Thank you Sayan.

Thank you Joachim.

for a textual analysis of texts in late imperial China, see  the database http://paolosantangelo.altervista.org/emotions.htm

best

paolo

thanks Paolo.

thanks Mahtab, but I am looking for more then two classes(just positive or negative), that captures the user's current mental state.

yes, states of mind are classified at levels of families, and then on specific emotions. but my database is for historical-anthropological purpose, not for experimental psychological tests

You might be interested in datasets with texts and their annotations I developed for emotional research (see "Opinion Mining and Lexical Affect Sensing").

Thesis Opinion Mining and Lexical Affect Sensing

Here

http://nlp.cs.swarthmore.edu/semeval/tasks/task14/summary.shtml

you can find a data set of newspaper titles, each one associated with a ground truth (human annotated) of six emotions. Thus, for each sentence you willl have not a single emotion but a weighted vector of emotions.

in fact, you should rely on a particular affective model.

Following this project:

https://www.researchgate.net/project/SEMO-SEMantic-EMOtion-Recognition-Model

you will find in few weeks more results and data

Hey man, take a look at my paper, it's not currently published anywhere, it will be soon, the dataset I use contains facebook posts from a bunch of fanpages; therefore, likes, shares, and the well known REACTIONS which are EMOTIONS. You could check it out in my github.

https://github.com/rgap/simbig2016-facebook-reactions/tree/master/2_paper

Check out the new TidyText R package by Silge & Robinson - Article tidytext: Text Mining and Analysis Using Tidy Data Principles in R

https://www.tidytextmining.com/