From 2011 until 2023, our lab was the lead developer and host of a project called Sciveyor (previously known as evoText and powered by software that was, for a time, called RLetters). This was a web application enabling end-users to perform text analyses against our corpus, which, in its prime, held around two million articles from journals across the sciences.
We made the decision to sunset the project in mid-2023. With the rise of interactive notebook-based programming in Python and R, it has become less and less important to provide these kinds of user-friendly front-ends. The lack of ability for users to customize the ways in which the algorithms were run against the text led to a kind of inflexibility. Further, as the project was started in 2011, it had accumulated an unmanageable level of technical debt. Finding time and financing to engage in a complete, bottom-up rewrite of the system seemed like an increasingly impossible proposition.
We’ve moved internally to a combination of Python-based notebook programming and smaller, lightweight collections of scripts that we develop for the needs of particular projects. Some of these are publicly available, largely on Prof. Pence’s Codeberg account. We also still have access to our entire corpus, which we maintain and use to pursue a number of ongoing projects. We still encourage interested users to get in touch with us if they would like to work on analyses of the scientific literature!