README for Hungarian Facebook political comments corpus
http://corpus.nytud.hu/trendminer/

Release date:
2014-11-10

Contact:
Márton Miháltz <mmihaltz@gmail.com>
RILMTA

Files:

comments_annotations.tsv.tgz: 
Contains all 1.9M fb comments + NLP annotations.
Contains 3 directories of files, each file contains about 500 comments. 
Each comment is headed by a line: #START_<fb_page_id>_<fb_post_id>_<fb_comment_id>_<comment_author_fb_user_id>
Succeeding lines are tab-separated columns describing the tokens of the comments: <token> <token_lemma> <PoS-tag> <morphological_analysis_string>
Blank lines denote sentence boundaries (last blank line means end of that comment).

comments_corpus.tgz
Contains 1 tsv file, which contains 20.9M annotation units over the comments.
Columns: <fb_page_id> <fb_post_id> <fb_comment_id> <sentence_index> <start_token_index> <end_token_index> <tag>
The first 3 columns reference ids in the comments file above (3-tuples of page, post and comment ids uniquely identify comments in the corpus).
Sentence and token indices start from 0.