Методология создания шаблонов для русского языка в knowledge-based системах IE

Rules developed by experts are widely used in knowledge-based information extraction (IE) systems. The rules are productions consisting of two parts. The left-hand side of a rule is a template that matches a certain syntactico-semantic structure and the right-hand side is an action that is executed when left-hand side template is matched against a particular text fragment. A typical action is to annotate a matched text fragment with a new syntactico-semantic structure. Once the text is annotated in such a way, knowledge is extracted and stored in a particular format. Rule-based approach is described in a number of works. To our knowledge, there are no papers providing the set of rules developed for a particular domain or describing a methodology used to develop the rules themselves. In this paper we describe our methodology for building extraction rules. In particular, we describe the process of building a more complex left-hand side part (further in the paper we will refer to left-hand side as template).