New research using Twitter helps to better understand public opinion of drug related crime
27 March 2018
CSRI Research Assistant Daniel Harborne has been spending time in Hokkaido University Japan, collaborating with academics there on computer science research.
Last summer Daniel spent two months at Araki Labs, Graduate School of Information Science and Technology, Hokkaido University for an internship which was partly sponsored by Cardiff University's Global Opportunities Centre. The lab's focus is Natural Language Processing and Common-sense Reasoning, which are two areas within Artificial Intelligence.
During the internship, Daniel used research from the DAIS ITA project relating to situational understanding, which is another part of AI and decision support, and IBM's Controlled English technology to develop a dialogue interface to operate a city surveillance/information dashboard.
Situational understanding is the bringing together of distributed sensors and systems to analyse the current state of the world, or a slice of it, and to predict future states. The aim is to provide decision support to humans and other machine systems. In addition, part of situational understanding is also taking feedback from those humans/other machines to improve the systems performance or resilience. Situational understanding is a key part of systems used within the military and public sectors in tactical situations, as well as operations like disaster relief efforts.
Daniel recently made a second visit to Hokkaido University, this time staying for a month as a visiting researcher to collaborate with the same colleagues on new research. The research was focused on using twitter to assess the public opinion on drug related crimes including possession and dealing, and comparing that to the strength of punishment the law applies to those crimes.
This research has two implications:
- We can see whether law makers are punishing crimes in line with public opinion.
- We can explore the learning of public morals and ethics, which will help guide the development of AI that act as the public wishes/expects them to.
You can read more about this research on the IBM resource page, by clicking here.