A Systemic Approach to Appraisal: Identifying Opinion and Sentiment in Text


December 1, 2004


Casey Whitelaw


University of Sydney


Sentiment analysis aims to identify the subjective content of text: what kinds of opinions are being expressed, and how? This is a relatively new field with many applications in information extraction and natural language processing.

Systemic Functional Linguistics deals explicitly with how we use language to express sentiment, using a set of systems called Appraisal. This talk introduces a framework for extracting Appraisal, based explicitly on this theory and using new and existing computational techniques.

The talk will cover an introduction to SFL and Appraisal, an explanation of the computational processes involved, and present results from recent experiments in using this approach to classify movie and product reviews.


Casey Whitelaw

Casey Whitelaw is a 26-year old final year PhD student studying at the University of Sydney. His research focuses on applying aspects of SFL theory to practical large-scale text classification tasks using machine learning. During his candidature, Casey has published over ten papers in the fields of named entity recognition, sentiment analysis, and text classification. Most recently, Casey has been an invited scholar at the Illinois Institute of Technology, working with Shlomo Argamon on automated analysis of appraisal.