iWEEMS: Interactive Word Embeddings for Early Modern Science

Abstract

This paper introduces a collection of 15-word embedding models supplemented by an interactive visualization web application for their explorations. The models are type-based word embeddings trained on automatically lemmatized texts from several Latin corpora with a special relevance for the history of Early Modern science. The models and web app allow anyone to explore semantic trends in the vocabulary of Early Modern science between the 16th and 17th centuries and across multiple academic disciplines and milieus. In this paper, we document the methodology used to obtain the models and the rationale behind the web app design.