• Using Global Vectors for Getting Semantic Fields

    Vector representations for words are becoming popular in text mining. In particular, the text2vec algorithm (by Tomáš Mikolov) and the GloVe algorithm (by Jeffrey Pennington, Richard Socher and Christopher D. Manning) have attracted considerable attention - including mine. These algorithms generate a vector representation of words based on term cooccurrence...


  • Pride and Prejudice: Finding Westerwelle's speech

    Pride and Prejudice: Finding Westerwelle’s speech In a seminar on policy responsiveness I am teaching with my colleague Susanne Pickel, we explore the relationship between the political supply of policies, and the substance of policies citizens demand for. Before proceeding to a quantitative analysis, we wanted to do a qualitative...


  • The peak of refugees: Dispersion of terms across time

    The peak of refugees: Dispersion of terms across time Publishing blog entries here has been a plan for a while. The intention is to share findings we may gain from large-scale corpora. Most of the time, I will elaborate on a some scenario that the polmineR package may be useful...


  • PolMine Project - Registration as CLARIN Centre

    The PolMine Project is now a registered CLARIN centre, i.e. it is listed here. As a category C CLARIN centre, we offer metadata for the data resulting from the PolMine project that can be harvested by a web service. Most importantly, our data can now be explored through the Virtual...


  • polmineR update - version 0.6.2

    A new release of the polmineR-package is available at CRAN. Apart from behind-the-scenes-refactoring that makes the code more consistent, there availability of progress bars for time-consuming functions has been extended. Improving the usability has been a prime aim of this update.