Hi folks,
I would like to create/have a word count library for words that differentiate mails within the job context vs. mails with a private content.
The background is that I am working in a team that will (after having permission from respondents) analyse text data from emails -- however we will never have access to the text (that is we are allowed to have some word count algorithms over the texts and will never record or see the text).
Hence the idea to analyse, how much of the mail communication of one person has a job relevance or private relevance.
Either this would be possible by an already existing library of words occurring in job-emails or by creating one. But for the latter it would be fruitful to have some corpus of actual emails texts (labeled or not).
Perhaps such databases exist somewhere....
Best,
Holger
Hi Joachim Pimiskern
thanks a lot. Unfortunately, our respondents will not stem from on organization--it's a sample from the overall population. So all the exciting network stuff is not possible.
My current idea is this (please comment if you /any other wants to)
1) I found the "Enron database" of 500,000 emails within Enron: https://www.cs.cmu.edu/~./enron/enron_mail_20150507.tar.gz
2) My goal is to text-mine these mails and extract the most common words within these mails (or n-grams)
3) Let a jury evaluate the "discriminatory value" of which word (i.e., in how far a given word is unique for being used in a job context or whether it could also very likely be used in a private mail.
I have no clue whether this makes any sense. Unfortunately, I am sure there are language issues (i.e, transporting conclusions from English mails to the German context). Otherwise there may be some German words like "sehr geehrte(r)....", "Mit freundlichen Grüßen" or "Hochachtungsvoll" that may indicate job mails. I guess we will add them to the list manually.
Best,
Holger
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Analysis
- Data Analysis
- Dataset
- Big Data