Currently, I'm a few months in a PhD track on the applications of process mining to the archival sciences. The idea of the research project is to devise algorithms that would enable government officials to (semi) automatically search through lots of different digital documents (including e-mails, text messages, and others) which would help them gain an understanding of complex political processes and actions carried out by the government in the past.
Although I think this is an interesting subject, I also have ideas for papers in other research areas. My academic background lies in mathematics. Now that I learn more about finite-state machines, push-down automata, and other related concepts through a course on computational complexity theory, I find myself forming thoughts about defining the Kolmogorov complexity of certain classes of finite automata. Though this may have some overlap with process mining, the subject matter of a paper in this direction would diverge a bit from the original overarching research question.
Questions: In general, to what extent should the papers written for a PhD dissertation form a coherent whole? To what degree is variation allowed among the articles? Does this maximum degree of variation depend on the field in which one is doing the PhD?
