Recently I joined the TEAM Project, which focuses on research networks on the web. The project deals with issues like recommendation, text disambiguation, and metadata validation. In my part, I will do something a bit different: I will take a look at how a research fields is represented in such research networks.
So far, academic fields have been analyzed using metadata that came with the published articles. That includes, amongst others co-authorship, categories, keywords, and most importantly, citations. With this kind of metadata, it is possible to map out a research field from the position of the authors. Now, employing user generated data from research networks, it is possible to take a look at a field from a whole new viewpoint: that of the reader.
You might ask: why is that interesting? Well, meta-data from articles always only give you one part of the story. Co-citation and co-authorship analysis surely are sound ways to look at a field; but what if there are two groups of authors in different fields working on the same topic that just never publish together and never cite each other? In that case you will not get the connection between them. Most probably they will be using different language, so text analysis won’t help either. In come the readers: they might have identified that the authors are working on the same topic despite all the issues mentioned above. Furthermore, they might have grouped them together or used the same tags to describe their articles. If we analyze these groups and tags, we can find the connection, thus extending the field beyond its original borders.
That is not all; other interesting questions include: How are articles shared among researchers, and what does that say about interdisciplinarity in a field? Are the articles that are often read the ones that are often cited? As you can see, I am pretty enthusiastic about this. I could go on why I think that readership analysis is a good idea, but I am more inclined to get some early feedback: What other issues would be interesting to look at? What problems do you see with this kind of analysis?