Evaluating NLP toxicity tools: Towards the ethical limits
Giner Miguelez, Joan
Ramon i García, Eloi, dir. (Universitat Autònoma de Barcelona. Departament d'Enginyeria Electrònica)
Universitat Autònoma de Barcelona. Escola d'Enginyeria

Data:	2020
Descripció:	70 pag.
Resum:	In the last years we have seen and big evolution in the field of neuronal networks, and the field of natural language processing (NLP). Solutions as voice assistants write assistance, or chatbots are present, every time more often, in our daily work. In addition, these techniques are used for more sophisticated analysis as sentimental classification or hate-speech detection. In contrast, the detection of gender or racial biases in these solutions has created problems. This problem has opened a debate around the limitations and potentials of these solutions. The goal of this work is to evaluate the present tools around the sentimental analysis that are available at the moment of writing. To achieve this, we have selected a set of tools and we have compared its usability over a specific dataset focused on biased detection. In addition, we have developed a tool to evaluate these models in a real-world application by integrating these models into Content Management systems. The developed tool has the goal to help in the moderation of the content in the CMS, is developed over a popular CMS distribution (Drupal). Finally, we present a debate around the ethics and fairness in sentiment analysis using NLP.
Drets:	Aquest document està subjecte a una llicència d'ús Creative Commons. Es permet la reproducció total o parcial, la distribució, i la comunicació pública de l'obra, sempre que no sigui amb finalitats comercials, i sempre que es reconegui l'autoria de l'obra original. No es permet la creació d'obres derivades.
Llengua:	Anglès
Titulació:	Enginyeria de Telecomunicació / Telecommunication Engineering [4313797]
Pla d'estudis:	Màster Universitari en Enginyeria de Telecomunicació / Telecommunication Engineering [1170]
Document:	Treball de fi de postgrau

71 p, 2.1 MB

El registre apareix a les col·leccions:
Documents de recerca > Treballs de recerca i projectes de final de carrera > Enginyeria. TFM