Utilising Semantically Rich Big Data to Enhance Book Recommendation Engines

This paper proposes a novel approach to book recommendation: we utilise big data created by thousands of book social cataloguing website users and treat it as a collectively written meta-annotation of a book. After learning semantic similarity between a large collection of books by applying algorithms of natural language processing (probabilistic topic models and semantic neural networks) to the symbolic text of meta-annotations, we show how to construct more precise and descriptive semantic recommendation engines and enrich content-based recommendation engines.