TU Darmstadt / ULB / TUbiblio

Digital Transformation of Science: AI-Assisted Collaborative Reading and Evaluation

Buchmann, Jan ; Dycke, Nils ; Zyska, Dennis ; Gurevych, Iryna
Hrsg.: Ubiquitous Knowledge Processing Lab, TU Darmstadt (2023)
Digital Transformation of Science: AI-Assisted Collaborative Reading and Evaluation.
doi: 10.26083/tuprints-00023118
Report, Erstveröffentlichung, Verlagsversion

Kurzbeschreibung (Abstract)

Peer review is a common instrument of quality control in the academic world. New scientific knowledge is only accepted and, in many cases, only published when it has passed this barrier. However, in its current state, peer review has shortcomings, as it is time-intense and often unreliable.

To help make peer reviewing faster and more reliable, and to give guidance in this process to young researchers, we are developing CARE (Collaborative Augmented Reading Environment, formerly PEER), an Artificial Intelligence (AI) assisted software to support researchers in the annotation phase of reading and evaluation of scientific publications. To provide the maximum benefit to researchers from any field, we design CARE to adapt to (i) the user, (ii) the domain of research, and (iii) the document at hand.

This report first introduces the CARE software tool in greater detail. It then presents the setup and results of two studies performed at the Center for Advanced Internet Studies (CAIS), a survey and a user study. These had the aim to elucidate the requirements of the CAIS community members in CARE, and to test the usability of the software. The report summarizes the main findings, showing that the tool is useful to the study participants, and provides an outlook on the future of CARE.

Typ des Eintrags: Report
Erschienen: 2023
Autor(en): Buchmann, Jan ; Dycke, Nils ; Zyska, Dennis ; Gurevych, Iryna
Art des Eintrags: Erstveröffentlichung
Titel: Digital Transformation of Science: AI-Assisted Collaborative Reading and Evaluation
Sprache: Englisch
Publikationsjahr: 2023
Ort: Darmstadt
Kollation: 4 Seiten
DOI: 10.26083/tuprints-00023118
URL / URN: https://tuprints.ulb.tu-darmstadt.de/23118
Kurzbeschreibung (Abstract):

Peer review is a common instrument of quality control in the academic world. New scientific knowledge is only accepted and, in many cases, only published when it has passed this barrier. However, in its current state, peer review has shortcomings, as it is time-intense and often unreliable.

To help make peer reviewing faster and more reliable, and to give guidance in this process to young researchers, we are developing CARE (Collaborative Augmented Reading Environment, formerly PEER), an Artificial Intelligence (AI) assisted software to support researchers in the annotation phase of reading and evaluation of scientific publications. To provide the maximum benefit to researchers from any field, we design CARE to adapt to (i) the user, (ii) the domain of research, and (iii) the document at hand.

This report first introduces the CARE software tool in greater detail. It then presents the setup and results of two studies performed at the Center for Advanced Internet Studies (CAIS), a survey and a user study. These had the aim to elucidate the requirements of the CAIS community members in CARE, and to test the usability of the software. The report summarizes the main findings, showing that the tool is useful to the study participants, and provides an outlook on the future of CARE.

Freie Schlagworte: peer review, assistance, NLP, AI, UKP_p_PEER, UKP_p_seditrah_QABioLit
Status: Verlagsversion
URN: urn:nbn:de:tuda-tuprints-231189
Sachgruppe der Dewey Dezimalklassifikatin (DDC): 000 Allgemeines, Informatik, Informationswissenschaft > 004 Informatik
Fachbereich(e)/-gebiet(e): 20 Fachbereich Informatik
20 Fachbereich Informatik > Ubiquitäre Wissensverarbeitung
TU-Projekte: DFG|GU798/28-1|PEER: Eine computerg
Hinterlegungsdatum: 24 Jan 2023 13:03
Letzte Änderung: 06 Feb 2023 13:11
PPN:
Export:
Suche nach Titel in: TUfind oder in Google
Frage zum Eintrag Frage zum Eintrag

Optionen (nur für Redakteure)
Redaktionelle Details anzeigen Redaktionelle Details anzeigen