All the people rely on recommendations from other people. These recommendations can be either by reviews of movies and books that are printed in newspapers and magazines, recommendation letters, or general surveys. This is called a social process that can be assisted and augmented by recommender systems. Recommender systems that have become increasingly popular are a way for people to make their decisions. These decisions are based on other individualsā recommendations and choices can be made without sufficient knowledge or experience of alternatives. Typically recommendations are considered as inputs of recommender systems.
MovieIt recommender system is such a system that uses Facebook infrastructure to allow its users to talk about the movies that they have watched, suggest them to their friends and also the place they watched the movie, etc. Recommendations from friends have been the most powerful way to find about new things such as new movies.
Facebook is a social network that people can establish or maintain connection with others. With MovieIt, Facebook users can log in their Facebook accounts and suggest their friends the movie they have recently watched and also the place they watched the movie.
An important factor in recommender systems is the person who provides the recommendation. It is essential to know if we have similar tastes with the person who recommends a movie, book, etc. In big groups of people that they do not know each other, it is valuable to automatically match people with similar tastes.
In MovieIt case, people provide information to their Facebook friends. Although Facebook is a platform to meet new people, Facebook participants mostly interact with people they already know and with the same age range, such as an old friend or a classmate. This means that the probability of having people among your friends with similar tastes is high. So when a MovieIt user watches a movie, enjoys it and recommends it to his/her friends, it is likely that many of his/her friends enjoy it too.
There are some commonly used algorithms in recommender systems. Two of them are mentioned following:
1. K-nearest neighbourhood approach: With the calculation of Pearson Correlation, a particular user’s neighbourhood with similar taste or interest can be found in a social network. This can be done with the collection of data of top-N nearest neighbours of the particular user. By using certain techniques, the data can be calculated to predict the preference of the user.
2. Collaborative Filtering: A large amount of information about behaviour, activity or preferences of users will be collected and analysed. What a user will like will be predicted by a comparison between the user and other users with similar tastes. Item-to-item collaborative filtering is one of the most common types of collaborative filtering. Another type of collaborative filtering that uses the social process of asking a friend for recommendation is called user-based collaborative filtering.
Collaborative filtering recommender systems provide suggestions for users according to the information which is almost quite relevant to them. It has been proved that these systems are one of the most effective techniques which help people to find the information valuable to them.
Cyber Tube somehow uses the last type of collaborative filtering called user-based collaborative filtering. Cyber Tube is a social system that applies a social process of friends providing recommendation for each other in a social infrastructure, Facebook.