Item Recommendations via Concept Relatedness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing recommendation systems often provide either over-inclusive or under-inclusive results, failing to deliver personalized content to users due to reliance on hierarchical taxonomies and lack of relevant data for new items, leading to inefficient user experiences and resource utilization.
Innovation Solution
A computer system that utilizes a non-hierarchical taxonomy to determine item concept relatedness scores, allowing for more accurate recommendations by classifying and comparing items at an appropriate level of abstraction, thereby providing relevant content based on user preferences and behavior.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If hierarchical taxonomy is used for item classification, then system complexity is reduced and ease of operation is improved, but recommendation precision and relevance deteriorate due to over-inclusive or under-inclusive results
Solution Approach 1:
The patent segments the item classification system into multiple independent concept vectors, where each vector represents a distinct aspect or attribute of items. Instead of using a single hierarchical taxonomy, the system creates separate concept spaces for different dimensions (e.g., product type, brand, feature, usage scenario), allowing for more nuanced and precise recommendations by combining relevant concepts from multiple vectors.
Solution Approach 2:
The patent transitions from a single-dimensional hierarchical taxonomy to a multi-dimensional concept space. Each item is represented by multiple concept vectors that capture different aspects simultaneously. This dimensional expansion allows the system to navigate complex item relationships without the limitations of hierarchical depth, providing both precision and operational efficiency.
2Measurement precision
If more data points are collected for new items, then recommendation relevance improves, but computing resources and processing time increase
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing concept vectors for all items in the database before they are needed for recommendations. These concept vectors capture essential item characteristics in advance, allowing the system to generate relevant recommendations for new items with limited data by comparing them against the pre-established concept space, thereby avoiding intensive real-time computing.
Solution Approach 2:
The patent uses copying by creating concept vector representations that capture the essential characteristics of items. Instead of processing raw item data repeatedly, the system creates compressed concept vector copies that preserve key information while requiring minimal computational resources for comparison and recommendation generation.
3Device complexity
If hierarchical taxonomy is used, then data structure simplicity is improved, but adaptability to diverse item concepts and user preferences deteriorates
Solution Approach 1:
The patent introduces dynamics by making the concept vector structure flexible and adaptable. Unlike rigid hierarchical taxonomies, the concept vector model allows for dynamic addition, removal, and modification of concept dimensions based on emerging item types and user preferences. The system can adapt to new categories and relationships without restructuring the entire data model.
Solution Approach 2:
The patent applies universality by designing a concept vector framework that can represent diverse item concepts across multiple domains using the same underlying structure. Each concept vector serves multiple functions: it enables classification, similarity measurement, recommendation generation, and user preference modeling, making the system highly versatile without requiring domain-specific data structures.
Data Source
AI summary
Techniques are provided for determining that an item should be provided for presentation to a user based on an item concept relatedness score. For example, a computer system may partition a plurality of items into a plurality of item concepts, whereby each item is assigned to one and only one item concept. The computer system may receive a list of co-selected items, whereby each item of the list is assigned to an item concept. Based at least in part on this list, the system may determine an item concept relatedness score between a first item concept and a second item concept of the plurality of item concepts. The computer system may then receive a selection of a first item assigned to a first item concept. Based at least in part on the item concept relatedness score, the system may provide the second item for presentation to a user.