Referent-specific sentiment strength for Chinese social-media text
SEAM computes a continuous sentiment-strength value for a given post and a referent you specify. Evaluations of different referents within the same post are separated, and every value is traceable to the original text.
The current SEAM research service is configured for Chinese-language text.
It is not a general sentiment classifier and does not output a polarity label.
The implementation is proprietary and is not distributed. Entity extraction and alias normalization are not exposed by this service.
The hosted instance applies the event-specific sentiment lexicon built for the study and is intended for texts from the same event or comparable domain.
Access by arrangement
Online computation has not yet been verified.