UTS - Movie Recommendation Systems: Collaborative & Content Based

This report provides an introduction to movie recommendation systems, highlighting their significance in addressing information overload and automating information filtering based on user interests. It delves into three primary recommendation methods: collaborative filtering, content-based filtering, and hybrid approaches. Collaborative filtering leverages the preferences of similar users, while content-based filtering relies on item descriptions and user profiles. Hybrid systems combine both methods to enhance accuracy and overcome limitations like cold start problems and sparsity. The report discusses the advantages and disadvantages of each approach, emphasizing the practical implementation of these methods in real-time movie recommendations based on user reviews and ratings. Desklib offers a platform for students to access this and other solved assignments.

INTRODUCTION
Research over intelligent agents of information attracted greater attention by the businesses,
which targeted on customers’ interests. In information highway of today, electronic markets’
information overload and automation are the relevant problems. It needs obtaining and
refining this tonnes of information from the internet to get the small and accurate piece of
information. Businesses have been spending meaningful efforts to automate filtering of
information to present organized information after to the user. Movie recommender system is
one of such techniques to refine the information according to the interests of the user.
Interactive agents make use of the models through gaining the feedback regarding movies
recommended by the users (Ricci et al., 2011). The objective of the movie recommendation
system is to understand and analyze, what kind of movies are most liked by the users, so that
the similar kind of movies are played for the users. There are many machine learning
techniques are implemented, such as decision trees, cluster analysis, artificial neural
networks, and Bayesian Classifiers.
There are several movie recommender methods and most commonly used are collaborative
filtering, content-based filtering and hybrid methods.
1. Collaborative Filtering Method
Collaborative filtering is the most commonly used approach to design recommender
system. This method is implemented on the basis of assumption that users, who agree
and like something will most likely to agree and like the same thing in the future. This
method is most commonly used for movie recommendation systems. By using the
same logic, most liked movies are found through the feedback by the users and the
same kind of movies is interested to play (Chen et al., 2011). This system generates
necessary recommendations through collecting information related to profiles of
rating from different users. Cold start problems can be dealt through nearest
neighbour algorithms, based on item and user and improve results of
recommendation, by using the data available. Collaborative filtering methods are of
two kinds model based and memory based.
2. Content-based Filtering Method
Content based filtering is one more designing recommender systems. These methods
have the basis of item description and preferences profile of the user. It is suitable for
the movie, since the there is enough data known about the movie, such as name,
casting, director of the movie are known, but the much about the user (Melville et al.,
2002). This method treats recommenders based on the movie content and treats
recommendation as classification problem, specific to the user and learns a classifier
for the likes and dislikes of user, on the basis of movie features.
3. Hybrid Recommender System
Hybrid recommender system is more developed and refined hybrid approach that
combines both the filtering methods, collaborative and content-based. It has multiple
approaches, implemented in different ways, through predictions based on
collaborative and content or through the approach of unifying multiple models into
one model. Hence, recommendation of movie makes use of content based and
collaborative approaches. Hybrid method provides movie recommendations more
accurately, rather than individual approaches. The best part is that the hybrid method
enables the recommender system to overcome certain regular issues in recommender

systems, like knowledge engineering bottleneck, sparsity problem and cold start
problems.
These three approaches are implemented in real time to recommend movies to watch for its
users, based on their movie preferences, such as movie reviews and movie ratings.
REFERENCES
Francesco Ricci, Lior Rokach and Bracha Shapira, (2011). Introduction to Recommender
Systems Handbook, Recommender Systems Handbook, Springer, pp. 1-35
Chen, H. Gou, L. Zhang, X. and Giles C. (2011). Collabseer: a search engine for
collaboration discovery, in ACM/IEEE Joint Conference on Digital Libraries (JCDL).
Prem Melville and Vikas Sindhwani, (2010). Recommender Systems, Encyclopedia of
Machine Learning.
Elahi, Mehdi, Ricci, Francesco and Rubens, Neil (2016). A survey of active learning in
collaborative filtering recommender systems. Computer Science Review
Prem Melville, Raymond J. M, and Ramadass Nagarajan. (2002). Content-Boosted
Collaborative Filtering for Improved Recommendations. Proceedings of the Eighteenth
National Conference on Artificial Intelligence (AAAI-2002), pp. 187–192, Canada:
Edmonton.