Miguel Won presents a paper on the 23rd. Here’s the abstract:
“Politicians often guide their public statements by recurrently making use of key-phrases. The wise use of such key-phrases is a common rhetorical technique in political discourse (oral or written), that allows political actors to efficiently transmit the most important ideas associated to their political position. Additionally, according to the context where a political statement is given, different key-phrases can intentionally be used. Analysing these differences opens the possibility of a better understanding of the political intentions for a particular actor (or group of actors), according to the contexts associated to his/her statements.
The present work focuses on the problem of developing and validating statistical methods for extracting and comparing the use of key-phrases in Portuguese political texts originated from different sources. As a first step, we have developed a simple method based on sound statistical principles advanced in seminal studies in the areas of information retrieval and natural language processing, to automatically extract key-phrases. The developed technique was applied in political speeches originated from the same political actors, but given in two different contexts: Parliament debate sessions, and quotations/citations, as taken from on-line news articles. For the Parliament speeches, we have used the faithful transcriptions of the speeches at Portuguese parliament debate sessions from the last completed legislature (the XII legislature of the Portuguese government), collected from the Portuguese parliament website, together with information on each member of parliament (MP) and member of government. In respect to the media source, we relied on a database from an online service that collects quotations from a vast pool of Portuguese news media sources (VOXX service from Sapo Labs).
We have additionally started the collection of opinion articles from the main print media ( newspapers and newsmagazines) with online content. This constitutes an ongoing project that aims to provide political scientists with an indexed and searchable corpus of the political opinion in the Portuguese media. Like above, we have implemented automatic processing techniques to extract relevant Key-Phrases associated to each article.”