Joint Item Attribute Selection for Informational Displays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current informational displays, such as search results pages, face challenges in optimizing the relevance and utility of items and their attributes due to limited space, often resulting in redundancies and lack of coherence when separately ranking items and attributes, failing to consider dependencies between them.
Innovation Solution
A machine-learned display selection model is trained to jointly select items and attributes for inclusion in informational displays, using a nested submodular objective function to optimize the selection process, ensuring coherence and minimizing redundancy by modeling dependencies across items and attributes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If separate ranking techniques are used to rank items and attributes independently, then the selection process is simpler and more computationally efficient, but the overall coherence and relevance of the informational display deteriorates due to redundancies and lack of consideration for dependencies between items and attributes
Solution Approach 1:
The patent combines the separate ranking processes for items and attributes into a single joint ranking model. The machine learning model simultaneously considers both items and their attributes together, optimizing their selection as a unified structure rather than independent components, thereby maintaining coherence and relevance while managing computational complexity through integrated processing.
Solution Approach 2:
The joint ranking model serves multiple functions simultaneously: it ranks items based on relevance, selects appropriate attributes for each item, and ensures coherence across the entire display structure. This multi-functional approach replaces the need for separate specialized ranking processes while achieving better overall display quality.
2Reliability
If a joint machine learning model is used to simultaneously select items and attributes, then the coherence and relevance of the informational display is improved by considering dependencies between them, but the device complexity and computational requirements increase
Solution Approach 1:
The joint ranking model is segmented into distinct processing components that handle different aspects of the ranking task. The model processes items and attributes through separate but coordinated pathways, allowing for manageable complexity while maintaining the joint optimization framework. This segmentation enables the system to handle the computational burden in modular fashion.
Solution Approach 2:
The patent employs parameter changes by transforming the complex joint ranking problem into a more manageable form through feature engineering and parameter optimization. The machine learning model adjusts parameters dynamically to balance the complexity of joint processing with computational efficiency, enabling the system to handle the increased model complexity through adaptive parameter tuning.
3Ease of operation
If limited space is allocated in the informational display, then the presentation is more concise and user-friendly, but the number of items and attributes that can be displayed is reduced
Solution Approach 1:
The joint ranking model implements partial action by selectively displaying only the most relevant items and attributes within the limited space. Rather than attempting to display all available information, the model prioritizes and presents a curated subset that maximizes relevance and user utility, achieving effective communication without overwhelming the user with excessive content.
Solution Approach 2:
The model dynamically adjusts parameters such as the number of items to display and the selection of attributes based on relevance scoring and space constraints. This parameter optimization allows the system to maximize the information density within limited space while maintaining user-friendliness and relevance.
Data Source
AI summary
The present disclosure provides systems and methods that use machine learning to improve whole-structure relevance of hierarchical informational displays. In particular, the present disclosure provides systems and methods that employ a supervised, discriminative machine learning approach to jointly optimize the ranking of items and their display attributes. One example system includes a machine-learned display selection model that has been trained to jointly select a plurality of items and one or more attributes for each item for inclusion in an informational display. For example, the machine-learned display selection model can optimize a nested submodular objective function to jointly select the items and attributes.