Team Thaurus is a mix of PhD and master's students from Universidad Politécnica de Madrid whose research interests include natural language processing, signal processing, and machine learning.
For this challenge, our priority is not only to capture and keep the user's attention — not just to create the most knowledgeable chatbot, but we want to offer an experience closer to what another human being could offer.
Marcos Estecha-Garitagoitia — Team lead
Estecha-Garitagoitia is a second-year PhD student in telecommunication systems at the Speech Technology and Machine Learning Group at Universidad Politécnica de Madrid. His master thesis was on the topic of spoken language recognition. His research interests include dialogue controllability and knowledge-grounded conversations, focusing on emotions and persona profiles and commonsense reasoning respectively. He also worked on developing automatic evaluation metrics based on Graph Neural Networks (GNNs) for dialog systems. Marcos wasmember of the Genuine2 team that participated in the Alexa Prize SGC4. Marcos is also collaborator at the DSTC10 andDSTC11 challenges for automatic dialogue evaluation.
Cristina Conforto Lopez
Conforto Lopez is pursuing a master’s in telecommunication engineering. She is a member of the IEEE Student Branch ETSIT-UPM and has always been interested in exploring the current advances in technology. After developing her bachelor thesis on the topic of tiny machine learning, she took a special interest in natural language processing. She is also learning about automatic evaluation of dialogue systems at turn and dialogue level.
Estevan has a bachelor’s in telecommunications engineering and is currently pursuing a master's in machine learning and data science. For his final degree thesis, he designed a chatbot using RNN and Transformer models which was focused on identifying toxicity within a conversation or a given comment. Sergio participated in the DETOXIS challenge, oriented to the identification of toxic comments in social networks, organized by Iberlef 2021 as part of GTH UPM team.
Claudia Garoé Fernández García
Fernández García has a bachelor's in telecommunications engineering and is currently pursuing a master's. She did an internship in Accenture where she worked developing and maintaining cloud infrastructures (in Microsoft Azure and AWS). She also participated in several school events such as the 2021 Hackathon or the 2018 TelecoEmprende, both competitions in which programming, teamwork and creativity are key. Although always interested in technology and eager to learn, she recently found a great interest in AI and natural language processing.
Garrachon is a master's student who developed a Alexa skill called Buddy Thomas (Lazarillo Tomás in Spanish). Garrachon used many NLP models to process the data, developing also the whole user-interaction and knowledge retrieval model. He is working on a research project in connection with the university to develop a conversational agent in Spanish and learn state of the art techniques for compressing and optimizing models.
Fiorella Anneth Jhonson García
Jhonson Gracía is pursuing a master's in signal theory and communications, signal processing, and machine learning for big data. She is researching how to automatically generate new persona profiles grounded on commonsense databases. She has also several years of dedication to software engineering at Indra, both in front-end and back-end development.
Mario Rodríguez Cantelar
Rodríguez Cantelar is a third-year PhD student in automation and robotics in Intelligent Control Group at Centre for Automatic and Robotics (CAR UPM-CSIC). . His research interests are NLP and dialogue generation and management, focused on emotions and personality in open domain multilingual conversations. He also worked on the integration of software and components for a first prototype of human-robot conversational interaction. Rodríguez Cantelar was also member of the Genuine2 team that participated in the Alexa Prize SGC4.
Luis Fernando D'Haro — Faculty advisor
D'Haro is an associate professor at UPM. He co-organized DSTC2015-2022 challenges, JSALT2020, WoChat2016-2018 and DBDC4-5, for advancing dialogue systems and their automatic evaluation. He also co-organized Interspeech 2014, HAI2016, IWSDS2018, and was general chair for IWSDS2020. In 2021, he was the faculty advisor for the Spanish team Genuine2 at the Alexa Socialbot Grant Challenge (SGC4). His current research includes automatic evaluation and controlled multimodal and multilingual generation for open-domain dialogue systems, and language and speaker recognition.