This is a guest post contributed by Stina Johansson, Bibliometrician at Chalmers University of Technology Library.
As a librarian and bibliometric analyst at Chalmers University of Technology, I am surrounded by data describing our university’s research in collaboration with other universities, the industry or other parts of society. More than once have I called our local repository (Chalmers Publications Library) a gold mine. In these data intensive times, not only are we, as analysts, able to track impact through citations, but we are also able to look at and understand the nature of our research and its’ outreach through a broader perspective. We can analyze, for example:
- the spread of research results in social media through alternative metrics,
- the nature of collaboration through co-authorship and,
- as a complement to publications’ metadata, metadata covering projects going on at the university.
Through local and global systems, these types of metadata are becoming more and more useful and available to us. All these data sources combined give us the possibility to look at research from a broader perspective, and to understand impact from a broader perspective than through classic bibliometrics alone. Using the data that is available, we are able to look for patterns and trends when it comes to the university’s role in society, especially its outreach to society, and to academia and the industry sector through various forms of collaboration.
How do we choose to work with this data? Are there as many alternative methods as there are data sources? We, a group of librarians at Chalmers Library with a passion for metadata, have found that different methods of data visualization have helped us both in terms of handling our data, and in analyzing and presenting it to the university. In our experience, one of the most interesting aspects of visualizations is that they, when successful, encourage discussions and work as stepping stones to further and more in depth questions about the nature of our research.
The first visualizations we presented, using data from our local repository, were geospatial maps plotting our Chalmers’ national and international research collaboration, measured through co-authorship. These were interesting because they showed a geospatial pattern, but they also encouraged a conversion on research collaboration. Now we have an image of the spread of collaboration but can we also answer questions about how our collaboration patterns have evolved over time? What impact does this collaboration have compared to that? In this respect, the images we have created showing geospatial patterns have helped us move forward in our work.
One more recent experiment is a network analysis I’ve got underway in collaboration with a Chalmers PhD student, studying the sociology of science. In this visualization project, we use publications’ metadata from our local repository and we focus on the author field of publications. Not only do we want to make a good visual representation of the social network existing within our repository, we also want to explore patterns and trends in the network – both simply by looking at it, but also through social network analysis measures (describing the density of the network, detecting subgroups within the network, important roles like stars, bridges and gatekeepers among individual researchers). We have already seen that co-authorship patterns, and perhaps social practices concerning authorship, vary between different departments at our university. This of course makes us want to dig deeper into our network.
Questions we have posed when looking at the network visualization are:
- What types of practices can we detect?
- What different types of roles are visible in the community?
- How can we use social network analysis as an alternative method to understand our research community better?
- How would a set of ‘social network analysis metrics’, complement our classic bibliometric metrics?
- The university is concerned with equality between male and female researchers; how can we, through local data and social network analysis, apply a gender perspective to the author based networks images we have created?
At this point we pose more questions than we answer, and is that not a sign of development? And a sign that this method is interesting in terms of what is does to stimulate discussion on what we can do with our metadata?
As I’ve expressed, there are many pros using data visualization techniques, but of course there are also limitations and responsibilities. They say that you should be careful when communicating through images, because images are both effective and powerful, and can manipulate the mind.
Think of a line graph, a classic visualization- is there a more effective way of showing a trend? And how easily can it be manipulated? Fairly easily, as it turns out: through changing the scale of the image or manipulating the data used to make it (like we see in Figure 2).
Through our experiences experimenting with data visualization methods at Chalmers Library, we have grown to appreciate and respect data visualization as a powerful tool, most of all as a help for us to better understand our data, presenting it to the university in an easily readable format, and to appreciate the images created as stepping stones for further analysis. Yet we have just started this quest, and we are eager to see where our visualization projects lead us.