Browse Teach Curate

Item Association Graphs in Markov-Chain Sequential Recommendation

A sequential-recommendation model that predicts a user's next item by fusing an item association graph into a mixed Markov chain. Building on Fossil (FISM item-similarity factorization for long-term interest plus a Markov chain for short-term interest), it factorizes item-item association information and integrates it into the item sequence representation, improving next-item ranking with few extra parameters and easing data sparsity and cold-start.

Rosie Text

Item Association Factorization Mixed Markov Chains for Sequential Recommendation DongYu Du, Yue

Item Association Factorization Mixed Markov Chains (IAFMC) is a sequential-recommendation algorithm that predicts a user's next item from their historical behavior sequence. It extends Fossil, which …