11th GATE Training Course: Large Scale Text and Social Media Analytics with GATE

By Genevieve Gorrell

Every year for the last decade, the GATE team at Sheffield have been delivering summer courses helping people get to grips with GATE technology. One year we even ran a second course in Montreal! It's always a challenge deciding what to include. GATE has been around for almost a quarter of a century, and in that time it has organically grown to include a wide variety of technologies too numerous to cover in a week-long course and adapt to the changing needs of our users during one of the most technologically exciting periods in history. But under the capable leadership of Diana Maynard and Kalina Bontcheva, we've learned to squeeze the most useful material into the limited time available, helping beginners to get started with GATE without overwhelming them, as well as empowering more experienced users to see the potential to push it into new territory.

Recent years have seen a surge of interest in social media. These media offer potential for commercial users to deepen their understanding of their customers, and for researchers to explore and understand the ways in which these media are affecting society, as well as using social media data for various other research purposes. For this reason, we have positioned social media as a central theme for the course, which most students seem to find accessible and interesting. It provides an opportunity to showcase GATE's Twitter support, such as the GATE Twitter Collector, and to draw examples from our own work on social media within the Societal Debates theme of SoBigData. For example, GATE's Mimir semantic search engine, in addition to core GATE natural language processing functionality, has been instrumental in enabling us to work with the very large social media corpora we made use of in our work on abuse towards politicians and media influence on Brexit. However, there are also plenty of examples illustrating how GATE can be applied to other popular areas, such as analysis of news or medical text.

I've been teaching GATE's machine learning offering for most of the time the course has been running, and therefore I've had the opportunity to explore different ways of helping people to get a handle on what can seem an intimidating topic to those who aren't already familiar with it. Machine learning is challenging to teach to a mixed audience because it's such a large field and the time is limited. It's also an important one though, as it's increasingly a part of the public discourse, and many students are excited to learn about the ways they can incorporate machine learning into their work using GATE. Johann Petrak has taken the lead on keeping the GATE Learning Framework up to date with the latest developments in this rapidly evolving field, and I'm always proud and excited to teach something new that's been added since the last course.

It's evident from the discussions during lunch and tea breaks that students are eager to talk to us about how they are using GATE, and how they would like to use it. I think one of the most valuable things about the course is the opportunity it provides for the students to talk to us about what they are doing with GATE, and for us to be inspired by the range of uses to which GATE is being put. Here is some of the feedback we received from students this year:

"Last week was one of the most useful courses I have done. Overall I think it was pitched really well given the range of technical abilities."

"Thank you all for such an informative and well-delivered course. I was a little worried about whether I'd be able to pick it up as I don’t have a background in programming, but I learned so much and the trainers were all very helpful and patient."