Figure 1
From: Sieve-based relation extraction of gene regulatory networks from biological literature
Architecture of the proposed sieve-based relation extraction system. The system consists of nine sieves. The first two prepare data for processing, then six sieves try to recognize events and relations, and the last sieve cleans the extracted relations. Every input document is processed sequentially by each of the sieves and at the end a list of extracted relations is returned as a result.