In an era when recommender systems aspire to reduce information overload, we analyze how recommender systems can be implemented to overcome current limitations. This thesis presents a novel framework for a semantic recommender system that not only copes with existing problems but also presents a strategy that computes customized recommendations using a variety of tools including semantic contents. To this end, we have identified the need for developing semantic recommender systems, which are able to extend existing systems and perform a semantic search in an effort to find the most suitable scientific papers in the field of Computer Science. For this purpose, we developed three different semantic recommender techniques rooted in annotation systems and its semantic matching components. The techniques, which are entitled REI, REII, and REIII, are based on GATE, Alchemy API, and a combination of both tools. These recommender techniques are capable of exploring an annotated database in an attempt to trace and rank the most relevant documents in a particular query. Precision and recall are subsequently measured and compared to a similar query conducted in Google Scholar indicating that this research is promising and can improve on current semantics-based recommender systems.
Zircon - This is a contributing Drupal Theme Design by WeebPal.
