloads, thus maintaining search precision while minimizing energy loss through repeated network transactions.
3Measurement precision
If conventional image search engines rely on precise search parameters, then the search can be accurate, but users who are not precisely certain what they are searching for cannot effectively use the system
Solution Approach 1:
The patent enables self-service by allowing the system to automatically refine and adjust search results based on user feedback without requiring users to manually adjust search parameters. Users simply indicate their preferences through simple interactions (like/dislike), and the system autonomously optimizes the search sequence, thereby maintaining search accuracy while dramatically improving ease of operation for users uncertain about their search criteria.
Solution Approach 2:
The patent creates a dynamic search experience where the system adapts to user needs in real-time. Rather than requiring precise static parameters upfront, the search dynamically evolves based on user responses, automatically adjusting the sequence of images presented. This dynamic approach maintains measurement precision through continuous refinement while improving ease of operation by removing the burden of parameter specification from the user.
Data Source
AI summary
System and computer-implemented image search engine of analyzing tags associated with a sequence of images presented to a user to present a current object of interest of the user is disclosed. An image from among a plurality of images is presented on an electronic display. The image is associated with a set of tags. An input is received indicating a user's preference for the image. A plurality of tags is processed based on the preference and the set of tags to determine a next set of tags from the plurality of tags. A next image is determined from the plurality of images based on the next set of tags. The next image represents a physical object, different from a physical object represented by the previous image. A sequence of images is generated by repeating the above process with the next image in place of the previous image for present a user's current object of interest.