Understanding Latent Factors Using a GWAP

TL;DR

This paper introduces a game-based method to automatically generate semantic descriptions for latent factors in recommender systems, making them more interpretable for users.

Contribution

It presents a novel output-agreement game that effectively captures real-world characteristics of latent factors through user participation.

Findings

Collected descriptions align with real-world factor characteristics

User study validates the effectiveness of the game approach

Enhances interpretability of latent factor models

Abstract

Recommender systems relying on latent factor models often appear as black boxes to their users. Semantic descriptions for the factors might help to mitigate this problem. Achieving this automatically is, however, a non-straightforward task due to the models' statistical nature. We present an output-agreement game that represents factors by means of sample items and motivates players to create such descriptions. A user study shows that the collected output actually reflects real-world characteristics of the factors.