Speech Function classifier is created for analysis of dialogue structure (based on the theory by Eggins&Slade). Overall,there are 32 possible labels. One label is assigned to one utterance. Moreover, there is a predictor for 5 most probable next labels
At first, preprocess a dialogue for analysing its scructure. Input: separate lists with utterances and a sequence of speakers.
import sf_classifier
sf_classifier.classify_labels(utterances,speakers)You will get a list with predicted speech functions as a result.
import predictor
predictor.print_most_probable_labels(speech_function)