1. INTRODUCTION
With the overwhelming information on the Internet and limitations of one-fit-all search
engines, advanced tools are required to enable users to find the right information and
make choices meeting their needs and expectations, thus enhancing their engagement
and overall satisfaction with online services. Recently, recommender systems have
been increasingly popular in assisting users with their choices. A recommender system
can be abstracted to consist of a user model, a community, an item (product) model, a recommender algorithm, and an interaction style [Zanker and Jessenitschnig 2009].
The user model provides all of the information for personalizing the user’s experience.
It captures the user interactions with items in user profiles. Mainly, these user interactions consist of explicit and implicit information about the user’s interest or preference
for items. Typically, recommender systems use ratings as a mechanism to proactively
express their interests in items and seamlessly collected clickstream data for inferring
users’ interests or preferences. This explicit and implicit information are usually referred to as explicit feedback or explicit rating and implicit feedback, or implicit rating
[Konstan et al. 1997; Jannach et al. 2011]. There has been significant research activity in this area since the 1990s. However, relatively little attention has been given to
questioning how user feedback is applied to recommender systems. Several recommendation algorithms do not account for the variability in human behavior and activity.
Often, they are hardwired for explicit ratings rather than implicit ratings.
1. INTRODUCTION
With the overwhelming information on the Internet and limitations of one-fit-all search
engines, advanced tools are required to enable users to find the right information and
make choices meeting their needs and expectations, thus enhancing their engagement
and overall satisfaction with online services. Recently, recommender systems have
been increasingly popular in assisting users with their choices. A recommender system
can be abstracted to consist of a user model, a community, an item (product) model, a recommender algorithm, and an interaction style [Zanker and Jessenitschnig 2009].
The user model provides all of the information for personalizing the user’s experience.
It captures the user interactions with items in user profiles. Mainly, these user interactions consist of explicit and implicit information about the user’s interest or preference
for items. Typically, recommender systems use ratings as a mechanism to proactively
express their interests in items and seamlessly collected clickstream data for inferring
users’ interests or preferences. This explicit and implicit information are usually referred to as explicit feedback or explicit rating and implicit feedback, or implicit rating
[Konstan et al. 1997; Jannach et al. 2011]. There has been significant research activity in this area since the 1990s. However, relatively little attention has been given to
questioning how user feedback is applied to recommender systems. Several recommendation algorithms do not account for the variability in human behavior and activity.
Often, they are hardwired for explicit ratings rather than implicit ratings.
การแปล กรุณารอสักครู่..
