Building a WSD module within an MT system to enable interactive resolution in the user’s source language
3470Authors
nan
Abstract
. Ambiguous words pose very serious problems to existing machine translation systems. The Translation Checker, a system part of Translution Central addresses this prob- lem by allowing users to disambiguate words in their own language, with little or no knowledge of the target language. In order to achieve this, a multilingual dictionary is being developed using EuroWordNet. Languages are too ambiguous to feasibly present users with all the senses available for a word. To this end, a suite of language processing modules has been developed to reduce the ambiguity of words. The implemented modules and an evaluation of their influence on English, French, German, Italian and Spanish corpora are presented. The results of the evaluation show that the proposed approach dramatically re- duces the ambiguity of the language. 1.
Keywords
Full Content
EAMT 2005 Conference Proceedings 205 Building a WSD Module within an MT system to enable interactive resolution in the user’s source language Constantin Orasan*, Ted Marshall+, Robert Clark+, Le An Ha*, Ruslan Mitkov* * Research Group in Computational Linguistics, University of Wolverhampton, UK + Translution, UK C.Orasan@wlv.ac.uk, Ted.Marshall@translution.com, Rober.Clark@translution.com, L.A.Ha@wlv.ac.uk, R.Mitkov@wlv.ac.uk Abstract. Ambiguous words pose very serious problems to existing machine translation systems. The Translation Checker, a system part of Translution Central addresses this prob- lem by allowing users to disambiguate words in their own language, with little or no knowledge of the target language. In order to achieve this, a multilingual dictionary is being developed using EuroWordNet. Languages are too ambiguous to feasibly present users with all the senses available for a word. To this end, a suite of language processing modules has been developed to reduce the ambiguity of words. The implemented modules and an evaluation of their influence on English, French, German, Italian and Spanish corpora are presented. The results of the evaluation show that the proposed approach dramatically re- duces the ambiguity of the language. 1. Introduction Ambiguous words pose a very serious problem to existing machine translation systems because in many situations the translation engines do not know how to handle these words. This problem can be particularly serious for organisations which heavily rely on machine translation for their everyday operation. The work presented in this paper is part of a larger project to develop technologies which will enable people and or- ganisations to improve communications by re- moving language barriers.1 The technology is based on automatically redirecting e-mails, web pages and electronic documents to a centrally based translation facility termed Translution Central. Users simply write emails in their own language in the normal way, press the Send but- ton, and the recipients will receive the email automatically translated into their own lan- guage. Similarly, incoming emails will be trans- lated into the user’s own language. The initial release of the product will sup- port five European languages: English, French, 1 This project was initiated by Translution in 2002. German, Spanish and Italian. Translution has developed three different product suites, aimed at different sectors of the market, Translution Light, Translution Pro and Translution Corpo- rate2. This paper presents the Translation Checker, a tool integrated in the Translation Central which enables users to specify the meaning of polyse- mous words without the need of knowing how they translate in the target language or defining the domain of the source document. The paper is structured as follows: Section 2 discusses the structure of the translation checker. An evalua- tion of the system is presented in Section 3. The paper finishes with conclusions. 3 2 More information about these products can be found at http://www.translution.com 3 This is a collaborative work between Trans- lution and the University of Wolverhampton the main objective being to develop tools which allow users to improve Machine Translation quality designed with the non-linguist in mind. Orasan, et. al. 206 EAMT 2005 Conference Proceedings …
Content truncated to 500 words
Labeled Text
. Ambiguous words pose very serious problems to existing machine translation systems. The Translation Checker, a system part of Translution Central addresses this problem by allowing users to disambiguate words in their own language, with little or no knowledge of the target language. In order to achieve this, a multilingual dictionary is being developed using EuroWordNet. Languages are too ambiguous to feasibly present users with all the senses available for a word. To this end, a suite of language processing modules has been developed to reduce the ambiguity of words. The implemented modules and an evaluation of their influence on English, French, German, Italian and Spanish corpora are presented. The results of the evaluation show that the proposed approach dramatically reduces the ambiguity of the language. 1.
Publication Details
- Paper ID:
- 3470
- Year:
- 2005
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan