This question probably has a somewhat unusual backstory, which is perhaps worth describing before asking the question. It hopefully clarifies and motivates the question itself.
A couple of months ago, I asked the following question on Open Data (Beta) Stack Exchange.
For a research project, I am currently looking for datasets that consists of large amounts of e-mails, text messages, or OCR-processed letters that were written as part of a conversational exchange between two people.
There are quite a few open datasets that include a (significant amount) of emails. For instance, one has the Enron dataset, the Jeb Bush dataset, and the Hillary Clinton dataset. However, these datasets are not exactly what I'm looking for.
I'd like to have a dataset that consists of a large amount of (written) conversations between exactly two people. The emails should alternate between these two people, with one email (or letter or text message) for each person, each time.
Question: do such open datasets exist? If so, where can I find them?
Unfortunately, it yielded little attention, and no answers thus far. I also aimed at asking a similar question on Open Source SE, but soon learned questions of that kind were not welcome on that SE venue.
The research project in question entails analyzing written correspondence by means of techniques from the field of artificial intelligence. The goal is to identify missing messages among a chain of letters (or e-mails) that have been sent back and forth by two people. The messages may be missing because they were deleted or destroyed, either accidentally or on purpose.
It turns out that finding appropriate datasets that were generated through the correspondence between two actual people is not easy. Such datasets often have restrictions imposed on them, either because of copyrights or because of issues pertaining to the privacy of the correspondents and the people they mention in their writings.
Recently, I therefore thought of approaching this challenge from a somewhat different angle. Instead of trying to obtain a dataset that consists of texts written by people in the real world, I will aim at obtaining and analyzing a dataset that consists of fictional letters exchanged between two people.
A natural place to look is the epistolary novel genre. Although I have looked into the wiki article on the subject, I am by no means an expert on the topic (my educational background lies in mathematics and data science). It seems to me that most of the epistolary novels consist of fictional letters written by more than two people. Moreover, it appears the genre was more in vogue in the 19th century.
This is somewhat tricky, because I strongly suspect the AI methods work best if I work with a dataset that consists of written texts that are (a) written by only two (albeit fictional) people, and (b) written in a manner that is as "modern" as possible. It also helps if the novel is written in English. Each letter or email should be a reply to the previous one. This is where my question comes from.
Question: What is the most recent epistolary novel written in English that consists entirely out of back and forth letters or emails (written correspondence) between two characters - if such novels still exist?
In case such novels aren't written anymore nowadays, I'd like to hear the most recent example of such a novel. If they are still being written, I am interested in a recent example - preferably the most recent one before this question was originally posted (July 3, 2023).
