Phrase-Based Statistical Machine Translation for MANOS System

Phrase-Based Statistical Machine Translation for MANOS System Bo XU, Z.B Chen., W WEI W PAN and Z.D.YANG National Lab of Pattern Recognition,Institute of Automation, CAS P.O.Box 2728, Beijing China, xubo@nlpr.ia.ac.cn Abstract MANOS (Multilingual Application Network for Olympic Services) project. aims to provide intelligent multilingual information services in 2008 Olympic Games. By narrowing down the general language technology, this paper gives an overview of our new work on Phrase-Based Statistical Machine Translation (PBT) under the framework of the MANOS. Starting with the construction of large scale Chinese-English corpus (sentence aligned) and introduction four methods to extract phrases, The promising results from PBT systems lead us to confidences for constructing a high-quality translation system and harmoniously integrate it into MANOS platform. Keywords: Statistical Machine Translation (MT), Speech and Language Processing, Multilingual information processing, 1 Introduction Beijing has won the right to stage the Games of the XXIX Olympiad in year 2008. With the promise of highlighting the theme of "green, high-tech and people's Olympiad", a "High-tech Olympics Action Plan" was released to accelerate its scientific research efforts in a move to provide the world the "best Olympiad featuring the latest technologies''. As one of the key projects in “Digital Olympiad”, Multilingual Intelligent Information Service Network System- MANOS is to realize convenient information service related to the Olympics at any time, in any place and with facilities of various kinds by porting state-of-the-art available speech and language processing technologies. This paper gives a survey of MANOS in Section 2. Section 3 describes the key technology in MANOS- --Phrase-Based SMT. Finally; we aim to harmoniously port SMT system into MANOS in Section 4. 2 Outline of MANOS system MANOS system has been sketched as an Olympic- oriented intelligent multilingual information network service center in which some advanced multilingual speech and language processing measures, such as speech interactive, machine translation, information retrieval & management, dialog management, and so on, could be integrated. Figure 1 shows a concept structure of MANOS system. Bearing in mind the usability of state-of- the-art spoken language processing Figure 1 A concept Framework of Olympic-oriented Multilingual Network Information Service System technologies, MANOS system is located on the Olympic-related information inquiry service based on three different channels, such as wire or wireless telephone, voice-enabled multilingual information kiosk, and voice-enabled wearable terminal (PDA). The scope of information services covers the traffic, i - 23 hotel reservation, ticket booking, medical, shopping, weather forecast, Games program, travel sites and so on. Multilingual service, as one of add-value service, are designed to cross various service if the vocabulary and scenery is feasible for current technology. As the entrance of multilingual information service, speech interaction, machine translation and speech translation (ST) all could find their positions for different purposes. Especially in our designed Conceptual Olympic Terminal speech translation plays as possible practical application in whole MANOS system, which include three components: 1) Automatic speech recognition (ASR); 2) Machine Translation (MT); 3) Text to speech (TTS) technology. With long-time accumulation on speech interactive technology, the project team have …

Content truncated to 500 words