Automatic Detection of Key Verbs in Literary Texts

Automatic detection of keywords serves as a viable technique of topic (theme) identification (modeling). We argue that some lexical features are more meaningful for content analysis than others and their discerning can save computational time. Though verbs have already shown good performance as topic modeling features based on clustering techniques, their role is much more significant in introducing new information in the text due to their prevailing position in the Rheme of the sentence. Thus, applying machine reading tools we aim to explore the ideas introduced by key verbs in the fiction by Margaret Atwood and assess the similarity among the texts based on the new information (ideas, messages), i.e., on the Rheme, rather than the Theme.