The Axiomteam is asked to give a user case/test case such that the relevance of the project as a whole can be explained. However, we’re no experts in a) this period, or b) these authors. So, best to i) look for questions that can be answered by quantitative results (then we don’t have to read up), and/or ii) hook it up to The Model (i.e. The Classical Model of Science).
So, two lines of research:
A. The Model Approach
Search for ‘Scientia’ and ‘Wetenschap’, then for ‘Method’, ‘Veritate’, etc. (search for CMS-terms). Then (by trial and error) look if you can find a pattern; maybe only scientists come up, and not the poets, or maybe the scientists can be grouped according to certain criteria purely based on terms they use. (A trial and error approach of course.)
B. The Quantitative Approach
1) Hein van den Berg discovered something curious: there are periods in which Descartes writes virtually no letters, and periods in which he seems to have nothing else on his hands. This pattern seems to be based on publications: whenever Descartes publishes a book, he starts writing letters. It could be interesting to compare different authors with respect to this habit: do they mostly write before they publish to get ideas, or after publications to sell and defend their book?
2) Another possible pattern to look for is whether certain scientists can be grouped together on the basis of their expertise using only Co-citation and Correspondent Network data, i.e. distinguish biologist from mathematicians on the basis of the people with whom they correspond and whom they mention in their letters. Or see whether authors that understood misunderstood someone were also not corresponding with that author.
N.B. Most of the quantitative based research questions (such as discovering centers of excellence, or research schools on the basis of the letters) cannot be answered as long as the corpus is not complete!
Some suggestions for improvement of the tool:
1) Improve visualization tool ‘Geographical Map’ such that a) one can see the direction of the letter, b) the map does not get blurred into one big red dot, c) it is possible to compare time spans of the selected letters (i.e. that one can e.g. choose to represent all letters from 1600-1620 to be yellow, all letters from 1620-1640 to be orange), d) you can select and read a letter by clicking on a line.
2) Improve visualization tool ‘Correspondent Network’ such that a) one can increase the distance between authors names to make it more readable, b) one can select/deselect correspondents on the basis of the number/locations of letters sent/received, c) one can compare two or more correspondents networks at the same time. (Both as a web, but also simply as a list, to see whether there is any overlap between correspondents.)
3) Improve visualization tool ‘Co-citation Graph’ along similar lines as 2), where c) should then be seen as ‘to see whether there is any overlap in people being mentioned’. Also, instead of only letting size determining who is mentioned most, displaying the number of times someone is mentioned would be nice as well.
4) It seems the search results are stemmed, but not lemmatized. Is this correct? (Lemmatized would definitely be a huge improvement.)
5) Have important dates of the author also represented in the timeline. Such as date of publication of certain works, such that one can see whether letters were written before or after a publication.
6) When selecting senders/recipients it would be nice to be able to select on ‘minimum or maximum number of letters’ sent/received. (To exclude certain correspondents.)
7) Also, what do the lines in the ‘Co-citation graph’ mean?
How to proceed?
First meet with ePistolarium people during the kickoff of GLAMMap, maybe some of our suggestions have already been taken care of – this would save time when we start with our Action plan.