Preparation Notes

.

.

Workshop 4:

Intelligent Technologies for Virtual Communities

Findings

Introduction

Due to the current age of explosion information and entertainment technologies, everyday people have dozens of choices to make, especially in the web. The web offers myriads of possibilities, such as thousands of interactive documents to read, numerous chat rooms to join, and billions of items to purchase. These vast numbers of possibilities vary widely in quality, and evaluating all these alternatives takes time. As a result, recommendations for people who are familiar with these choices are essential.

Recommendation is a communicative act. It is usually based on the preferences of the recommender. A recommendation may be directed to specific individuals or "broadcast" to anyone who is interested. A recommendation is a source for the receiver in helping them to make decisions from the universal of alternatives.

In the Internet world, people are turning to the computational recommender systems, which automates or supports part of the recommendation process. It assume part of the recommender role, and offer recommendations to users based on their preferences. Thus, it allows people to share and create recommendations easier.

Issues for Computational Recommender Systems

There are four main issues identified to characterize the design space for recommender system. The issues would be examine one bye one below:-

Preferences

Recommendation is usually based on preferences. Therefore, the automated recommender system must obtain preferences from people concerning the relevant domain. Generally, the guidelines for the domain relevancy is according to the previous users to the system in seeking the recommendations.

Roles & Communications

The role of the recommender could be filled by either or both human or computational system depending on the needs. And, the communication method of the recommender could be directed to a specific person or "broadcast" to the public. Sometimes, the systems might allows users to have the opportunity to provide feedback to the recommender.

Algorithms for Computing Recommendations

Automated recommender system might have problems in determining whose preference to use in computing a recommendation. This is because all the people who have expressed their preferences are placed in a large, multi-dimensional space, and it is not easy to weight and combine these alternatives together.

Human-Computer Interaction

The most simple and common example for presenting recommendations are in an ordered list. The more complicated examples include 2D and 3D visualizations, as well as visual annotations of existing information spaces.

Types of Recommender Systems

There are many types of recommender systems that has developed during 1990s. The few major types of recommender systems are discussed below:-

Content-Based System

It is used only the preferences of the seeker. They recommend items that are similar to the items that the user like in the past. Their focus is on the algorithms for learning user preferences and filtering a stream of new items for those which most closely match user preferences.

Recommendation Support System

This system does not automate the recommendation process. They do not have to represent preferences or compute recommendations. They served as a tools to support people in sharing recommendations, helping both those who produce recommendations and those who looks for recommendations.

Social Data Mining System

This system cares about implicit preferences from computational records of social activity, such as UseNet messages, system usage history, citations or hyperlinks. These systems also focus on human-computer interaction issues involved in visualizing the results. These visualization often have been presented to aid the navigation of information spaces in the World Wide Web (WWW).

Collaborative Filtering System

This system requires recommendation seekers to express preferences by rating a dozen or two items. Hence, this merges the role of recommendation seekers and preference provider. This system focus on algorithms for matching people based on their preferences and weighting the interests of people with similar taste to produce a recommendation for the information seeker.

Issues

 

The issues that raised from the above research on computational recommendation systems are stated below:-

• How can one determine which is the best computational recommendation system used?

• What are the advantages of computational recommendation system?

• What are the disadvantages of the computational recommendation system?

• How is the future of the computational recommendation system seems?

 

References

 

	Quality	Title & Author
	*****	Beyond Recommender Systems: Helping People Help Each Other - Loren Terveen & Will Hill
	*****	E-Commerce Recommendation Applications - J. Ben Schafer, Joseph A. Konstan & John Riedl
	****	Interaction Design for Recommender System - Kirsten Swearingen & Rashmi Sinha
	****	The Role of Transparency in Recommender Systems - Rashmi Sinha & Kirsten Swearingen
	***	Recommender Systems - Paul Resnick & Hal R. Varian
	**	HUBRIS: HUman Benchmarking of Recommender Systems
	*	Personalization and Recommender Systems in the Larger Context - Clifford Lynch

Preparation Workshop        |    One    |    Two    |    Three    |    Four    |    Five    |    Six    |                    Top