Đang chuẩn bị liên kết để tải về tài liệu:
Báo cáo khoa học: "System Demonstration of On-Demand Information Extraction"

In this paper, we will describe ODIE, the On-Demand Information Extraction system. Given a user’s query, the system will produce tables of the salient information about the topic in structured form. It produces the tables in less than one minute without any knowledge engineering by hand, i.e. pattern creation or paraphrase knowledge creation, which was the largest obstacle in traditional IE. This demonstration is based on the idea and technologies reported in (Sekine 06). A substantial speed-up over the previous system (which required about 15 minutes to analyze one year of newspaper) was achieved through a new approach to. | System Demonstration of On-Demand Information Extraction Satoshi Sekine New York University 715 Broadway 7th floor New York NY 10003 USA sekine@cs.nyu.edu Abstract In this paper we will describe ODIE the On-Demand Information Extraction system. Given a user s query the system will produce tables of the salient information about the topic in structured form. It produces the tables in less than one minute without any knowledge engineering by hand i.e. pattern creation or paraphrase knowledge creation which was the largest obstacle in traditional IE. This demonstration is based on the idea and technologies reported in Sekine 06 . A substantial speed-up over the previous system which required about 15 minutes to analyze one year of newspaper was achieved through a new approach to handling pattern candidates now less than one minute is required when using 11 years of newspaper corpus. In addition functionality was added to facilitate investigation of the extracted information. 1 Introduction The goal of information extraction IE is to extract information about events in structured form from unstructured texts. In traditional IE a great deal of knowledge for the systems must be coded by hand in advance. For example in the later MUC evaluations system developers spent one month for the knowledge engineering to customize the system to the given test topic. Improving portability is necessary to make Information Extraction technology useful for real users and we believe lead to a breakthrough for the application of the technology. 1 This work was conducted when the first author was a junior research scientist at New York University. Akira Oda 1 Toyohashi University of Technology 1-1 Hibarigaoka Tenpaku-cho Toyohashi Aichi 441-3580 Japan oda@ss.ics.tut.ac.jp Sekine Sekine 06 proposed On-demand information extraction ODIE a system which automatically identifies the most salient structures and extracts the information on the topic the user demands. This new IE paradigm becomes