Many Collaborative Filtering (CF) algorithms are item-based in the sense thatthey analyze item-item relations in order to produce item similarities.Recently, several works in the field of Natural Language Processing (NLP)suggested to learn a latent representation of words using neural embeddingalgorithms. Among them, the Skip-gram with Negative Sampling (SGNS), also knownas word2vec, was shown to provide state-of-the-art results on variouslinguistics tasks. In this paper, we show that item-based CF can be cast in thesame framework of neural word embedding. Inspired by SGNS, we describe a methodwe name item2vec for item-based CF that produces embedding for items in alatent space. The method is capable of inferring item-item relations even whenuser information is not available. We present experimental results thatdemonstrate the effectiveness of the item2vec method and show it is competitivewith SVD.
展开▼
