Recommender Systems
Are We Really Making Much Progress? (Dacrema, Cremonesi, and Jannach, n.d.)
The authors analyzed various recent publications on recommendation systems techniques, and found that these have:
- Weak baselines
- Establish weak methods as baselines
- Are outperformed by simple, sometimes non-neural approaches
The simple approaches that work well include ItemKNN, a collaborative-filtering approach that uses k-nearest neighbours and item-item similarities:
\begin{equation} s_{ij} = \frac{r_i \dot r_j}{\lvert r_i \rvert \lvert r_j \rvert + h} \end{equation}
To alleviate these issues:
- Use appropriate evaluation methods
- Evaluate on appropriate datasets (size is important)
- Release reproducible code
Bibliography
Dacrema, Maurizio Ferrari, Paolo Cremonesi, and Dietmar Jannach. n.d. “Are We Really Making Much Progress? a Worrying Analysis of Recent Neural Recommendation Approaches.” http://arxiv.org/abs/1907.06902v1.