Using sonification for mining time series data

In recent years, there is a growing interest in mining time series databases by both automated and interactive tools. In this paper, we present an interactive methodology for mining of time series data using a novel sonification technique which uses some important properties of time series and tonal music to achieve effective (accurate) and efficient (fast) results. We have created an experimental website, where participants were asked to perform some basic data exploration and mining tasks by listening to a musical display of several time series. The initial results indicate that the proposed methodology for musical representation of data allows, on one hand, to efficiently perform some decisionmaking tasks "on the fly" - by only listening to some short music examples, and on the other hand, it provides an alternative data representation for blind or visually impaired users or users who are due to their professional or personal activities (e.g., driving) cannot use their sense of vision for watching a visual display of data, but still need to get some important time-based information by using their other senses.