Introduction

Abstract

Chapter 1 introduces the problem of extracting information from natural language unstructured documents, which is becoming more and more relevant in our “document society”. Despite the many useful applications that the information in these documents can potentiate, it is harder and harder to obtain the wanted information. Major problems result from the fact that much of the documents are in a format non usable by humans or machines. There is the need to create ways to extract relevant information from the vast amount of natural language sources.

After this, the chapter presents, briefly, background information on Semantics, knowledge representation and Natural Language Processing, to support the presentation of the area of Information Extraction [IE, “the analysis of unstructured text in order to extract information about pre-specified types of events, entities or relationships, such as the relationship between disease and genes or disease and food items; in so doing value and insight are added to the data.” (Text mining of web-based medical content, Berlin, p 50)], its challenges, different approaches and general architecture, which is organized as a processing pipeline including domain independent components—tokenization, morphological analysis, part-of-speech tagging, syntactic parsing—and domain specific IE components—named entity recognition and co-reference resolution, relation identification, information fusion, among others.