In-house seminar
The seminar aims to address the issue of diagnostic captioning, in particular applied to the automatic drafting of mammogram reports. In the first part of the seminar, therefore , an overview of the state of the art relating to diagnostic captioning will be provided, with a particular focus on methodologies, datasets and evaluation. In the second part of the seminar, after a brief discussion on the choice of mammograms as a field of application, a first approach to the generation of reports for the Italian language and the problems encountered will be described. Specifically, an approach used for the generation of a corpus composed of mammograms and related reports for Italian and a first experiment of generation of textual reports starting from the images of the exams will be described. The last part of the seminar will finally be focused on some open questions and possible future research paths.
Speaker(s): Giulia Benotto
Graduated in Computer Science at the University of Pisa, she received her PhD in Computational Linguistics at the same University, with a thesis on the study of the limits and potential of the distributional approach with respect to lexical semantics and, in particular, to the analysis of paradigmatic relationships. She has been a research fellow at the Institute for Computational Linguistics “A. Zampolli” of the National Research Council of Italy from 2011 to 2016 and has dealt, in particular, with applying distributional theories to various issues relating to Digital Humanities (assisted translation, authorship attribution). From 2017 to 2022 she worked at various private companies, dealing with the development and management of chatbots, using both proprietary and IBM Watson technologies. Since February 2022 she has been a researcher at the Institute for Computational Linguistics “A. Zampolli” of the National Research Council of Italy.