Knowledge Enhanced Translation-based User Preference Model (KTUP)
Patent Name: Knowledge Enhanced Translation-based User Preference Model (KTUP)
Filing No: 10201905197X (SG)
Filing Date: 6 Jul 2019
Country to be Filed: SG
Description: This invention proposes a novel model Translation-based User Preference Model (TUP), which specially accounts for various preferences in translating a user to an item. We further enhance TUP by introducing Knowledge Graph (KG), namely KTUP. The key idea is that there exists multiple (implicit) relations between users and items, which reveal the preferences (i.e., reasons) of users on consuming items. KTUP endows the preferences with explicit semantics with relation types in KG, capturing the intuition that the type of item attributes plays a crucial role in user decision-making process. Technically speaking, we transfer the relation embeddings as well as entity embeddings learned from KG to TUP, simultaneously training the KG completion and recommendation tasks. Two tasks shall be mutually enhanced via jointly training. Extensive experiments on two datasets demonstrate the effectiveness and interpretability of our proposed model.