Conversations Based on Search Engine Result Pages
Episode details
Episode description
How might we convey the information that is traditionally returned by a search engine in the form of a complex search engine result page (SERP) in a meaningful and natural conversation? In the talk, Maarten starts from recent work on so-called background based conversations, where a conversational agent has access to additional background information to help it generate more natural and appropriate responses. Then, he talks about ongoing work on our next step: SERP-based conversations. He explains the task definitions, describes pipelines (subtasks), baselines, datasets, etc. Finally, Maarten describes the differences between background-based and SERP-based conversations and their relations to other, related tasks. Work on SERP-based conversations is in its early stages, leaving lots of opportunities for follow-up research. Based on joint work with Zhumin Chen, Jun Ma, Chuan Meng, Christof Monz, Pengjie Ren, Svitlana Vakulenko, and Nikos Voskarides. Talk slides: https://www.microsoft.com/en-us/research/uploads/prod/2020/01/44612_compressed.pdf See more on this and other MSR AI Distinguished Lecture Series talks: https://www.microsoft.com/en-us/research/videos/ai-distinguished-lecture-series/
Watch also from the same channel
More episodes from this channel
You might be interested in
