As the latest result from a decade-long collaboration between ARG-tech and the BBC, our Evidence Toolkit app is today releasing new functionality allowing users to select articles of their choice helping distinguish fake news from true. A key part of this release is a new ‘Reason Checker’ which represents the first large scale deployment of argument mining technology in the wild. The full Evidence Toolkit app is now available on the BBC Taster platform.
The Centre for Argument Technology is advertising two new fully-funded PhD studentships.
Harnessing Big Data to Accelerate Evaluation of Infection Outbreak
Social media and local news hold huge untapped potential as indicators of infectious disease outbreaks and resistance threats, yet the automatic processing and sense-making of vast amount of natural language unstructured data is still an enormous challenge. The project will combine two innovative and emerging AI technologies – Argument Mining (Lawrence et al. 2017) and Ethos Analytics (Duthie et al. 2016) – in which areas the Centre is working with both IBM and the BBC. By building techniques to assess the content and credibility of such data at very large scale, we will deliver an early warning system for infection outbreaks. We will partner with Health Protection Scotland to align with the MRC’s Infections Priority Challenge (MRC Delivery Plan 2016-2020).
Duthie, R., Budzynska, K & Reed, C. (2016) “Mining Ethos in Political Debate” in Baroni, P., Stede, M. & Gordon, T. (eds) Proceedings of the Sixth International Conference on Computational Models of Argument (COMMA 2016), IOS Press, Potsdam, pp299-310
Lawrence, J., Snaith, M., Konat, B., Budzynska, K. & Reed, C. (2017) “Debating Technology for Dialogical Argument: Sensemaking, Engagement and Analytics”, ACM Transactions on Internet Technology, 17 (3)
Contact Dr Kasia Budzynska or apply online. Open to applicants from the EU.
Closing date: 20 April 2018.
The Artificial Co-Investigator: AI Systems as Team Members in Discovery for Medicine
We are becoming increasingly familiar with the idea of AI as a team member in demanding cognitive environments (see this article commissioned from us by BBC News, for example). In previous work we have shown how AI reasoning systems can synthesise medical research hypotheses that go on to be testable and publishable (Quinlan et al., 2012). Our goal in this project addresses the MRC’s Discovery for Medicine agenda, and TTL in particular (MRC Delivery Plan 2016-2020), through an ambitious unification of these two threads of work, using recent results in deep learning to build AI systems that contribute to collaborative research teams in natural language (Pease et al., 2017) – and we’ll know we’ve succeeded when the AI system is listed as a co-author on published results.
The era of precision medicine is promising to deliver individualised patient treatment, but, whilst individual genetic traits can be identified, the ability to direct specific treatment is lacking. The UK is delivering some of the largest datasets worldwide that can be analysed, such as the release of data by UK Biobank. These datasets offer the opportunity for AI to find credible combinations of genetic mutations that may be candidates for drug development.
In a collaboration between the University of Dundee and the Advanced Data Analysis Centre (ADAC) at the University of Nottingham, the student will (a) develop deep learning systems for identifying regularities in genetic datasets from the UK Biobank; (b) build on existing infrastructure to develop new software for human-machine hybrid research teams; and (c) explore innovative ways of explaining and justifying research hypotheses.
Pease, A., Lawrence, J., Budzynska, K., Corneli, J. & Reed, C. (2017) “Lakatos-style collaborative mathematics through dialectical, structured and abstract argumentation”, Artificial Intelligence, 246, pp181-219.
Quinlan, P., Thompson, A. & Reed, C. (2012) “An analysis and hypothesis generation platform for heterogeneous cancer databases” in Proceedings of the 4th International Conference on Computational Models of Argument (COMMA 2012), IOS Press, Vienna.
Contact Prof Chris Reed or apply online. Open to applicants from the EU.
Closing date: 20 April 2018.
For either position, candidates should have, or be expected to achieve, a degree preferably in artificial intelligence, computational linguistics, linguistics, cognitive science, computer science or related areas. Programming experience is an advantage. The starting date of the project will be 1st Sept 2018 or as soon as possible thereafter and will run for 4 years.
Stipendary payments are at UKRI levels (£14,777 for academic year 2018/19; tax-free). Applicants should meet MRC’s eligibility requirements (see www.mrc.ukri.org).
Both positions will be held in the Centre for Argument Technology in Computing at the University of Dundee (www.arg.tech), under the supervision of Prof. Chris Reed and Dr. Katarzyna Budzynska.
At the most recent REF, computer science research in the School was rated third in Scotland, with over three quarters of its research rated world leading or internationally excellent (4* or 3*). Dundee has been ranked amongst the top places in the world for scientists to work (The Scientist), and has one of the lowest costs of living in the UK.
We’re excited to be launching software with BBC News today.
Following up on the success of our work livetweeting argument analytics with BBC Radio 4′s Moral Maze and an associated Test Your Argument app last October, we have a new deployment of software to encourage thinking about thinking in partnership with BBC News. BBC News and BBC Academy are responsible for School Report, an educational initiative for improving critical thinking skills and helping young people distinguish real news from fake. BBC School Report is being made available to 1000 schools. For 11 to 15 year olds, the BBC is partnering with Aardman Animations in providing an educational game around the theme of fake news. For the older age group of 16-18 year olds, the BBC is partnering with the Centre for Argument Technology to deliver The Evidence Toolkit, for helping users analyse news articles to help them better understand what’s fake and what’s not. With a second phase to be released in a couple of days, this software represents the world’s first large-scale deployment of argument mining technology in the wild, and we’re thrilled that our research in the area can contribute to an educational mission led by the BBC focusing on such a pressing issue as fake news.
The full story is available with the BBC.
Chris Reed has written an article that has appeared with Newsweek, and today featured on its Tech & Science homepage. It focuses on the role of AI in helping to teach and to contribute to human argument & debate.
This week we have two visitors to the Centre. Nuria Franco-Guillen from Griffith University is visiting all week and is giving a seminar on Thursday morning at our regular ARGi session (10am in the ARG lab). She works with John Parkinson at Griffith on a project tracking the national conversation around key political issues and they are interested in using some of our argument mining techniques to process very large data sets.
On Friday we are joined by Alejando Ramos Soto from the University of Santiago de Compostela and currently a visiting researcher at the University of Aberdeen. Following our work delivering analytics for the BBC, Alejandro is here to work on the connection between Natural Language Generation Techniques and debate analytics. He’ll be giving a talk in the ARG lab at 11.30 on Friday.
A stimulating week!
As well as edited copy from ARG-tech, the page also has links through to the ARG-tech developed Test Your Argument application hosted on BBC Taster.
The Centre for Argument Technology (ARG-tech) at the University of Dundee has signed an agreement with the BBC to deliver a suite of argument technology to be piloted in conjunction with BBC programming: a special edition of the Moral Maze on BBC Radio 4 broadcast at 8pm on 11 Oct 2017 and repeated on 14 Oct; a one-off TV debate hosted by Anne Robinson and broadcast at 9pm on Monday 16 Oct on BBC2, and a further episode of the Moral Maze from the BBC Archive, all tackling the issue of abortion on the occasion this month of the 50th anniversary of the Abortion Act in the UK.
Following a decade of collaboration between ARG-tech and BBC Radio 4′s Moral Maze, this new agreement will see funding of work to prepare three exciting new argument technology applications. First, a set of argument analytics to be published on the Moral Maze website will give audiences an insight into the debate, and a web page on BBC Radio 4 will use them to offer tips on how to improve their own arguments. We will also be livetweeting some of the analytics as the programmes are aired. Second, BBC Taster will host ‘Test Your Argument’ for users to try their hand at building better arguments. And third, ARG-tech will host ‘Debater,’ a system that allows users to take on the role of the chair of the Moral Maze and recreate their own new debates. For more information, check out an article commissioned by the BBC as a part of their Expert Network (at www.bbc.co.uk/news/technology-41010848) and ARG-tech’s own site at www.arg.tech.
Chris Reed has been commissioned by the BBC to write an article on Argument Technology, which appears on the BBC News website today. The article forms a part of the BBC’s Expert Network, whereby background and analysis are solicited from universities, think-tanks and other organisations that can offer context and depth to content across the BBC. The piece is aimed at a general audience and offers a unique opportunity to the field for raising the profile of computational models of argument, argument mining and argument technology in general.
Our collaboration with Jean Wagemans of UVA continues and during his recent visit to ARGtech he also gave a talk ‘A factorial approach to argument classifaction’. Jean is currently working as a senior researcher at the Amsterdam Centre for Language and Communication (ACLC) of the same university. He is a co-author of the Handbook of Argumentation Theory and has published articles on classical dialectic and rhetoric, pragma-dialectics, and the characteristics of scientific argumentation. Find out more http://uva.academia.edu/jeanhmwagemans.