One of the main aims of the Immersive Automation project is to study automation of editorial processes. Although a lot of things can be automated, it is not worthwhile to automate everything. In order to learn more about the difficulties of automation, and which parts of editorial processes are in fact worth automating, the IA research team developed its first prototype, the multilingual election bot Valtteri.
As a result of news criteria analysis, text template writing, and careful programming, Valtteri was launched a couple of days before the 2017 municipal elections in Finland. By utilising election result data from the Finnish Ministry of Justice, Valtteri combines relevant facts with text templates, and generates short news stories. The first version of Valtteri can write news in Finnish, Swedish, and English.
Because computers need plenty of examples to learn from, it was necessary for us to create a prototype which can produce massive amounts of texts. Based on our experiences and what we have learnt, we will continue to develop more sophisticated text generators.
Find Valtteri the Election Bot at vaalibotti.fi
Blog post April 7 2017: Meet our first prototype – Valtteri the election bot