Defining Comparison in a Computational Linguistics Framework

Maria Milosavljevic
CSIRO-MIS, Intelligent Interactive Technology Group

Tuesday 20 April at 11am

Abstract

The process of comparison allows us to build categorisations of things in our world, to distinguish one kind of thing from another, and to form new categorisations when presented with something unfamiliar. Moreover, comparison allows us to describe these categorisations in a way which reflects our understanding of the structure of things in the world. The understanding of concepts is promoted when the learner recognises relationships between things; this is best achieved in descriptions by using comparison.

In this talk, I will describe some different forms of comparison found in a corpus analysis of descriptions of entities. Each form of comparison reflects a different discourse purpose. For example, comparison can be employed to distinguish similar entities (eg. "This printer differs from that in that..."), or to illustrate the properties of an entity by referring to a known entity or class of entities which share(s) those properties (eg. "It's about the size of a match mox"). In addition, comparison can be used more implicitly when describing the features of an entity, and is a useful tool in other tasks such as misconception correction.

This talk will also cover the automatic generation of comparisons in descriptions of entities. It will present a broader discussion of the role of comparison in textual descriptions of entities, taking into account the possible comparators in the situational context surrounding the description. Some computational mechanisms for the automatic production of textual comparisons of entities will also be presented.

Short resume

Maria Milosavljevic is a research scientist at CSIRO Mathematical and Information Sciences in Sydney, Australia. She is a member of the Intelligent Interactive Technologies group, and works in the field of natural language generation. Maria has been employed by CSIRO for five years, and spent three years of this on study leave at the Microsoft Research Institute completing her PhD. She is applying language technology to the automatic construction of documents from a data source, including hypertext documents. Her PhD focused on the use of comparisons in descriptions of entities.

 

Back to HAIL Home Page