Andrew Salway - past projects

NTAP: Networks of Texts and People
Funded by: The Research Council of Norway, 2012-17
Principal Investigator, Scientific Coordinator: Dag Elgesem, Andrew Salway
My role: NTAP developed methods and tools to detect, analyse and visualize the distribution, flow and development of knowledge and opinions across online social networks. I was the co-proposer and scientific co-ordinator of this project in which my group collaborated with Prof. Dag Elgesem (PI) and his colleagues in the Department of Information Science and Media Studies, University of Bergen. I worked particularly on: designing and building blog corpora comprising text and network data (Salway et al. 2016; Salway and Hofland 2015; Salway, Hofland and Touileb 2013); developing a novel data-driven text mining technique, based on the distributional hypothesis, to identify salient information structures in discourses (Salway and Touileb 2014; Salway, Touileb and Tvinnereim 2014; Touileb and Salway 2014); and, combining text analytics, network analytics and visualization for digital media analysis (Diakopoulos et al. 2014; Diakopoulos, Zhang and Salway 2013; Salway, Diakopoulos and Elgesem 2012).

LINGCLIM: Linguistic Representations of Climate Change Discourse
Funded by: The Research Council of Norway, 2013-16
Principal Investigators: Prof. Kjersti Fløttum, Department of Foreign Languages, University of Bergen
My role: LINGCLIM generated new knowledge about the role of language in online climate discourse with a team from linguistics, psychology, political science and media science. My role was to design and execute corpus-based linguistic analyses as per research plans developed jointly with the project team, and to interpret and write up results for conference presentations, journal papers and a chapter in a book about the project (Fløttum, Dahl and Salway in review; Salway 2017; Salway, Elgesem and Fløttum 2015; Fløttum et al. 2014).

TRIPOD: automatic image caption creation
Funded by: EU 6th Framework, 2006-2009
Principal Investigators: Dr Ross Purves and Dr Mark Sanderson
My role: I led research and development in a workpackage concerned with extracting and integrating information about tourist sites from the web.
Outputs: I developed a Java web service for extracting semi-structured facts from the web, written up as Salway et al (2010).

Television in Words (TIWO)
Funded by: EPSRC GR/R67194/01, £55,315, January 2002-2005. 
Principal Investigator: Dr Andrew Salway
Aim: To develop a computational understanding of narrative in multimedia systems. 
Application: Technology to assist the production of audio description, and technology to reuse audio description for video retrieval and browsing applications for films.
Partners: Royal National Institute of the Blind, the BBC, ITFC and Softel
Outputs: Five publications to date.  Novel technologies for browsing and analysing films.  Technology transfer is under discussion with BBC Access Services, BBC Archive and Philips.  Three PhD dissertations.

Code Zebra
Funded by: various Canadian agencies and international companies, 2002-2005.
Project Leader: Sara Diamond (then Artistic Director, Banff New Media Institute; now President Ontario College of Art and Design).
Aim: Language visualisation to study and enhance collaboration in online chat environments.
My role: I led workshops with scientists, artists and engineers on corpus-based analysis of online conversations and consulted on technologies for text classification, neural networks and software development strategies.
Output: Software for online chat with enhanced language visualisation.

Analysing Image-Text Relations
Collaboration with: Dr Radan Martinec, University of the Arts, London
Aim: To develop a semiotic theory of image-text relations that is grounded in semiotic theory and that can be applied in the development of multimedia computing systems for image retrieval, hypermedia browsing and multimedia generation. 
Output: Publication in Journal of Visual Communication.

Information Extraction from the Tate Gallery website
Collaboration with: Rachel Bhandari, Tate Gallery
Aim: To identify linguistic patterns in a 1,000,000 word corpus of painting captions and artist biographies that could be exploited for automatic image indexing.
Output: Seminar at Tate Britain and a refereed international conference paper.

Knowledge-based Annotation and Browsing of Dance Videos
Collaboration with: Department of Dance Studies, Uni. of Surrey
Aim: to analyse the language used by experts to describe and interpret dance sequences, and to extract information from such texts in order to index digital dance video
Output: PhD dissertation, 'Video Annotation: the role of collateral text.'  A prototype dance video annotation and browsing system.

Scene-of-Crime Information System (SOCIS)
Funded by: EPSRC GR/M89041/01, £326,316, 1999-2002
Principal Investigators: Prof. Khurshid Ahmad (Uni. of Surrey) and Prof. Yorick Wilks (Uni. of Sheffield)
Aim: To explore the link between language and vision computationally.
Application: Automatic generation of representations of digital images from the spoken descriptions of experts, leading to an image retrieval system for crime-scene investigations.
My role: I contributed to the proposal when I was a PhD student.  In 2000 I was responsible for managing the project RA and PhD students, chairing project round table meetings with UK scene-of-crime officers and managers, and conducting data elicitation exercises.

Safe Design with Information Systems (Safe-DIS)
Funded by: EPSRC / DTI GR/H95235, £220,000, 1994-7
Principal Investigators: Prof. K. Ahmad (Uni. of Surrey) and Dr R. Price (Wallingford Software)
Aim / Application: To enhance the role of information systems in making safety-related information available to design engineers in a timely and effective manner.
My role: I worked as Research Assistant with responsibility for project management (there was another RA and a PhD student) and for organising meetings with an industry round table.  The project was awarded top marks for ‘Management and Use of Resources’.  My research focus was on a corpus-based study of the ‘language of safety’.