THE ROLE OF LINGUISTIC INFORMATION FOR SHALLOW LANGUAGE PROCESSING

Abstract. Many methods in computational linguistics rely on shallow processing to achieve their goals. The advantage of these methods in comparison to deep processing is that they do not require the building of elaborate representations of the text to be processed or to perform reasoning on this data, and as a result they can be more easily implemented. This paper shows how shallow methods for automatic summarisation can improve their performance by adding different types of linguistic information. 1