Abstract: Question answering is a complex task that requires the use of technology from a wide range of natural language processing tasks. In this presentation we will focus on the technology of AnswerFinder, the question answering system that we are developing at Macquarie University. From the engineering point of view, AnswerFinder aims to produce an environment for the development and testing of question answering technology and its deployment on specific tasks. From the theoretical point of view, AnswerFinder is an excuse to develop document retrieval systems, question classifiers, named entity recognisers, query-driven summarisers, text entailment systems, paraphrase detection systems, and answer validation systems, among other applications. To address the problem of paraphrasing, AnswerFinder represents the logical contents of sentences and questions in graphs. After a quick overview of AnswerFinder we will explore the graph notation and how these graphs are used to learn question-answering rules.