The recommendation systems are popular and beneficial to both service providers and users. It helps to increase profits costs for many service providers. So, we need a recommendation system because of the overpopulation. It helps people to choose the best product or services in highly populated situations. Further, it helps to decide which movies to watch, which song to listen to, which news to read, which hotel to eat, etc. The recommendation systems gather a huge amount of information about users’ preferences of several items like online shopping products, movies, tourism, restaurants, etc. It records user reviews for watched movies, purchased products, etc. Recommendation systems are widely used in various fields or various domains such as health, entertainment, sports, e-commerce, media, etc. This paper provides a descriptive overview of the recommendation system for movies from one of the domains of entertainment. To recommend movies, the system first collects the ratings for target users and then recommends the top list of items to the users. They can check existing reviews before watching a movie.
In the previous paper, they used the collaborative filtering method, content-based method, and hybrid method for recommending movies used by the supervised machine learning algorithm. It helped to analyze and derive data from the dataset for recommending movies. We got a lot of information and ideas from the previous paper. It helps to find results when combining our final data with the earlier research. They obtained the results for recommending movies based on the final reports of data. Now we are approaching to recommend movies based on the same as the from the previous paper. But we have proposed some technical changes in the earlier paper and we will predict the results. Then combining this paper with the previous one, we will find out the similarities or differences in them.
Data analysis is a major problem not only in recommendation systems but also in various fields. Data analysis is difficult, manually. Apart from manpower, we need machines for analyzing and an assortment of data features. By using computers we analyze data and get better results which process is called machine learning algorithm. The algorithms are classified into four types. 1. Supervised machine learning 2. Unsupervised machine learning 3. Semi-supervised machine learning and 4. Reinforcement. There are several types in each type of machine learning algorithm. The system has used the algorithms for data analysis and to get the result.
We have introduced the movie recommendation system by using machine learning algorithms, which depend on the system’s acceptability. This paper on movie recommendation systems tells how to solve and overcome the problem. Here, many methods are implemented to get accuracy for recommending movies.
1. Eyjolfsdottir, E. A.et al., introduced an expert system for movie recommendation is movieGEN. Here, the machine learning algorithm and cluster analysis are used to implement the system based on a hybrid recommendation method. A perfect SVM (support vector machine) helps to provide users with personal information and predicts their movie preferences for the movie recommendation system. The movie selects from the data set, clusters the movies and generates questions to the users by Support Vector Machine (SVM). Finally, movies are recommended for users based on their refined movie set of users’ answers. The system has been customizable by the process of traversing the parameter space [1].
Goyani and Chaurasiya proposed recommendation system is much more popular in main areas and it helps to take proper decisions for people. The beneficial information is found from the various data available for users by the recommendation method. There are two types of following recommendation methods are collaborative and content-based filtering. Collaborative filtering is similarities between users and the content-based filtering method is depends on a particular user’s activity. A better recommendation system has been done by hybrid recommendation method when overcoming the limitations of collaborative filtering and content-based filtering method. Found the similarity between the users for recommendation through utilizing various similarity measures. Different similarity measures had reviewed in the paper such as Facebook which recommends friends, LinkedIn which recommends jobs, Pandora recommends music, Netflix recommends movies, Amazon recommends products, etc., The companies profit and customers benefits increased by using the recommendation system.
Zhang. J, et al., developed a novel based on collaborative filtering approaches and it’s called KM-slope-VU. It is used for fast and scalable movie recommendations. Also, the movie recommendation site has been personalized. The proposed algorithm is used in real-life data to collect and evaluate users' feedback on the recommended movie. Particularly, they have adopted K-means from an unsupervised machine algorithm to divide users into several clusters and the users keep that in each cluster to represent their virtual opinion leader. However, the work represents each whole cluster based on conceived virtual opinion leaders. The proposed algorithm’s prediction accuracy is lower than SVD (singular value decomposition). Because of some outdated movies and the small number of data samples collected by the movie recommendation system. The real feedback evaluation was obtained from the MovieLens dataset. It’s higher than the performance. They have given two methods for future work. First one is to make the latest films for users instead of outdated movies in the dataset. And second one is to optimize the virtual users for better represents of entire real users in a cluster by using fuzzy C-means, which could further improve the recommendation accuracy
You must be logged in to post a comment.