Wednesday, June 24, 2009

i3 reports: identifying email types

Another report from the i3 conference that is taking place at Robert Gordon University, Aberdeen, Scotland. The second paper in the first session today was from Malcolm Clark, who reported on a part of his research, which is examining people's use of email. In particular he was looking at email structure and layout, looking for clues as to what features and forms in the emails were being used to make decisions about what to do with email. This is useful for studying how people filter and identify emails. I can imagine people producing emails can also earn from this (including spammers... but also library and information people: getting better attention for your emails)
He used the term genre to describe something which has a recognisable purpose (in terms of topics, arguments etc.) and form (which you can tell from readily observable features, such as language, communication medium or structural features).
He used eyetracking equipment to follow where eyes moved and how they behaved in reaction to what was on the screen. There were 24 participants. There were various kinds of email e.g. call for papers, library notice, information on a seminar, spam. Each item was presented 4 times, structured in different ways, including ones in which the structure was kept but the email text was replaced with XXs or 9s, one with the content and no structure, and one with all the structure & content removed. People were asked to identify what kind of email it was. He looked at various items such as genres correctly identified, amount of time spent on each. In terms of recognising what type of email itwas:
87% of the ones in their original form were correctly categorised, 77% of those with the original format by the content converted to X and 9s. If the structure (e.. coumns, paragraph sacing) was taken away, 68% of messages were recognised. 27% still recognised the type of email when it had no structure andwas all Xs and 9s - apparently certain types of email were recognised because they normally had lots of numbers (replaced by 9s). People skimmed the shape of the text: different areas were focused on depending on the structure. With emails with no structure there was a lot of scanning behaviour. Some useful features were identified for types of email: e.g. Calls for papers: dates, centred blocks, titles
Cinema listsings: blocks and numerical content
Library: people went straight to details of a book at the end of desciptive passage (perhaps flags up that if you currently circulate listings with abstracts, it may be that people don't want the abstracts!)
Spam - looking for keywords, address, emboldened text
One of the interesting methodological questions was whether you could use a constructivist concept like "genre" with a quantitative approach (observation in the lab).
Photo by Sheila Webber: Reception last night, with alumni of RGU, to celebrate 40 years of the Department

No comments: