|go to week of Oct 26, 2014||26||27||28||29||30||31||1|
|go to week of Nov 2, 2014||2||3||4||5||6||7||8|
|go to week of Nov 9, 2014||9||10||11||12||13||14||15|
|go to week of Nov 16, 2014||16||17||18||19||20||21||22|
|go to week of Nov 23, 2014||23||24||25||26||27||28||29|
|go to week of Nov 30, 2014||30||1||2||3||4||5||6|
Aspiring writers are often told that in order to learn to write well, they need to read high quality texts. Most of us, however, find it surprisingly hard to pinpoint what aspects of well-written texts make them so good. In this talk I will present our most recent work on automatic discovery of elements of text quality with a focus on discovery of communicative goals and text specificity. Our results suggest that the quality of texts form several genres could be predicted automatically.
Our models of communicative goals rely entirely on structural features to detect sentences with similar communicative goals across a collection of texts. Our experiments show that coherent texts exhibit strong structural patterns in pairs of adjacent sentences and that text coherence can be accurately estimated from these patterns. We have applied this coherence model to academic and journalistic writing. We have found significant differences in the communicative goals expressed in great and typical writing in these domains and I will discuss some of these differences.
This is joint work with Annie Louis.
Ani Nenkova is an Assistant Professor of Computer and Information Science at the University of Pennsylvania. Her main areas of research are automatic summarization, discourse, and text quality. She obtained her PhD degree in Computer Science from Columbia University in 2006. She also spent a year and a half as a postdoctoral fellow at Stanford University before joining Penn in Fall 2007.