Metadata of French Electronic Theses (theses.fr), focused on altmetrics (2020)

For the vast majority of the manuscripts it hosts, the theses.fr archive provides the following metadata: author(s), dissertation title, document type, defense date, domain(s), keywords, language, and the identity of the dissertation director or supervisor. Additionally, two dynamic metrics are displayed on each manuscript’s presentation page: the number of views and the number of downloads. They were collected in 2020. A document is considered “viewed” when a user accesses the webpage displaying both its metadata and the URL link to the PDF.

However, it is important to note that the PDF can be accessed and downloaded directly from a search engine like Google, without requiring navigation through the archive or the use of Google Scholar. This process does not generate views, which likely explains why, as we will observe, the number of downloads consistently exceeds the number of views. One limitation of such traffic metrics is that they do not indicate whether the documents are actually read.

Metadata were collected on June 23, 2020, for 99,743 manuscripts available in the open repository, using a web scraper written in Python based on the Selenium library.