Search

Préface; International audience; The special issue on Bibliometric-enhanced Information Retrieval presents works at the crossroad between Bibliometrics and Information Retrieval, two domains closely related on the map of Information Science.This special issue eventually includes eight papers: three are extended papers presented at BIR and five are extended papers presented at BIRNDL. It covers a variety of topics including: Authorship identification and writing style analysis (Rexha et al. 2018); Locality sensitive hashing to measure the overlap bibliographic databases for paper indexing (Abdulhayoglu and Thijs 2018); Automatic discovery of cross-topic collaborations (Cagliero et al. 2018); Identification of problems and their solutions in scientific articles (Heffernan and Teufel 2018); Accurate identification of cited text spans (e.g., to generate summaries) (Ma et al. 2018); Detection of automatically generated sentences and texts within genuinely written papers (Tien and Labbé 2018); Citance-based retrieval and summarization using Information Retrieval and Machine Learning (Karimi et al. 2018); Analysis of search stratagems for interactive information retrieval in bibliographic databases (Kacem and Mayr 2018).

International audience; The Bibliometric-enhanced Information Retrieval (BIR) work-shop series has started at ECIR in 2014 and serves as the annual gath-ering of IR researchers who address various information-related tasks onscientific corpora and bibliometrics. We welcome contributions elaborat-ing on dedicated IR systems, as well as studies revealing original charac-teristics on how scientific knowledge is created, communicated, and used.This editorial presents all accepted papers at the 7th BIR workshop atECIR 2018 in Grenoble, France.

De captivantes questions de recherche résident à la croisée entre la scientométrie et d'autres disciplines. Cet exposé synthétisera mes tentatives d'exploration de ces frontières avec l'informatique, la psychologie et la sociologie des sciences. Lors de cette excursion, nos escales aborderont des sujets originaux et variés : quantification d'éponymes, équilibre travail-loisir des chercheurs, dynamiques de carrières en informatique, biais affectant les conférences basées sur l'évaluation par les pairs, crowdsourcing alimentant les bibliothèques clandestines de l'édition scientifique... Je souhaite communiquer mon enthousiasme pour la recherche interdisciplinaire médiée par la scientométrie, tout en soulignant les opportunités offertes aux chercheurs en science des données.

International audience; Information Retrieval Systems (IRS) generally display results as a list of documents.One may think that a deeper structure exists within results. This hypothesis is reinforced bythe fact that most of the graphs produced from real data (e.g., graphs of documents) sharesome structural properties, and in particular a community structure. We propose to use theseproperties to better organize the set of returned documents for a query from a given IRS. For thispurpose, the retrieved document set is modeled as a bipartite graph (Documents ↔ Terms) onwhich the Kodex community detection algorithm is applied. This paper presents Kodex and itsevaluation : regarding F1 measure, Kodex overcomes baseline Okapi BM25 by 22%.; Les Systèmes de Recherche d’Information structurent en général leurs résultats sousla forme d’une liste de documents. Nous pensons qu’il existe une structure plus riche dansces résultats. En effet, la plupart des graphes obtenus à partir de données réelles (entre autre,les graphes de documents) partagent certaines propriétés structurelles, en particulier uneorganisation en communautés que nous proposons d’exploiter afin de mieux organiser l’ensembledes documents restitués pour une requête. Pour ce faire, l’ensemble des documents restitués estmodélisé par un graphe biparti (Documents ↔ Termes) sur lequel est appliqué notre algorithmeKodex de détection de communautés. Cet article présente Kodex et son évaluation : sur la mesureF1 , Kodex améliore significativement la baseline Okapi BM25 de 22 %.

Information Retrieval Systems (IRS) generally display results as a list of documents.One may think that a deeper structure exists within results. This hypothesis is reinforced bythe fact that most of the graphs produced from real data (e.g., graphs of documents) sharesome structural properties, and in particular a community structure. We propose to use theseproperties to better organize the set of returned documents for a query from a given IRS. For thispurpose, the retrieved document set is modeled as a bipartite graph (Documents ↔ Terms) onwhich the Kodex community detection algorithm is applied. This paper presents Kodex and itsevaluation : regarding F1 measure, Kodex overcomes baseline Okapi BM25 by 22%., Les Systèmes de Recherche d’Information structurent en général leurs résultats sousla forme d’une liste de documents. Nous pensons qu’il existe une structure plus riche dansces résultats. En effet, la plupart des graphes obtenus à partir de données réelles (entre autre,les graphes de documents) partagent certaines propriétés structurelles, en particulier uneorganisation en communautés que nous proposons d’exploiter afin de mieux organiser l’ensembledes documents restitués pour une requête. Pour ce faire, l’ensemble des documents restitués estmodélisé par un graphe biparti (Documents ↔ Termes) sur lequel est appliqué notre algorithmeKodex de détection de communautés. Cet article présente Kodex et son évaluation : sur la mesureF1 , Kodex améliore significativement la baseline Okapi BM25 de 22 %.

International audience; The COVID-19 pandemic has thrust preprints into the spotlight, highlighting their advantages and disadvantages. The lack of peer review allows publication to occur with unprecedented speed, but this has raised concerns among biomedical scientists about the quality of the reported research. The study had three primary objectives in regard to COVID-19 preprints: 1. Test the feasibility of automatically and publicly evaluating preprints on a large scale; 2. Assess the prevalence of common quality problems in preprints; and 3. Compare the quality of preprints to published papers. While a substantial fraction (36.2%) of preprints addressed study limitations, the proportion that met other quality criteria was much lower, with only 20% addressing sex as a variable, ~14% sharing open code or data, 7.6% including non-colorblind safe images, and 7.3% showing misleading bar graphs. Both authors and non-authors interacted with the automated Tweets containing reports for preprints. This project shows that it is feasible to conduct large-scale automated screening of preprints for common quality criteria and provide feedback to study authors and readers before publication. These reports can publicly raise awareness about factors that affect study quality and reproducibility, while helping authors to present their research in a more transparent and reproducible manner.

"In this paper we describe a case study where researchers in the social sciences (n=19) assess topical relevance for controlled search terms, journal names and author names which have been compiled by recommender services. We call these services Search Term Recommender (STR), Journal Name Recommender (JNR) and Author Name Recommender (ANR) in this paper. The researchers in our study (practitioners, PhD students and postdocs) were asked to assess the top n preprocessed recommendations from each recommender for specific research topics which have been named by them in an interview before the experiment. Our results show clearly that the presented search term, journal name and author name recommendations are highly relevant to the researchers topic and can easily be integrated for search in Digital Libraries. The average precision for top ranked recommendations is 0.749 for author names, 0.743 for search terms and 0.728 for journal names. The relevance distribution differs largely across topics and researcher types. Practitioners seem to favor author name recommendations while postdocs have rated author name recommendations the lowest. In the experiment the small postdoc group favors journal name recommendations." (author's abstract)

The advent of large-scale citation services has greatly impacted the retrieval of scientific information for several domains of research. The Humanities have largely remained outside of this shift despite their increasing reliance on digital means for information seeking. Given that publications in the Humanities probably have a longer than average life-span, mainly due to the importance of monographs in the field, we propose to use domain-specific reference monographs to bootstrap the enrichment of library catalogs with citation data. We exemplify our approach using a corpus of reference monographs on the history of Venice and extracting the network of publications they refer to. Preliminary results show that on average only 7% of extracted references are made to publications already within such corpus, therefore suggesting that reference monographs are effective hubs for the retrieval of further resources within the domain.