Implementation of a subjective evaluation application for recommender systems

Nowadays, a huge amount of information is available online, meaning that seeking an element in the World Wide Web, instead of being a easy way to collect data, could turn into a laborious job which would include the visualization of many irrelevant pages. Recommender Systems aims to facilitate the browsing through long catalogs, selecting a limited number of items (i.e. films, newspaper articles, books) according to the preferences identified to the active user. The functionality of this type of system rests on the output given by its algorithm, the core of the recommendation engine. Currently, there are many kinds of algorithms applicable; furthermore, as a field of interest of many (specially in the commercial websites case), it has been established a continuous search for new and more accurate ones. Given this context, the present work regards the development of a subjective evaluation application for recommender systems, which will permit to carry out surveys among groups of users, large enough to estimate the reliability and efficiency of recommendation algorithms. Not only the implementation of a user interface to perform questionnaires about films, called Milo Web Application, was accomplished but also a self­sufficient recommendation engine, named Whisperer, was deployed, implying that the last one can be imported and used by other web applications, independently from the context its inserted (not necessarily the movie scope).