A perhaps unexpected output from the Library and Information Science Research Coalition’s DREaM project is a corpus of introductions to research methods, with videos, slides and reports.
A number of the workshop themes particularly tickled my fancy, and I’m gradually working my way through them, starting with social network analysis – see my Storify of the session. A DREaM online community has been created and the project is also running with the idea of a workshop cadre – no doubt they will be analysing themselves! (Update: see below for the results.)
The other sessions at the event also look worthy of a closer look, in particular those on ethnography and discourse analysis, to give some ideas in relation to textual and sentiment analysis of backchannels. Thankfully the workshop was expertly amplified. Scan the CoveritLive archive for a flavour of discussions, including comments from 18 remote attendees or follow up individual sessions from the organisers’ review or the participants’ reviews. I found those by David Jarman and Paula Goodale particularly useful.
A common theme seems to be the availability of ‘easy to use’ software enabling those with Higher Maths at best to have a go. The tricky bit is interpreting the results.
- data mining (workshop 3) – inc text mining to eg classify attitudes in social media, link analysis for clustering – essentially creating models from data
- discourse analysis (workshop 1) – documentary sources such as records of events used as an example; discourse can be contextual, rhetorical, action-oriented, constructed and constructive…is Twitter discourse? analysis of what people say (or write) through understanding of the context in which it is said: the social norms embedded in that context, and how language is used to construct a way of seeing the world
- ethnography (workshop 1)
- techniques from history (workshop 2) – this exercises some underused muscles in my brain and also ties in with my thinking around the anti-social
- horizon scanning (workshop 3)
At workshop 2 Mike Thelwall gave a useful intro to link analysis, altmetrics and sentiment analysis. Head of the Statistical Cybermetrics Research Group at Wolverhampton, Mike is author of an introduction to webometrics and loads of interesting looking publications. He also spoke at the #nsmnss launch event, giving an overview of using the social web to study something else (video | slides).
Webometrics is defined as gathering, processing and analysing large scale data from the web, a research method mainly used within information science. Similar outputs in other disciplines, eg link analyses in computer science, networks in sociology, but based on different principles. I can’t help feeling this is a typical librarianship story…anyway, useful as a first pass at a research question, but need to be aware of its limitations, in particular in relation to sampling.
Link analysis can be used to look at online academic communication – which academic sites interlink? what interlinking patterns exist? which pages have the most online impact? Bonus: link network diagrams!
- SocSciBot – Web crawler and link analyser
- Webometric Analyst – link analysis, network diagrams and more for web sites, Twitter and YouTube
- IssueCrawler is another link analysis tool, much used in sociology
Sentiment analysis always looks interesting. One project looked at 30 major events on Twitter, finding a slight increase in negativity during popular events, although many tweets neutral/reporting, raising the question of whether Twitter really reflects how people feel. Sentiment is implicit – may be more explicit on eg blogs.
An analysis of the DREaM network
At the first workshop participants were asked:
- Whether they were aware of the knowledge/expertise of others in the room.
- Whether they had had direct acquaintance with individuals in the room prior to the workshop.
Initial findings were that the network was fairly well connected, with 20 cliques/clusters. The acquaintanceship network was slightly more connected than the knowledge and expertise network – see the slides for details.
In her preview interview Louise Cooke states:
The results from the initial analysis were interesting, but led us to question whether the networks would change over the course of the three DREaM workshops…Initially, the networks were very centralised around a few key individuals. Ideally, we’d like to see that subsequently new connections have been made, and a more ‘equal’, but also more densely connected, pattern has emerged.
Louise has updated her analysis – again, see the slides for details. By the third workshop the network was much more densely connected and less dependent on a key individuals – the question now is how to sustain and develop this, as well as how to take the DREaM community forward.
I’ve also done a quick Storify of the session, being a bit of a Storify addict at the moment. I still find it fiddly, but it does complement my rather texty blog.
Hazel Hall subsequently wrote a paper on the analsysis – see Mapping a community: a SNA case study.