University of Cambridge > > Technical Talks - Department of Computer Science and Technology  > Finding Signals in Twitter with ML/NLP at Bloomberg

Finding Signals in Twitter with ML/NLP at Bloomberg

Add to your list(s) Download to your calendar using vCal

  • UserMinjie Xu - Bloomberg Software Engineer, Social Media Analytics
  • ClockTuesday 25 April 2017, 13:00-14:00
  • HouseLT1, Computer Laboratory.

If you have a question about this talk, please contact Jan Samols.

Ever wondered how to: • Find out the most talked-about companies, topics on Twitter? • Get alerted when something breaks out? • Obtain a quick overview of what’s going on (e.g. when Donald Trump posts again)?

Bloomberg is all about up-to-date financial data and analysis. In the current social media era, Twitter has proven itself an indispensable source of information as we frequently see information posted or shared on Twitter being big market movers.

Thanks to a proprietary access deal with Twitter, we can receive every single tweet that matches a set of rules in near real-time, on top of which we then perform a series of ML/NLP analytics, extracting meaningful signals and filtering out irrelevant noise.

To this end, we have built a range of products around Twitter, including trending companies, company sentiment, topic streams, etc.

In this talk, Minjie will be covering several topics: a) A general overview of how Bloomberg processes Twitter data b) Highlight our in-house Twitter topic streams system c) Present several interesting Machine Learning topics therein and if time allows, d) Discuss a more recent project that involves learning good document embeddings (via variational auto-encoders) for use in clustering

Food and the opportunity to network after the talk in FW26 .

This talk is part of the Technical Talks - Department of Computer Science and Technology series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity