The BIR and BIRNDL workshops are a unique place where people from several di erent domains, Bibliometrics, Information Retrieval and more broadly Natural Language Processing can exchange ideas and work together to develop new approaches for the processing of scienti c corpora. During our PhD theses at Paris-Sorbonne university, we both worked on the linguistic analysis of corpora, in the framework of enunciative linguistics. Iana Atanassova focused on the semantic dimension of discourse in scienti c corpora, and in particular for applications in Information Retrieval, and Marc Bertin focused on the discursive forms present in citation contexts. This work on citation acts, with a purely linguistic approach at the beginning, is the starting point of more than 10 years of work. The link between these two approaches and our collaboration started around the same period as the rst BIR workshop in 2014. At that moment Iana worked on the development of a search engine at MyScienceWork and Marc was a post-doctoral fellow at Montreal working with Yves Gingras and Vincent Lariviere [8]. The BIR and BIRNDL workshops have become regular events where we can discuss our research progress and share ideas around bibliometrics and the processing of scienti c papers. Iana is proud to have been the keynote speaker of BIR 2019 [1]. Considering bibliographic references in papers, we can observe a great variety in the citation contexts but also many regularities that need to be studied an categorized in order to better understand the linguistic nature of citations and build solutions for ontology population. We have approached the study of citation contexts rstly by examining the regularities in their lexical distributions in the IMRaD structure [2] and by further characterizing the di erent sections of papers by Factorial Correspondance Analysis [3] and later by K-means and Hierarchical clustering methods [6]. We also studied phenomena around multiple references in the same sentence [4] and negative citations [5]. The InTeReC corpus was proposed [7] to address some of the problems in developing machine learning approaches for citation classi cation and we conducted a preliminary study to compare such approaches with rule-based ones [10]. Finally, we ques? A companion video is hosted at https://youtu.be/hLgTB0av_b0.
tioned the text spans that should be considered as citation contexts [
The BIR community is growing and is becoming the place to be to share and discuss new and original ideas around mining scienti c papers. We are very happy to be part of this family.