Seminar 13/03/2018

Analysis of textual data in a temporal perspective
ILC-CNR - Aula Seminari IBF SG 5
Those who study the history of ideas rely on a long tradition aiming at reconstructing (ex post) the fundamental stages of the development of a discipline and recognize the most influential and authoritative authors of the past. In this research we have proposed an alternative reading of the history of ideas through a statistical analysis of the contents conveyed by the articles published in scientific journals and, therefore, a representation of the topics that, at the specific time of publication, were relevant for the dominant scientific communities.
An interdisciplinary group of Italian scholars worked on a research project funded by the University of Padova (PRAT CPDA145940 Tracing the History of Words. A Portrait of a Discipline Through Analyses of Keyword Counts in Large Corpora of Scientific Literature) that analyzed authoritative journals of some disciplines of the sector of the Humanities and Social Sciences (Philosophy, Sociology, Psychology, Linguistics and Statistics) in order to explore the temporal evolution of theories, concepts, methods and fields of application through the temporal evolution of the key words present in the articles published.
Starting from the observation of the occurrences of the keywords over time, first of all we wanted to verify the existence of a temporal pattern in the data (correspondence analysis); then both methods for the study of the temporal trajectories of the single key words (curve clustering) and methods for the automatic identification of the main themes (topic detection) were used to study the evolution of ideas over time.
The fact of observing the trajectories designed by the occurrences of the keywords to study their destiny in a dynamic way has also opened up an interesting theoretical perspective: the research dealt with innovative concepts of "quality of life" and "life cycle" of words that go beyond the more traditional objective of Linguistics, namely the identification of their first attestation.
Arjuna Tuzzi

University of Padova