Here's the meaning of prefixes of various parts of the Treebank:
"rev" - sentences from product reviews,
"sklad" - sentences from skladnica,
"polevaltest" - sentences used in PolEval 2017 sentiment task (poleval.pl),
"jun18" - 500 sentences gathered from web, open domain, of complex semantic and syntactic structure, often mixed sentiments, added 18th of June 2018,
"neg" - another 500 sentences gathered from web, as in jun18 but most of them negative sentiments to balance the data set


For each prefix, we have following four files:
"labels" - sentiment labels
"parents" - IDs of dependency parents of each token
"rels" - dependency relations of each token
"sentence" - space tokenized sentence
