tailieunhanh - Báo cáo khoa học: "Automatic Event Extraction with Structured Preference Modeling"

This paper presents a novel sequence labeling model based on the latent-variable semiMarkov conditional random fields for jointly extracting argument roles of events from texts. The model takes in coarse mention and type information and predicts argument roles for a given event template. This paper addresses the event extraction problem in a primarily unsupervised setting, where no labeled training instances are available. | Automatic Event Extraction with Structured Preference Modeling Wei Lu and Dan Roth University of Illinois at Urbana-Champaign luwei danr @ Abstract This paper presents a novel sequence labeling model based on the latent-variable semiMarkov conditional random fields for jointly extracting argument roles of events from texts. The model takes in coarse mention and type information and predicts argument roles for a given event template. This paper addresses the event extraction problem in a primarily unsupervised setting where no labeled training instances are available. Our key contribution is a novel learning framework called structured preference modeling PM that allows arbitrary preference to be assigned to certain structures during the learning procedure. We establish and discuss connections between this framework and other existing works. We show empirically that the structured preferences are crucial to the success of our task. Our model trained without annotated data and with a small number of structured preferences yields performance competitive to some baseline supervised approaches. 1 Introduction Automatic template-filling-based event extraction is an important and challenging task. Consider the following text span that describes an Attack event . North Korea s military may have fired a laser at a . helicopter in March a . official said Tuesday as the communist state ditched its last legal obligation to keep itself free of nuclear weapons . . . A partial event template for the Attack event is shown on the left of Figure 1. Each row shows an 835 argument for the event together with a set of its acceptable mention types where the type specifies a high-level semantic class a mention belongs to. The task is to automatically fill the template entries with texts extracted from the text span above. The correct filling of the template for this particular example is shown on the right of Figure 1. Performing such a task without any knowledge about