About Project: Specificity

General-Specific Classifier

In writing, not every sentence is fully in detail. General and ambiguous sentences add not only interest, but also a human element not yet captured by autosummarization. What makes a sentence general? What do readers want from a sentence? A specificity classifier is a first step in capturing this non-lexical aspect of writing and further mimicking human style in automated writing.

Final goal: Create and train a classifier that will accurately discern whether a sentence is general or specific.
Learning goals:
- How do we collect and interpret data from human annotators?
- How else might it be represented?How are features for classification selected?
- How is vagueness expressed?
- What is the relationship between ambiguous information that can be found in context and that is deliberately withheld?
- How is content analysis performed?