This social bot will be trustworthy, engaging, and safe by generating responses that are consistent to internal state and external knowledge.
We aim to build a socialbot that is trustworthy, engaging, and safe. Trustworthiness means that generated responses of the socialbot are consistent with internal state and external knowledge so that the users can always trust the bot.
Hong Wang - Team leader
Wang is a fifth-year CS Ph.D. candidate at UC Santa Barbara. Wang's research mainly focuses on Machine Learning and NLP, especially on the dialogue system, question answering and information retrieval. As for industrial experience, Wang has interned as at Amazon Alexa, Meta AI and Microsoft Research, accumulating hands-on experience for large-scale data preparation, model design, training, evaluation and deployment. Additionally, Wang has led a team of 5 PhD talents to participate the fierce Amazon Taskbot Challenge, competing with 10 university teams from 3 continents and successfully getting into the final round.
Saini is a BS/MS student at UC Santa Barbara. He’s the team’s youngest member, but don’t underestimate him. Before he started competing in the Alexa Prize, he trained multimodal neural networks for Google Maps; developed low-latency, aerospace-focused image processing algorithms; interned with Alexa (twice); and launched spacecraft. His years of experience designing reliable real-time systems, deep passion for interdisciplinary approaches to AI research, and entrepreneurial drive give him unique perspectives on open-domain dialogue.
Wang is a first-year Ph.D. student at University of California Santa Barbara. Before that, Wang obtained their master and bachelor degree from Rutgers University and Xi'an Jiaotong University, respectively. Wang has been dedicated to the research on natural language processing and dialogue systems for 3 years with a series first-author publications on top-tier Computer Science conferences, including EMNLP, AAAI, and SIGIR. Wang is trying to incorporate the pre-trained language scale language models into building up strong chatbots and eventually unifying the training paradigm for the two tasks.
Zhukova is a third-year graduate Linguistics student at UC Santa Barbara and a former Knowledge Engineer Intern at Alexa AI-Knowledge. She uses insights from linguistics to collect and better understand natural language data. Her research interests cover sociocultural linguistics (language and gender, language and identity) and computational social science (virtual assistants, digital avatars and emoji). Marina is experienced in developing annotation guidelines to ensure the quality of data for pre-trained language models for data automation, as well as in configuring knowledge architecture ontologies and designing voice and visual response to improve the customer experience.
Xifeng Yan - Faculty advisor
Yan's research focuses on the development of fundamental concepts and new principles of data mining, design intelligent algorithms and build scalable systems. His current concentrations include: modeling, managing, and mining large-scale, heterogeneous graphs; structural and statistical analysis of texts and their applications; intelligent solutions to cross-domain problems in bioinformatics, business intelligence, computer security, computer systems, and social science.