How good your recommender system is? A survey on evaluations in recommendation
- PDF / 2,173,846 Bytes
- 19 Pages / 595.276 x 790.866 pts Page_size
- 6 Downloads / 192 Views
ORIGINAL ARTICLE
How good your recommender system is? A survey on evaluations in recommendation Thiago Silveira1 · Min Zhang1 · Xiao Lin1 · Yiqun Liu1 · Shaoping Ma1 Received: 16 May 2017 / Accepted: 5 December 2017 © The Author(s) 2017. This article is an open access publication
Abstract Recommender Systems have become a very useful tool for a large variety of domains. Researchers have been attempting to improve their algorithms in order to issue better predictions to the users. However, one of the current challenges in the area refers to how to properly evaluate the predictions generated by a recommender system. In the extent of offline evaluations, some traditional concepts of evaluation have been explored, such as accuracy, Root Mean Square Error and P@N for top-k recommendations. In recent years, more research have proposed some new concepts such as novelty, diversity and serendipity. These concepts have been addressed with the goal to satisfy the users’ requirements. Numerous definitions and metrics have been proposed in previous work. On the absence of a specific summarization on evaluations of recommendation combining traditional metrics and recent progresses, this paper surveys and organizes the main research that present definitions about concepts and propose metrics or strategies to evaluate recommendations. In addition, this survey also settles the relationship between the concepts, categorizes them according to their objectives and suggests potential future topics on user satisfaction. Keywords Recommender system · Evaluation · Novelty · Diversity · Serendipity · Unexpectedness
1 Introduction Recommender Systems (RSs) have been largely studied for the past decade and have shown to be suitable for many scenarios. On the arrival of the internet and the era of e-commerces, companies are opting for having a RS as an attempt to boost sales. RSs provide predictions of items that the user may find interesting to purchase [2], in which most algorithms for this purpose focus on providing recommendations that fit the preferences of the user. RSs have shown to be useful for users and business. Users suffer from what is called the paradox of choice. Having many options to choose from lead to more difficulty in effectively making a choice [31]. Since e-commerces have a vast amount of items, users face a complication in finding what they desire. Therefore, RSs can help users, since good predictions can reduce the search space for the user, facilitating their decision making process [29]. Moreover, it is also * Min Zhang [email protected]; z‑[email protected] 1
Tsinghua National Laboratory for Information Science and Technology, Department of Computer Science and Technology, Tsinghua University, Beijing, China
advantageous for business, because it represents enlargement of sales. Specifically, RSs are able to increase sales of niche items. Popular items are visible to the users anyway, however niche items would not be very likely visible by all users, but personalized recommendations can fin
Data Loading...
