Kovachev, VenelinPuertas i Prats, EloiMartí Antonin, M. AntòniaIkauniece, Indra2019-06-072019-06-072018-07-01https://hdl.handle.net/2445/134758Treballs finals del Màster de Fonaments de Ciència de Dades, Facultat de matemàtiques, Universitat de Barcelona, Any: 2018, Tutor: Venelin Kovachev, Eloi Puertas i Prats i M. Antònia Martí Antonin[en] After every semester in University of Barcelona all students are asked to fill a survey about professors and subjects from the previous semester. Students provide evaluation by answering two different kinds of questions - quantitative (numeric), and qualitative (open text). It would be useful for the professors, the program coordinators, and for the directors of the departments to have an automatic quantitative overview of the textual answers. The goals of this project are twofold: 1) to create a supervised dataset for sentiment analysis and polarity detection of student opinions in two languages (Catalan and Spanish); and 2) to validate the dataset empirically and propose competitive baselines by investigating, implementing and comparing sentiment analysis algorithms and methods to automatically classify student comments as positive, negative or neutral.38 p.application/pdfengcc-by-nc-nd (c) Indra Ikauniece, 2018codi: GPL (c) Indra Ikauniece, 2018http://creativecommons.org/licenses/by-nc-nd/3.0/es/http://www.gnu.org/licenses/gpl-3.0.ca.htmlLingüística computacionalTractament del llenguatge natural (Informàtica)Treballs de fi de màsterAvaluació dels professorsDades massivesComputational linguisticsNatural language processing (Computer science)Master's thesesTeacher evaluationBig datainfo:eu-repo/semantics/masterThesisinfo:eu-repo/semantics/openAccess