Abstract :
This paper propose a novel context modeling
method Contextual Operating Tensor model, named COT,
which is motivated by the recent work of semantic
compositionality in Natural Language Processing (NLP). It
provide an efficient implementation inspired by the
powerful ability in describing latent properties of words, in
recommender systems, using a vector representation of each
context value seems a good solution to examine the effect
of contexts on user-item interactions. This distributed
representation inferred from all contexts has more powerful
ability in illustrating the operation properties of contexts.
Moreover, in the research direction of sentence sentiment
detection, a noun has semantic information as a latent
vector, and an adjective has semantic operation on nouns as
an operating matrix .This paper assume that contexts in
recommendation systems have a similar property of
adjectives and can operate latent characteristics of users and
items. Then, new latent representations of entities can show
not only characteristics of original entities but also new
proprieties under a specific contextual situation.
Keyword :
Natural Language Processing, Context Operating Tensor, latent representation, Recommender systems.