Contextual Set Selection for Recommendation Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computer recommendation systems face challenges in efficiently selecting a group of items that are contextually coherent, as the computational complexity of analyzing interactions among tens of thousands of items is prohibitively high, especially when including expert judgment and user preferences, leading to suboptimal recommendations due to resource constraints.
Innovation Solution
A hybrid approach combining machine learning models, such as expected outcome prediction and expert judgement prediction models, with contextual set analysis, where the system decomposes selection probabilities into conditional probabilities and uses latent item representations to reduce computational load and enhance coherence in item selection, allowing experts to curate item groups efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning models are used to improve recommendation accuracy, then recommendation accuracy is improved, but computing resources required increase
Solution Approach 1:
The patent segments the recommendation process into multiple stages: generating candidate items using machine learning models, filtering candidates through contextual set analysis, and selecting final recommendations. This segmentation allows the system to use computationally intensive ML models only on subsets of items rather than the entire catalog, reducing overall computing resource requirements while maintaining recommendation accuracy.
Solution Approach 2:
The system performs partial analysis by generating a limited set of candidate items (e.g., top 100-1000 items) rather than analyzing all items in the catalog. This partial action approach applies machine learning models only to these candidates, significantly reducing computing resources needed while still achieving high recommendation accuracy through focused analysis of the most promising items.
2Measurement precision
If the system analyzes interactions among tens of thousands of items to ensure contextual coherence, then recommendation quality is improved, but computational complexity increases
Solution Approach 1:
The patent divides the item set into candidate items and non-candidate items, applying complex contextual analysis only to the candidate subset. The system segments the analysis process into candidate generation, contextual filtering, and final selection stages, reducing computational complexity by limiting intensive interactions to a smaller candidate set rather than analyzing all tens of thousands of items.
Solution Approach 2:
The system applies different levels of analysis depth to different items: machine learning models and contextual set analysis are applied deeply to candidate items, while non-candidate items receive minimal or no analysis. This local quality approach ensures high contextual coherence for recommended items while avoiding the computational complexity of analyzing all items equally.
3Measurement precision
If experts curate item groups manually to ensure quality, then recommendation accuracy is improved, but time consumption increases
Solution Approach 1:
The system performs preliminary actions by automatically generating candidate item sets using machine learning models and contextual set analysis before expert review. This preliminary processing narrows down tens of thousands of items to a manageable candidate set, allowing experts to focus their curation efforts on a smaller number of pre-filtered items, thereby maintaining recommendation accuracy while significantly reducing time consumption.
Solution Approach 2:
The system enables self-service by allowing experts to interactively refine and curate item groups based on automatically generated candidates. Experts can add, remove, or modify items in the recommendation set through user interface interactions, with the system automatically re-evaluating and adjusting the set. This self-service approach combines automated processing with expert judgment, improving accuracy while reducing overall time consumption compared to purely manual curation.
Data Source
AI summary
An initial list of candidate items automatically evaluated and chosen for an end-user is provided. A selection of one or more items in the initial list of candidate items is received from an expert user different than the end-user to include in an item group set for the end-user. Eligible items are evaluated to identify an additional item to include in the item group set based at least in part on an expert judgement prediction machine learning model trained to predict based at least in part on the one or more items already in the item group set a likelihood of a certain item being evaluated would be selected for inclusion in the item group set. Based on the evaluation, the additional item is included in the item group set. Member items in the item group set are indicated.