Cookies on this website
We use cookies to ensure that we give you the best experience on our website. If you click 'Continue' we'll assume that you are happy to receive all cookies and you won't see this message again. Click 'Find out more' for information on how to change your cookie settings.
It2019s not big data that discriminates 2013 it2019s the people that use it

PI: Georg Gottlob

Department: Computer Science

Large enterprises store massive amounts of data, from which they wish to infer knowledge (at times in combination with external datastores) to improve their decision making. For instance, a bank wants to detect fraudulent transactions or decide whether a customer is creditworthy. A retail company would like to optimise their pricing decisions based on both internal data (e.g. historic sales records) and external data (current product prices of competitors) available on the Web. The steps needed to query the data are (1) data provisioning: to perform preparatory data wrangling tasks by selecting relevant data from various internal and external sources, transforming and merging the data, and storing the data in appropriate tables into a database; (2) data analysis using machine learning, data mining methods, or logic reasoning and (3) drawing inferences to make business decisions.

In the context of the VADA EPSRC-funded project, Prof Gottlob’s group is developing a novel unified methodology to do (1), (2), and (3) within a single rule-based system, and has built a pilot prototype of such a system. The technology is based on the data-processing language Datalog and the concept of transducers (devices that convert variations in a physical quantity into an electrical signal). Recent extensions of Datalog developed in the group have successfully demonstrated a good trade-off between expressive power and computational efficiency. Despite the early stage of development, these methods and initial system have attracted strong interest from diverse industrial contacts, especially in the banking and security areas.