News Classifier RepRisk

Customer

The firm RepRisk offers a web-based tool to evaluate environmental and social issues that could pose a reputation or financial risk to a corporation or investment portfolio. RepRisk collates the reaction of media, NGO's and other sources for controversies for finance institutes, corporations and project relevant situations. The stringent selection of news as well as data enriching and cross referencing ensures the relevance, quality and depth of the information provided.

Project

The daily flood of information from different media is overwhelming. An algorithm needed to be developed, to identify and evaluate articles more efficiently, without any loss of quality.

Solution

Astina developed a classifier algorithm, that allows usable information from the dataflood of the internet to be filtered and evaluated based on qualitative criteria. The semantic evaluation defines the relevance of individual attributes within the text. The algorithm learns through manual refinement and thus becomes more precise with each use. In the end, only the relevant articles are viewable in the database. These articles can then be categorised and searched for, based on their similarities.