Validation of video data mining techniques to extract information from an Indigenous Knowledge technology repository.

Over the past years, advances in multimedia data mining have gained much attention. Multimedia data mining can be defined as the process of finding interesting patterns from media data such as audio, video, image, text and electronic and digital ink that are not ordinarily accessible by basic queries and associated results (Shrishrimal, Deshmukh & Waghmare, 2014). Multimedia mining is a subfield of data mining, which is used to find meaningful information of implicit knowledge from multimedia databases (Vijayarani & Sakila, 2015). Coupled with emerging technologies data mining techniques applied to big data repositories have shown to be useful in many fields, among others in cultural heritage preservation.

The Indigenous Knowledge Technologies (IKT) research cluster of the Faculty of Computing and Informatics (FCI) at the Namibia University of Science and Technology (NUST) has collected huge amounts of multimedia mixed content data sets over one decade of fieldwork. Various researchers using their own naming convention, based on specific research activities and trips organised the current multimedia repository. Data was collected with various equipment yielding in files with different formats at times with overlapping or related content. The data were stored temporarily across different storage devices and is now unified on the research cluster’s server. The repository contains valuable information for many other research initiatives across disciplines which have not yet been mined. However, the video mixed content datasets need to be pre-processed in order to be able to extract semantic information with video mining techniques to be identified.