Resolving Co-reference Anaphora Using Semantic Constraints

The field of Artificial Intelligence strives to use the notion of human intelligence to build computer systems to solve various problems. The development and the use of language by humans is a highly intelligent trait depicted by humans for millions of years. With the advent of the internet, an overwhelming amount of written text becomes available for humans to process and there is thus an increased demand for automatic processing of text. The research for this PhD falls in the area of Natural Language Engineering, which deals with building computer systems which can be used to interact with machines using natural language rather than structured commands dictated by the vocabulary of the machine. More specifically, this PhD is about resolving words used to co-reference other words in a document, referred to as anaphora. A common form of anaphora in English is the use of pronouns and phrases such as ``the driver¿, ``the tall slim man¿ and others that contain the definite article ¿the¿. Anaphora use occurs in all natural languages and its resolution is thus crucial for text understanding. Humans display an immense ability to resolve co-referring words using a complex combinat