Topic Streams of Financially Important News

Speaker

Miles Osborne

Abstract

Twitter nowadays is more than just stories about cats in trees and what I had for dinner. It is now a serious source of market moving news - stories on Twitter that can cause real changes in stock markets. But with hundreds of millions of posts per day, how can we quickly identify stories that matter to our clients?

Miles will sketch how Bloomberg has used online machine learning methods to find and organise incoming stories into real time streams of related content, processing more than 8k tweets per second on a single core. Along the way he will talk about open problems such as trust, multilinguality and market manipulation.

Speaker Bio

Miles Osborne is a senior research scientist at Bloomberg, working in News Search. Prior to this he was a Reader at Edinburgh University and has spent sabbaticals at Johns Hopkins University and Google. His research interests span Social Media, Natural Language Processing, Machine Translation, Information Retrieval, Machine Learning and Streaming Algorithms.