Non-supervised academic documents grouping by topics: Methods and performance.
Researchers not only need to search for academic documents, but they also need an overview of the topics of the documents in their field so they can know what to search for. This is achieved using science maps, a visualization that provides an overview of the topics in a set of documents. We research how we can group together documents with the same topic. We also research if these groups are useful for information retrieval, and which kind of topics are better grouped. For the grouping we use non-supervised methods and document similarity networks (e.g. similarity of text, citations, authors, twitter, etc.. ).
