I wish to compare similarities/differences between human and software performance of sentiment analysis of a specific corpus of diplomatic letters. For the time being, I am interested in the overall letter sentiment (negative, neutral, positive) .
Hi, it appears that my question was not of interest to other researchers (6 reads, 0 answers). Anyhow, I have completed the intended study. The results will be published in a conference proceedings in June 2018. In the meantime, here is its abstract. Meliha