Sporleder, C. and A. Lascarides [2008] Using Automatically Labelled Examples to Classify Rhetorical Relations: An Assessment, Natural Language Engineering, 14(3), pp369--416.
Being able to identify the rhetorical relations, such as contrast or explanation, that hold between spans of text is important for many natural language processing (NLP) applications. Using machine learning to obtain a classifier which can distinguish between different relations typically depends on the availability of manually labelled training data, which is very time-consuming to create. However, rhetorical relations are sometimes lexically marked, i.e., signalled by discourse markers (e.g., because, but, consequently etc.), and it has been suggested (Marcu and Echihabi, 2002) that the presence of these cues in some examples can be exploited to label them automatically with the corresponding relation. The discourse markers are then removed and the automatically labelled data are used to train a classifier to determine relations even when no discourse marker is present (based on other linguistic cues such as word co-occurrences).
In this paper, we investigate empirically how feasible this approach is. In particular, we test whether automatically labelled, lexically marked examples are really suitable training material for classifiers that are then applied to unmarked examples (i.e., examples which naturally occur without a discourse marker). We also explore how training on automatically labelled examples compares to training on manually labelled, unmarked examples. Our results suggest that automatically labelled data are of very limited use for classifying rhetorical relations in unmarked examples.
@article{sporleder:lascarides:2008,
author = {Caroline Sporleder and Alex Lascarides},
year = {2008},
title = {Using Automatically Labelled Examples to Classify Rhetorical
Relations: A Critical Assessment},
journal = {Natural Language Engineering},
volume = {14},
number = {3},
pages = {369--416},
publisher = {Cambridge University Press}
}