Multi-Modal Adversarial Autoencoders for Recommendations of Citations and Subject Labels
This work addresses recommendation tasks in academic and information retrieval domains, but it is incremental as it builds on existing autoencoder methods with adversarial regularization.
The paper tackles the problem of improving recommendation systems for citations and subject labels by using multi-modal adversarial autoencoders, showing that adversarial regularization consistently enhances performance across 408 experiments, with results indicating that item co-occurrence semantics (relatedness vs. diversity) critically affect model effectiveness.
We present multi-modal adversarial autoencoders for recommendation and evaluate them on two different tasks: citation recommendation and subject label recommendation. We analyze the effects of adversarial regularization, sparsity, and different input modalities. By conducting 408 experiments, we show that adversarial regularization consistently improves the performance of autoencoders for recommendation. We demonstrate, however, that the two tasks differ in the semantics of item co-occurrence in the sense that item co-occurrence resembles relatedness in case of citations, yet implies diversity in case of subject labels. Our results reveal that supplying the partial item set as input is only helpful, when item co-occurrence resembles relatedness. When facing a new recommendation task it is therefore crucial to consider the semantics of item co-occurrence for the choice of an appropriate model.