Resolving Cold Start Problem Using User Demographics and Machine Learning Techniques for Movie Recommender Systems
2
Citation
17
Reference
10
Related Paper
Citation Trend
Abstract:
There is a substantial increase in demand for recommender systems which have applications in a variety of domains. The goal of recommendations is to provide relevant choices to users. In practice, there are multiple methodologies in which recommendations take place like Collaborative Filtering (CF), Content-based filtering and Hybrid approach. For this paper, we will consider these approaches to be traditional approaches. The advantages of these approaches are in their design, functionality and efficiency. However, they do suffer from some major problems such as data sparsity, scalability and cold start to name a few. Among these problems, cold start is an intriguing area which has been plaguing recommender systems. Cold start problem occurs when the recommender system is not able to recommend new users/items since there is data sparsity. Researchers have formulated innovative techniques to alleviate cold start and the existing research conducted in this area is tremendous since the problem materializes in different use cases. Cold start is categorized into three problems. The first problem is when new users needs product recommendations from the system. The second problem is when new products listed in the system need to be recommended to existing users. The last problem is when new users and new products are present and the recommender engine needs to generate relevant recommendations. In this thesis, we concentrate on the first problem, where a user who is completely new to the system needs quality recommendations. We use a movie recommendation platform as our use case to analyze user demographics and find similarities between existing and new users to produce relevant recommendations.Keywords:
Cold start (automotive)
Demographics
Recommender systems provide users with product information and suggestions, which has gradually become an important research tool in e-commerce IT technology, which has attracted a lot of attention of researchers. Collaborative filtering recommendation technology has been the most successful recommendation technology so far, but there are two major problems-recommendation quality and scalability. At present, research at home and abroad mainly focuses on recommendation quality, and there is less discussion on scalability. The scalability problem is that as the size of the system increases, the response time of the system increases to a point where users cannot afford it. Existing solutions often result in a significant drop in recommendation quality while reducing recommendation response time. In this paper, the clustering analysis subsystem based on the genetic algorithm is innovatively introduced into the traditional collaborative filtering recommendation system, and its design and implementation are given. In addition, when obtaining the nearest neighbors, only the clustered users of the target user are searched, making it a collaborative filtering recommender system based on genetic clustering. The experimental results show that the response time of the traditional collaborative filtering recommender system increases linearly with the increase in the number of users while the response time of the collaborative filtering recommender system based on genetic clustering remains unchanged with the increase in the number of users. On the other hand, the recommendation quality of the collaborative filtering recommender system based on genetic clustering is basically not degraded compared with that of the traditional collaborative filtering recommender system. Therefore, the collaborative filtering recommender system based on genetic clustering can effectively solve the scalability problem of the collaborative filtering recommender system.
Cite
Citations (2)
The use of Collaborative Filtering is becoming very popular in designing a simple yet efficient recommender system. A recommender system based on Collaborative Filtering basically predicts a user's interest in some item on the basis of the scores generated and the correlation calculated between the users. In this paper we propose a basic structure and steps of designing a recommender system that uses Collaborative Filtering (user based) along with applications of partitioning and clustering of data, thus designing a Restaurant Recommender System. The proposed system reduces the complexity and gives a clear view of the basic approach to build a recommender system from scratch.
Cite
Citations (3)
The proliferation use of Internet has made people to rely on virtual recommendations. Recommender systems help out in giving important recommendations. Collaborative filtering is the most successful and widely used approach in designing recommender systems since the introduction of the concept of recommender systems. This approach uses the known tastes and preferences of a set of users to make predictions or generate recommendations about the unknown tastes and preferences of the target user. This paper discusses various works which use collaborative filtering approach to design recommender systems. The paper also gives a comparison of these approaches.
Cite
Citations (2)
Information filtering system
Cite
Citations (12)
Recommender systems help to overcome the problem of information overload on the Internet by providing personalized recommendations to the users. Content-based filtering and collaborative filtering are usually applied to predict these recommendations. Among these two, Collaborative filtering is the most common approach for designing e-commerce recommender systems. Two major challenges for CF based recommender systems are scalability and sparsity. In this paper we present an incremental clustering approach to improve the scalability of collaborative filtering.
Information Overload
Information filtering system
Cite
Citations (17)
Recommender systems are used to help users discover the items they might be interested in, especially when the number of alternatives is big. In modern streaming websites for music, movies, and TV-shows, E-commerce, social networks, and more, recommender systems are widely used. These recommender systems are often looking at the ratings on items for the current and other users, and predicting a rating on the items the user have not seen. Others match the content of an item itself against a user profile. A mix of the two is often used to make the predictions more accurate, and this can also help to the problem when a new user sign up where we have no knowledge about him. This issue, is a well-known problem for recommender systems often described as the cold-start problem, and much research has been done to find the best way to overcome this. In this thesis, we look at previous approaches to recommender systems and the cold-start problem in particular. We have developed our application, Eatelligent, which is recommending dinner recipes based on our study of previous research. Eatelligent has been designed to examine how we can approach the cold-start problem e ciently in a real world application, and what kind of feedback we can collect from the users.
Cold start (automotive)
Cite
Citations (1)
Cold start (automotive)
Cite
Citations (5)
Recommender systems have been widely used as an important response to information overload problem which leads consumers to locate the right information at the right time accurately and rapidly.The most successful and popular technique of such systems is collaborative filtering.However,collaborative filtering faces a key challenge of the cold-start problem.This paper first illustrates the cause of the cold-start problem and then tables the significance of cold-start problem research.The core of this paper is to summarize the existing algorithms which are used for addressing the cold-start problem.In order to facilitate users to choose the right algorithm to tackle the cold-start problem in collaborative filtering,it compares the performance and respective advantage and disadvantage of different algorithms.
Cold start (automotive)
Information Overload
Disadvantage
Cite
Citations (14)
The rating and review based algorithms like collaborative filtering faces the cold start problem on addition of new items or new users in the recommender system because of lack of ratings and reviews. The primary aim of this work is to solve the cold start problem faced by rating based recommendation system and also increase the accuracy of recommender systems. We used the content-based filtering algorithm along with the cross-domain data taken from Facebook so that the problem of addition of a new user problem and a new item in recommender system can be solved. The combination of both content-based filtering and cross-domain data obtains better results than the collaborative filtering and shows a good way to deal with the cold start problem.
Cold start (automotive)
Cite
Citations (4)
Recommender system becomes very popular and has important role in an information system or webpages nowadays. A recommender system tries to make a prediction of which item a user may like based on his activity on the system. There are some familiar techniques to build a recommender system, such as content-based filtering and collaborative filtering. Content-based filtering does not involve opinions from human to make the prediction, while collaborative filtering does, so collaborative filtering can predict more accurately. However, collaborative filtering cannot give prediction to items which have never been rated by any user. In order to cover the drawbacks of each approach with the advantages of other approach, both approaches can be combined with an approach known as hybrid technique. Hybrid technique used in this work is weighted technique in which the prediction score is combination linear of scores gained by techniques that are combined.The purpose of this work is to show how an approach of weighted hybrid technique combining content-based filtering and item-based collaborative filtering can work in a movie recommender system and to show the performance comparison when both approachare combined and when each approach works alone. There are three experiments done in this work, combining both techniques with different parameters. The result shows that the weighted hybrid technique that is done in this work does not really boost the performance up, but it helps to give prediction score for unrated movies that are impossible to be recommended by only using collaborative filtering.
Information filtering system
Cite
Citations (16)