Iris.ai Focus tool is using neural topic modelling for clustering semantically similar documents. When you enter Stage 2 of the Focus process, Iris.ai has identified topics, described using a string of 10 words, that the papers can be grouped into. Each paper can fall into multiple topics and you may use the topics to choose include/exclude criteria of what is and isn’t relevant to you. The topics are modelled both from a general list of millions of papers, from your specific corpus and from the now filtered-down corpus, for increasing specificity.