An algorithm for recommending online dance teaching resources based on user clustering and preference

Qiu Yang
THE Central Academy of Drama,Beijing,China, 102209
Hao Ma
Department of Education, East China Normal University,Shanghai, China,200062

Abstract:

: A dance teaching algorithm that recommends online resources for ballroom dancing, swing, hip hop and tango based on user preferences. This tool helps dancers identify online resources where they are most likely to find their current dance interests in the future. The model was tested on 1 million users who have visited the website and found that they share mutual interests with other people who have similar tastes.

Dance teaching algorithms have become more popular in recent years as a way of promoting dance music exchange across locations while not being biased towards one location or another. In this paper we use clustering techniques to investigate user preference data collected from a public online discussion board to successfully predict what specific site a dancer is likely to visit during each stage of their learning process. This model allows us to recommend resources that are associated with the users' preferences. In particular we are interested in identifying online resources where dancers who have similar interests will meet each other, which we refer to as 'dance communities'.

The most popular clustering technique is k-means clustering. This is the most widely used technique for user preference prediction because it has a good track record across a wide range of applications, but unlike many other clustering methods k-means cannot represent things that are not linear. In this paper we investigate a related problem: user preference prediction using an unsupervised technique known as density estimation.