Heterogeneous Distance Functions for Data-Type-Aware Recommendations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing recommendation systems struggle to accurately measure distances between data items of different types without losing information by converting them to a single type, leading to suboptimal recommendations.
Innovation Solution
A parameterized heterogeneous distance function is employed that utilizes multiple sub-functions tailored to different data types, allowing direct measurement of distances between data items with diverse feature types, and is customized based on user behavior to enhance recommendation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If data items of different types are converted to a single type for distance measurement, then the measurement process becomes simplified, but information loss occurs
Solution Approach 1:
The patent changes the parameter of the distance function from a single unified metric to multiple type-specific metrics. Each data type (categorical, numerical, text) has its own distance calculation function with appropriate parameters, allowing accurate measurement without conversion and preventing information loss while maintaining operational simplicity through automated parameter selection.
Solution Approach 2:
The patent segments the distance measurement process into multiple specialized sub-functions, each handling a specific data type. Instead of converting all data to one type, the system divides the measurement task into categorical distance calculation, numerical distance calculation, and text distance calculation, preserving the unique characteristics of each data type while simplifying the overall process through modular design.
2Device complexity
If a single distance function is used for all data types, then the system complexity is reduced, but recommendation accuracy deteriorates
Solution Approach 1:
The patent creates a universal distance measurement system that handles multiple data types through a single heterogeneous distance function. This function automatically detects the data types of input features and selects appropriate distance metrics for each type, achieving both low system complexity and high recommendation accuracy by making the distance function multi-functional rather than requiring separate functions for each data type.
Solution Approach 2:
The patent makes the distance function dynamic by enabling it to adapt its behavior based on the input data types. The function dynamically selects which distance metrics to apply (cosine similarity for text, Euclidean for numerical, Jaccard for categorical) based on the characteristics of the data being measured, thereby maintaining simplicity while achieving high precision through adaptive behavior.
3Ease of operation
If data types are standardized to a single format, then processing becomes easier, but unique characteristics of different data types are lost
Solution Approach 1:
The patent applies local quality by assigning specific distance measurement properties to specific data types rather than using a uniform approach. Each data type (categorical, numerical, text) receives a distance function tailored to its characteristics, allowing the system to process data easily while preserving the unique qualities of each type through specialized measurement approaches.
Data Source
Figure 1~2
Figure 3~5
Figure 6
AI summary
Systems, methods, devices, and computer readable storage media described herein provide techniques for generating recommendations utilizing a heterogeneous distance function. In an aspect, a measure of relevancy between a first data item and a second data item is received. A setting of an adjustable parameter of a parameterized heterogeneous distance function is determined based on the measure of relevancy. The parameterized heterogeneous distance function comprises first and second sub-functions. The first sub-function calculates a distance between data items based on features of a first data type and the second sub-function calculates a distance between data items based on features of a second, different, data type. A recommendation system is caused to utilize the parameterized heterogeneous distance function to generate a recommendation based on received input. In a further aspect, the measure of relevancy is determined as a function of a measure of interactions and a measure of impressions.