Multilingual Natural Language Processing Applications: From Theory to Practice

Chapter 8. Entity Detection and Tracking

Xiaoqiang Luo and Imed Zitouni

8.1. Introduction

Information extraction (IE) is the task of identifying and extracting useful textual information from natural language documents. While “usefulness” is user- and application-dependent, we often care about “who did what to whom at when and/or for what reason (why)” from the input document. Clearly, the scope of information extraction can be arbitrarily broad, and sometimes it may even require world knowledge. To make problems tractable, we focus on two subtasks in this chapter:

1. detecting mentions from a document and identifying mentions’ attributes: a mention is a text chunk identifying a physical object (e.g., a person or an organization);

2. grouping ...