Secure Personalized Recommendation System for Mobile User - Information Security and Cryptology-ICISC 2010 - page 268

recommender systems [14]. In order to give recommendation, these systems take
different information like product ratings, history of purchase or the customer's
interests into account. The term collaborative filtering was introduced in the context
of the first commercial recommender system, called Tapestry [4], which was designed
to recommend documents drawn from newsgroups to a collection of users.
Because of the recommender systems are dependent on external sources of
information, they are vulnerable to attacks. Recommender systems have proven to be
an important response to the information overload problem, by providing users with
more proactive and personalized information services. And collaborative filtering
techniques have proven to be a vital component of many such recommender systems
as they facilitate the generation of high-quality recommendations by leveraging the
preferences of communities of similar users.
2.2 Personalization System
Personalization means knowing who the user is and can recognize a specific user
based on a user profile [16]. Personalization involves as a process of gathering and
storing information about users, analyzing the information and based on the analysis,
delivering the information to each user at the right time. User satisfaction is the
ultimate aim of personalization.
Personalization can be divided into content-based filtering (customization), rule-
based filtering and collaborative filtering.
Content-based filtering is an information seeking process in which contents are
selected to satisfy a relatively stable and specific information need. Rule-based
personalization use “If-then” process and based on a customer's demographics, past
purchases, or product attributes. Collaborative filtering (CF) is one of the most
successful recommender techniques. It is the method of making automatic predictions
(filtering) about the interests of a user by collecting taste information from many users
(collaborating). The underlying assumption of CF approach is that those who agreed
in the past tend to agree again in the future.
Challenges in collaborative filtering include scalability, sparsity, cold-start,
accuracy and security. We proposed the RPCF algorithm [8] which is the combination
of rule-based and collaborative filtering approach to give the recommendation results.
Our prior work [9] have addressed the scalability, sparsity and cold-start problem by
using RPCF algorithm and the experimental results showed the improvement of
accuracy and the quality of recommendation compared with the pure collaborative
filtering approach.
The open nature of collaborative recommender systems allows attackers who inject
biased profile data to have a significant impact on the recommendations produced. A
collaborative recommender database consists of many user profiles, each with
assigned ratings to a number of products that represent the user's preferences. A
malicious user may insert multiple profiles under false identities designed to bias the
recommendation of a particular item for some economic advantage. This may be in
the form of an increased number of recommendations for the attacker's product, or
fewer recommendations for a competitor's product.