New donors are identified through their activity on social media networks such as current and past tweets, posts and statements in social media that can reflect their emotional involvement in an event. These activities are prefiltered with state of the art machine learning algorithms in three dimensions such as geospatial, temporal and sematic. The filtered social media post are visualised in a dashboard with additional exploring capabilities and the ability to engage with potential donors through the dashboard. The focus of this project is on emergency donations for natural events or social conflicts, the most important donation motivation with 41% of all donations worldwide represent.

Most fundraising activities focus on traditional communication channels such as payment slips in letters. However, potential donors first have to give consent to receive such letters and usual methods to get in contact with potential donors are for example by reaching out to them in the public. This procedure comes with high costs as areas must be located where people might be interested in the campaign and high efforts have to be taken in order to attract somebody to the campaign in the public. In virtual space like in social media networks people show their interest in specific events that are more likely in willing to donate and are a different target group than people contacted with traditional communication channels.

A prototype platform is developed that combines the crucial functionalities for the success of online fundraising activities. Users can visualize semantically filtered social media posts that related to the targeted event with multiple features such as the sentiment of a tweet or the location on a map. Additional statistics, time charts and pie charts enable the user to analyze the impact of the natural disaster in the social media stream. When potential donors are identified users can send personalized messages to the user in order to attract them to their campaign.