Understanding perceptions to a refugee influx through analysis of radio content

PARTNERS: UN Country Team 

PROGRAMME AREA: Humanitarian Action 

LAB: Pulse Lab Kampala


An outbreak of conflict in July 2016 caused thousands of South Sudanese to flee to neighbouring countries, especially Uganda, where the number registered in 2016 reached over 600,000 people, mainly women and children. Due to the protracted nature of the crisis in South Sudan, the UN Country Team continues to consider options that enable both refugees and host communities to build resilience and improve self-reliance. Moreover, the South Sudanese influx presents increasing social, economic and environmental pressures on host communities that, unless addressed through innovative and targeted support, could result in conflict and instability. 

In response to the UN needs, Pulse Lab Kampala was tasked by the UN in Uganda with unearthing the attitudes and intentions of host communities towards refugees. The objective of the project is to use automated speech-to-text methodologies to analyse local radio content to provide insights around the refugee influx in Uganda.

The flow of analysis will be focused on public opinions expressed in the Acholi language, spoken in Northern Uganda and the Luganda language, spoken in the Central region of the country as well as capital city Kampala. The results of the analysis aim to inform:

a)  early warning systems on emerging issues (with special focus on conflict prevention);

b) programme implementation with analysis of public perceptions about the ongoing response to refugees and host communities; and

c) resource mobilization strategies with analysis of emerging needs related to refugees and host communities. 

The project builds on lessons learned by Pulse Lab Kampala and partners with the completion of a series of case studies probing that: a) public radio discussions include reports of incidents and first-hand experiences not formally recorded anywhere else; b) the unconstrained nature of radio discussions, as compared to other types of data collection methods, contains relevant information around perceptions expressed by people; c) real time analysis of radio content includes people otherwise excluded by the digital divide, and d) public radio content can be treated as a source of big data and analysed with machine learning techniques. 

STATUS: Ongoing