G. Krupka. Sra: Description of the sra system as used for muc-6. Proceeding of Sizth Message Understanding Conference (MUC-6), 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....method we present summarizes a document by seeking the main ideas. We follow [AOGL97] in the use of Natural Language Processing techniques to go beyond the words and instead focus on terms. Co location is used to disambiguate the meaning of words that rely on context. We use the SRA s NameTagTM [Kru95] to recognize named entities and WordNet [MBF 90] to associate synonyms. Once the terms are defined a term sentence matrix is formed. The sentences are viewed now as vectors analogous to Salton s vector space model for information retrieval [BYRN99] The job of the automatic summarization ....

G. Krupka. Sra: Description of the sra system as used for muc-6. Proceeding of Sizth Message Understanding Conference (MUC-6), 1995.

....unsupervised training to learn patterns from the MUC training corpus. However, unsupervised learning can degrade in the face of sparse data; the UMass system seemed to require one more order of magnitude of training data than was available in the latest MUC. The HASTEN system, developed by SRA [7], uses a somewhat di erent example based approach: they seek to broaden coverage by allowing statistically approximate matches, a strategy that lacks a syntactic basis, and may result in overgeneration. The presented methodology has been fully implemented as a set of tools that complement our ....

George Krupka. SRA: Description of the SRA system as used for MUC-6. In Proc. Sixth Message Understanding Conf. (MUC-6), Columbia, MD, November 1995. Morgan Kaufmann.

....to provide users (domain experts) with an easy to use interface. This is simpler for the users than writing complicated patterns. It actually hides some meaningless complications from the users and allow them to create patterns based on example sentences. Another interesting method is proposed by [4]. It is a system which incrementally learns the patterns. It stores patterns which are manually created based on examples. At any time of the development, it can find a pattern to the input sentences by choosing the most similar patterns from the already acquired patterns. The latter two methods ....

George Krupka. SRA: description of the SRA System as Used for MUC-6. In Proc. Sixth Message Understanding Conf. (MUC-6), 1995.

....system architecture; information extraction This work while done while the author was at the University of Pennsylvania. 1 Introduction There are two schools of thought about developing large, robust natural language processing (NLP) systems. The first is to build a system from scratch (e.g. Kru95] developing the necessary components as part of the larger structure of the system. The second approach, which is only now becoming feasible due to the availability of various free tools, is to use existing components and solve integration problems as they arise. When developing EAGLE, we chose ....

George R. Krupka. SRA: Description of the sra system as used for muc-6. In Proceedings of the Sixth Message Understanding Conference, November 1995.

....and associated templates, and with minimal automatic syntactic generalization, may not be able to achieve sufficient coverage [10] It is possible to compensate in part for the lack of training data by providing more information about each example. In HASTEN, the system developed by SRA for MUC 6 [15], the developer builds a structural description of each example, marking the type of constituent, the constraints on the constituent, and the semantic label of each constituent. This approach was able to achieve a good level of performance on the MUC 6 task, but requires some expertise on the part ....

George Krupka. SRA: Description of the SRA system as used for MUC-6. In Proc. Sixth Message Understanding Conf. (MUC-6), Columbia, MD, November 1995. Morgan Kaufmann.

....delimit table entries (Doorenbos et al. 1997) Kushmeric et al. 1997) Unfortunately, such systems cannot handle the large proportion of text data that is in narrative form. Considerable progress has been made in natural language processing text extraction systems (Weischedel 1995) Grishman 1995) (Krupka 1995). However, NLP techniques typically expect the text to be in the form of full, grammatical sentences. What is found on the web is often a series of brief sentence fragments such as the excerpt from a National Weather Service web page in Figure 1. A new parser for web pages, Webfoot, demonstrates ....

Krupka, G. SRA: Description of the SRA System as Used for MUC-6. In Proceedings of the Sixth Message Understanding Conference, Morgan Kaufmann Publishers, 221-236, 1995.

No context found.

G. Krupka. SRA: Description of the SRA system as used for MUC-6. In Proceedings of the Sixth Message Understanding Conference (MUC-6), Columbia, MD, 1995. NIST, Morgan-Kaufmann Publishers.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC