2019/07/13 by Dai Quoc Nguyen, Nguyen, Dai Quoc, Tu Dinh Nguyen +3
Computer Science · #Advanced Graph Neural Networks #Computation and Language (cs.CL) #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Recommender Systems and Techniques
paper · pdf · doi:10.48550/arxiv.1907.06080
openalex publication_date 2019/07/13 · openalex created_date 2022/07/28 · openalex updated_date 2026/07/28
Knowledge graph embedding methods often suffer from a limitation of\nmemorizing valid triples to predict new ones for triple classification and\nsearch personalization problems. To this end, we introduce a novel embedding\nmodel, named R-MeN, that explores a relational memory network to encode\npotential dependencies in relationship triples. R-MeN considers each triple as\na sequence of 3 input vectors that recurrently interact with a memory using a\ntransformer self-attention mechanism. Thus R-MeN encodes new information from\ninteractions between the memory and each input vector to return a corresponding\nvector. Consequently, R-MeN feeds these 3 returned vectors to a convolutional\nneural network-based decoder to produce a scalar score for the triple.\nExperimental results show that our proposed R-MeN obtains state-of-the-art\nresults on SEARCH17 for the search personalization task, and on WN11 and FB13\nfor the triple classification task.\n