A Review of Recommender System and Related Dimensions
The exponential rise in the number of users and information has resulted in the information overhead, which restricts timely access to intended information on the Internet. In this age of Big Data, the main issues are that the data is heterogeneous, massi
- PDF / 222,082 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 31 Downloads / 249 Views
1 Introduction Recommender system (RS) is subclass of information filtering system that deals with the problem of information overhead and helps in decision making in large information spaces [1]. In other words, we need to develop more personalized form of information access and discovery that has the capability to understand the needs of users and respond to these needs in a more efficient and objective way. RSs are used by business organization (e-commerce) to recommend product to their users. The product can be suggested on the basis of past buying behavior, likes, comments, demographics of the customer, top seller on a site, etc. Nowadays, most of the e-commerce companies provide web recommendation to the user by systematically enabling RS at the back end. RSs have been designed and developed using heuristics approaches [2], data mining techniques [3] and pattern mining (association rule, similarity measure) [4]. Popular RS includes Netflix and MovieLens [4, 5] for movies, Amazon.com [6] for books, CDs, and many other products, Entrée for restaurant, and Jester [7] for jokes.
1.1 Motivation and Problem Explanation Nowadays, due to the hasten evolution of diverse technology, accumulation and generation of digital data can be accomplished effortlessly. Revolution in several domains like information retrieval, approximation theory, machine learning, statisT. Anwar (B) · V. Uma Department of Computer Science, Pondicherry University, Puducherry 605014, India e-mail: [email protected] V. Uma e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2019 R. K. Shukla et al. (eds.), Data, Engineering and Applications, https://doi.org/10.1007/978-981-13-6347-4_1
3
4
T. Anwar and V. Uma
tics, and pattern recognition have given chance to unfold and mine unknown as well as interesting patterns from the data. RS is a field of information filtering, which has gained much attention in recent years. Existence of data in various formats, viz. multimedia elements of web pages, URL logs, user likes, most viewed, or purchased item details, poses a great challenge to RS. Based on these data, filtering, a useful and interesting process removes unwanted and redundant information from the information stream and delivers the information that the user is likely to search. Filtering method is chosen based on how the data is being mined or analyzed. The three filtering methods used in RS are content-based filtering (CBF), collaborative filtering (CF), and hybrid filtering (HF). CBF suggests item based on the individual’s choice made in the past and also based on the most viewed, bought, liked, and positively ranked items. The drawback of this approach is the cold-start problem. Initially, when sufficient information has not yet been gathered, which is otherwise called as cold-start problem, the CBF system cannot be effective as it requires large amount of data/information about users/items for precise and accurate filtering. CF approach works on retrieving and analyzing enormous volume of information on user’s behaviors,
Data Loading...
