Matthew Adendorff

Accepted Talks:

Large-scale media analysis for driving accountability

In an age where fake news is an emerging concern, digestion of digital media should be assisted by informative measures that are exemplary of integrity, responsibility, equal representation, and a lack of personal agenda. Such metrics are intrinsically complex and their derivation, calculation and implementation should be handled with care. At the core of developing such accountability measures is the ability to ingest, process, analyze and draw insights from digital media. To achieve this, Media Monitoring Africa has developed a cloud-based platform, built on Flask, with a full Python stack, to automatically ingest digital media from several countries (particularly South Africa and other African countries). The system then performs Natural Language Processing to highlight important sources, topics, and entities, and allows for advanced analytics on the resulting data to be performed.

This talk will present the rationale, data flow and cloud implementation of this platform, as well as its application to driving accountability and insights in South African media. The intended audience is broad, where members can see a detailed road-map of constructing a Python-based Big Data type system, the analyses that can be performed with it, as well as the impact it can have on a society's approach to digital media.