People have always relied on the
recommendations from their peers or the advice of experts to support their
decision making. Amazon.com has been using collaborative filtering for a decade
to recommend products to their customers, and Netflix valued improvements to
the recommender technology underlying their movie rental service at $1M via the
widely published Netflix Prize [6]. Research on recommender algorithms garnered
significant attention in 2006 when Netflix launched the Netflix Prize to
improve the state of movie recommendation. The objective of this competition
was to build a recommender algorithm that could beat their internal CineMatch
algorithm in offline tests by 10%. It sparked a flurry of activity, both in
academia and amongst hobbyists. The $1 M prize demonstrates the value that
vendors place on accurate recommendations [8].
Recommender Systems provide the users with the
suggestions of information that may be useful to the users to make their
decisions on various situations such as selecting movies to watch, books to
read, pages to like, friends to add etc. The techniques for recommendation
systems are classified as content-based filtering, collaborative filtering and
hybrid filtering.
The goal of a collaborative filtering is to
suggest some items or predict the utility of a certain item for a particular
user based on the user’s previous likings and the opinions of other like-minded
users [2]. Whereas, the goal of the content based technique is to suggest an
item based on the similarity of features of the item and the user. The hybrid
recommender system is the combination of both collaborative filtering and
content based filtering approach which either makes the content-based and
collaborative-based predictions separately and then combines them or by
unifying the approaches into one model.
Recommendation system is a part of daily life
where people rely on knowledge for making decisions. Recommender system usually
make use of either or both collaborative filtering and content based filtering
as well as knowledge-based system. Collaborative filtering is based on the
assumption that people who agreed in the past will agree in the future, and
that they will like similar kinds of items as they liked in the past.
Content-based filtering methods are based on a description of the item and a
profile of the user’s preferences. Hybrid Recommender Systems either makes the
content-based and collaborative-based predictions separately or then combines
them or by unifying the approaches into one model. Netflix is a good example of
the use of hybrid recommender system.
The researchers have described the variety of approaches,
which are used to provide recommendation like collaborative filtering, content
based filtering, and hybrid filtering. They have clarified the methods of the approaches.
One of the approach is model based approach which is a theoretical model
proposed of user rating behavior. Instead of using raw data, they directly make
predictions. Another approach is content based approach which have different methods
i.e. Wrapper methods, filter methods and embedded methods. Researchers have
also described the hybrid approach which is use to make predictions and these
approaches are categorized by three types i.e. Weighted hybrid, mixed hybrid
and cross-source hybrid [1].
Refrences:
[1]
|
B. Bhatt, P. P. J. Patel and P. H. Gaudani, A Review Paper on Machine
Learning Based Recommendation System, Dharmaj, Vallabh Vidhyanagar:
Department of Computer Engineering, IIET, GCET, 2014.
|
[2]
|
B. Sarwar, G. Karypis, J. Konstan and J. Riedl, "Item-Based
Collaborative Filtering Recommendation Algorithms," GroupLens Research
Group/Army HPC Research Group, Minneapolis, 2001.
|
[3]
|
C. A. Gomez-Uribe and N. Hunt, "The Netflix Recommender
System," ACM Transactions on Management Information Systems, New York,
2015.
|
[4]
|
D. Jannach, Recommender Systems: An Introduction, Dortmund, 2014.
|
[5]
|
F. Ricci, L. Rokach, B. Shapira and P. B. Kantor, Recommendation Systems
Handbook, Springer, 2011, p. 842.
|
[6]
|
J. Bennett and S. Lanning, "The netflix prize," KDD Cup and
Workshop, California, 2007.
|
[7]
|
J. Hosein, A. Sim and Saadatdoost, "A Naive Recommendation Model
for Large Databases," International Journal of Information and
Education Technology, 2012.
|
[8]
|
M. D. Ekstrand, J. T. Riedl and J. A. Konstan, Collaborative Filtering
Recommender Systems, vol. 4, Foundations and Trends in Human-Computer
Interaction, 2011.
|
[9]
|
S. G. Walunj and K. Sadafale, "An online recommendation system for
e-commerce based on apache mahout framework," Proceedings of the 2013
annual conference on Computers and people research, ACM (2013), Cinncinnati,
2013.
|
[10]
|
S. Gong, "A Collaborative Filtering Recommendation Algorithm Based
On User Clustering and Item Clustering," Journnal of Software, vol.
5, 2010.
|
[11]
|
Y. Lee, Recommendation System Using Collaborative Filtering, San José:
Master Thesis and Graduation Research; The Faculty of the Department of
Computer Science, 2015.
|
[12]
|
Z. Huang, X. Li and H. Chen, "Link Prediction Approach to
Collaborative Filtering," IEEE, Tucson, 2005.
|
Comments
Post a Comment