Making machine-learning tools accessible to language teachers and other non-techies: T-SNE-lab and rocanr as first examples

Machine learning advancements could greatly benefit the general public. Many programs or even trained weight matrices are available as open source software or free download. At its present form, however, most results are not directly accessible to most people. We present a front end for the t-SNE algorithm and a front-end for querying English language word embedding as examples to illustrate what's needed to make these results more accessible to people outside this field. The way their usage can be combined also demonstrates the author's envisioned development of such tools in the future, following the unix tools philosophy.