Explainable Restricted Boltzmann Machines for Collaborative Filtering

By Behnoush Abdollahi, Olfa Nasraoui. 2016

51 citations recorded by the source corpus.

Most accurate recommender systems are black-box models, hiding the reasoning behind their recommendations. Yet explanations have been shown to increase the user's trust in the system in addition to providing other benefits such as scrutability, meaning the ability to verify the validity of recommendations. This gap between accuracy and transparency or explainability has generated an interest in automated explanation generation methods. Restricted Boltzmann Machines (RBM) are accurate models for CF that also lack interpretability. In this paper, we focus on RBM based collaborative filtering recommendations, and further assume the absence of any additional data source, such as item content or user attributes. We thus propose a new Explainable RBM technique that computes the top-n recommendation list from items that are explainable. Experimental results show that our method is effective in generating accurate and explainable recommendations.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Knowledge-aware Autoencoders for Explainable Recommender Systems (2018, neutral): “Other interesting works include: a RS based on Restricted Boltzmann Machines which looks at the rating distribution to identify the most explainable items [1], a Latent Factor Model leveraging users reviews to compute more transparent recommendations [23] and, finally, a novel approach based on movi”
Explainable Recommendation: A Survey and New Perspectives (2018, methodology): “For the first approach, Abdollahi and Nasraoui [2017] adopted mean explainability precision (MEP) and mean explainability recall (MER) for evaluation. More specifically, explainability precision (EP) is defined as the proportion of explainable items in the top-n recommendation list relative to the n”
Using Explainability for Constrained Matrix Factorization (2017, methodology): “Recently, Abdollahi and Nasraoui [2] presented an explainable CF approach using Restricted Boltzmann Machines (RBM).”
Inverse Classification for Comparison-based Interpretability in Machine Learning (2017, background): “…Doshi-Velez and Kim 2017; Biran and Cotton 2017) usually distinguish approaches mainly on the characteristics of these explanations: given a classiﬁer to be interpreted, in-model interpretability relies on modifying its learning process to make it simpler (see for instance Abdollahi et al. 2016).”
A Survey on the Explainability of Supervised Machine Learning (2020, methodology): “The approach from Abdollahi and Nasraoui (2016) also uses neighborhood style explanations.”
Incorporating Interpretability into Latent Factor Models via Fast Influence Analysis (2018, unknown): “Later on, the idea was incorporated into the probabilistic method [3] and Restricted Boltzmann Machines [2] for interpretable recommendation under the collaborative filtering setting.”
TEM: Tree-enhanced Embedding Model for Explainable Recommendation (2018, methodology): “[1] used Restricted Boltzmann Machines to compute the explainability scores of the items in the top-K recommendation list.”
Explanation Mining: Post Hoc Interpretability of Latent Factor Models for Recommendation Systems (2018, background): “Abdollahi and Nasraoui [2] also explain Restricted Boltzmann Machines (RBMs) for collaborative filtering based on [32] by adding an additional visible layer with a node for each item with a value equal to the explainability score of an item to the active user.”