Media Unit Retrieval via Probability Distribution Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing media retrieval systems are inefficient as they rely on low-level image features that do not accurately represent user intent, and require explicit relevance rankings from users.
Innovation Solution
A system that maintains and updates a probability distribution based on user input to model the user's intention, selecting media units that are relevant to the user's target by sampling from this distribution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If low-level image features are used for media retrieval, then the system implementation is simple, but the retrieval relevance to user intent deteriorates
Solution Approach 1:
The patent transforms the feature representation parameters from low-level image features to semantic attributes that better represent user intent. This involves changing the parameter space to include semantically meaningful descriptors that capture the essence of what users are searching for, thereby improving retrieval relevance while maintaining system feasibility through automated semantic extraction.
2Measurement precision
If explicit relevance rankings are required from users, then feedback precision is improved, but user operation complexity increases
Solution Approach 1:
The patent applies partial action by requiring only implicit feedback (selection or non-selection of media units) rather than complete explicit rankings. This partial feedback is sufficient to update the probability distribution and guide the retrieval process, significantly reducing user effort while maintaining effective learning of user preferences through iterative refinement.
3Measurement precision
If iterative media presentation with explicit feedback is used, then retrieval accuracy is improved, but time consumption increases
Solution Approach 1:
The patent implements a feedback mechanism where implicit user selections update the probability distribution over semantic attributes. This feedback loop allows the system to iteratively refine its understanding of user intent and present increasingly relevant media units, achieving high retrieval accuracy through automated processing that minimizes time loss compared to manual explicit ranking.
Data Source
AI summary
Media unit retrieval methods, systems and computer program products are provided that allow a user to search for an item by iteratively presenting media units such as images representing items to the user and receiving user input consisting of selections of the presented media units (including possibly the empty selection). Features, or attributes, a user is interested in, for example semantic features, are inferred from the interaction and media units are retrieved for presentation based on similarity with user-selected media units, through sampling of a probability distribution describing the intent or interests, or combinations of approaches. Accordingly, the user-experience is akin to a conversation about what the user is looking for. Retrieval may be based on both selected and unselected media units and the selection may comprise making a selection with a single action. Further, a database of media units can capture similarity relationships for efficient media unit retrieval.