The train/test split has been preserved for the purposes of benchmarking, but
the sentences have been shuffled from their original order. Each sentence has been parsed into many
phrases by the Stanford parser. Each phrase has a PhraseId. Each sentence has a SentenceId.
• [login to view URL] contains the phrases and their associated sentiment labels.
• [login to view URL] contains just phrases. You must assign a sentiment label to each phrase.
The sentiment labels are:
0 - negative
1 - somewhat negative
2 - neutral
3 - somewhat positive
4 - positive
Need documentation, commend code, good results.
check attached training data.
please apply only if you are a professional
$50 and bonus for good work