InfoTypography: Information Visualisation through and for Text

In several fields where text is the main source of information, it is important to be able to represent associated data about the text alongside it. For example, in the speech recognition field, it is often necessary to display how certain we are that a word has been recognised accurately. Similarly, in advanced thematic search, we might want to provide information about how much a certain word, sentence, or paragraph, relate to the topic we are interested in.

In this work I will investigate a new visualisation technique that allows representing data that is associated to text in-situ (within the text itself), without requiring additional visual artefact. The technique, called InfoTypography, is based on the idea that we can modify typographic parameters of the letters (glyphs) to efficiently represent information. For example, the letters of a word can be made to have more prominent serifs if that word has been recognised with more certainty or if it is more relevant to the topic being searched.

The technique might offer singular advantages beyond the fact that it does not require additional visual artefacts to "pollute" the text. For example, there are many different typographic parameters that could be used to represent many different variables. The purpose of this research is to investigate the properties of the technique, to characterise the best ways to take advantage of it, and to build a system that will make it available to people who need to represent this kind of data. Finally, the technique will be put in practice through an application within the digital humanities field.

Impact to the Economy:

This proposal is aimed at enabling new ways to present and understand data through a novel technique based on typography. A fundamental component of the promise of big data and evidence-based decision making is the ability of humans to make sense of information in an efficient way. Novel techniques that are useful to represent data have therefore the potential to facilitate the big data and evidence-based decision making endeavours, especially if this is for types of data that have been traditionally difficult to represent (the case of text).

The proposed technique can be leveraged by the growing industry of data visualisation businesses, which is also increasingly relevant in the UK but that, unfortunately, is currently led mostly by non-European businesses. For an impression of the size and economic importance of the field, Tableau Software, a U.S.-based start-up visualisation company, went public recently (NYSE:DATA) and it has a current market capitalization of well over $US 4 billion. The application of this technique can provide advantages to U.K. based projects looking to visualise text-based data.

Impact to People:

This project will employ a research assistant and will count with the collaboration of a PhD student and an undergrad research intern. The assistant will closely collaborate with the PI and the PhD student to develop a new system and gain new evidence about the benefits and specifics of the technique. The RA will therefore gain experience in supervision and the development of cutting-edge visualisation systems, their application to real data, as well as in the design, execution and analysis of experiments. This wide-range of skills and balance between empirical and applied research will help the RA in the process of developing as an academic, with unique options to become a practitioner in the use of a new technique.

The PhD student will benefit from the experience of working within a team of top-level academics, and will extent the potential reach of his research. The undergraduate intern will get the opportunity to work on a real research topic, and learn to do research by doing.

Impact to Society:

The application of techniques to visualize data within text such as the one that I propose can enable analysis in economic areas that have not been able to fully benefit from the potential of information visualisation and visual analytics. This is particularly relevant for the digitisation and commercial exploitation of textual historical records (e.g., newspapers see: http://news.bbc.co.uk/1/hi/scotland/tayside_and_central/8688897.stm ).

Indirectly, applications of the technique in the digital humanities can also indirectly benefit our knowledge of the British culture, current society, and our identity. This is particularly reflected by the Palimpsest project, the case study chosen as first application of the technique, which explores literary Edinburgh.