Researchers at the University of Glasgow have over a decade of experience developing innovative big data analytics for search applications in both high volume and high velocity scenarios.
Terrier, an open source toolkit for building search engine applications and researching the technologies that underpin them, is the result of continuous research and development of innovative search technologies. Over the past decade, Terrier has grown into one of the top search platforms used for research and development internationally due to its extensive functionalities coverage and business-friendly open source licence.
Terrier is readily deployable on large-scale collections using Hadoop MapReduce technologies and provides an ideal platform for the rapid development and evaluation of scalable and effective search engine solutions for almost any type of data, whether it is for use on billions of Web documents, tweets, images or sensor streams. Indeed, Terrier has been used to tackle tasks such as expertise search; real-time event detection and tracking; summarisation of textual streams; analysis and search of medical records; as well as the development of search engine technologies for addressing information needs within future cities. The most recent open source release (v4.0) of Terrier includes the latest generation of fast index compression techniques, as well as the ability to develop and deploy advanced machine-learned ranking approaches (known as learning to rank) – which currently underpin major search engines such as Bing – allowing effective search ranking models combining a variety of features to be automatically learned.
Furthermore, Terrier is a driver of cutting-edge research within academia and industry. Within academia, the Terrier platform has been used to develop some of the most effective diversification and risk sensitive search techniques available today. Terrier has also been adopted as a research platform within industrial organisations such as Yahoo! Research, The National Archives, and the Italian National Centre for ICT in Public Administration for a wide variety of search tasks, ranging from Web search to sensitivity review for e-government. Moreover, search technologies built on Terrier have been shown to be effective in trials with end-users of commercial search engines such as TRIPDatabase.com.
Terrier is available to download for free from terrier.org.