Understanding Latent Factors Using a GWAP

Kunkel, J., Loepp, B., & Ziegler, J. (to appear). RecSys ’18: Poster Proceedings of the 12th ACM Conference on Recommender Systems.


Recommender systems relying on latent factor models are well-known for generating accurate recommendations, but often appear intransparent for 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 players enjoyed the game while the collected output reflects real-world characteristics of the factors.


Related publications

Ein Online-Spiel zur Benennung latenter Faktoren in Empfehlungssystemen

Recommending Running Routes: Framework and Demonstrator

On User Awareness in Model-Based Collaborative Filtering Systems


Research development

A Game with a Purpose for Understanding Latent Factors

A game for eliciting descriptions for latent factors of matrix factorization

Focus area