Wednesday, October 26, 2022
2:30-3:30PM ET
MIT Building 32-D463, Stata Center – Star Conference Room

Speaker: Alessandro Moschitti
Speaker Affiliation: Amazon Alexa
Host: Jim Glass
Host Affiliation: MIT CSAIL

Abstract:
Personal assistants, e.g., Amazon Alexa, Google Home, and Apple Siri, provide interesting challenges for Question Answering research. These systems operate in open domain, where the question complexity and variability are set by the information needs of millions of customers. Winning these challenges requires the use of web content in the form of unstructured text. In this talk, we will describe state-of-the-art NLP/IR technology, e.g., Transformer models, for building open-domain QA systems based on web data, which can answer millions of questions with impressive accuracy. In particular, we will (i) describe QA system architecture and components, (ii) methods to make the usage of Transformer models efficient, e.g., our Cascade Transformer, and (iii) sequence to sequence models for jointly modeling answer candidates, thus producing even more accurate responses. Finally, we will provide a practical demonstration with a standard Alexa device.

Bio:
Alessandro Moschitti is a Principal Research Scientist of Amazon Alexa AI Web Information, where he has been leading the science of Web-based Question Answering since 2018, and more recently, most ML/NLP/IR technology of the entire Alexa Information. He obtained his Ph.D. in CS from the University of Rome in 2003, and then did his postdoc at The University of Texas at Dallas for two years. In 2007, he joined the CS Dept. of the University of Trento, Italy for 15 years. Since 2009 to 2011, he participated to the Jeopardy! Grand Challenge with the IBM Watson Research center, and collaborated with them until 2015. He was a Principal Scientist of the Qatar Computing Research Institute (QCRI) for five years (2013-2018). His expertise concerns theoretical and applied machine learning in the areas of NLP, IR and Data Mining. He is well-known for his work on structural kernels and neural networks for syntactic/semantic inference over text, documented by more than 300 scientific articles. He has received four IBM Faculty Awards, one Google Faculty Award, and five best paper awards. He was the General Chair of EMNLP 2014, a PC co-Chair of CoNLL 2015, and has had a chair role in more than 60 conferences and workshops. He is currently a Senior Associate editor of ACM Computing Survey, long-term Associate Editor of JAIR, and General Chair of EACL 2023. He has led more than 25 research projects, e.g., with MIT CSAIL.

For more information please contact: Marcia G. Davidson, 617-253-3049, marcia@csail.mit.edu