The- C- M- U- Arabic-to-- English- Statistical- M- T- System课件.pptVIP

下载本文档

3
0
约6.69千字
约 17页
2019-05-17 发布于湖北
举报
版权申诉

The- C- M- U- Arabic-to-- English- Statistical- M- T- System课件.ppt

1、原创力文档（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。。
2、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击这里二次下载。
3、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。
4、该文档为VIP文档，如果想要下载，成为VIP会员后，下载免费。
5、成为VIP后，下载本文档将扣除1次下载权益。下载后，不支持退款、换文档。如有疑问请联系我们。
6、成为VIP后，您将拥有八大权益，权益包括：VIP文档下载权益、阅读免打扰、文档格式转换、高级专利检索、专属身份标志、高级客服、多端互通、版权登记。
7、VIP文档为合作方或网友上传，每下载1次，网站将根据用户上传文档的质量评分、类型等，对文档贡献者给予高额补贴、流量扶持。如果你也想贡献VIP文档。上传文档

The CMU Arabic-to-English Statistical MT System Alicia Tribble, Stephan Vogel Language Technologies Institute Carnegie Mellon University * The Data For translation model: UN corpus: 80 million words UN Ummah Some smaller news corpora For LM English side from bilingual corpus: Language model should have seen the words generated by the translation model Additional data from Xinhua news General preprocessing and cleaning Separate punctuation mark Remove sentence pairs with large length mismatch Remove sentences which have too many non-words (numbers, special characters) * The System Alignment models: IBM1 and HMM, trained in both directions Phrase extraction From Viterbi path of HMM alignment Integrated Segmentation and Alignment Decoder Essentially left to right over source sentence Build translation lattice with partial translations Find best path, allowing for local reordering Sentence length model Pruning: remove low-scoring hypotheses * Some Results Two test sets: DevTest 203 sentences, May2003 Baseline: monotone decoding RO: word reordering SL: sentence length model DevTest DevTest May 2003 NIST Bleu4 NIST Baseline 8.59 0.385 8.95 RO 9.02 0.441 9.26 RO + SL 9.24 0.455 ? * Questions What’s specific to Arabic Encoding Named Entities Syntax and Morphology What’s needed to get further improvements * What’s Specific to Arabic Specific to Arabic Right to left not really an issue, as this is only displayText in file is left to right Problem in UN corpus: numbers (Latin characters) sometimes in the wrong direction, eg. 1997 - 7991 Data not in vocalized form Vocalization not really studied Ambiguity can be handled by statistical systems * Encoding and Vocalization Encoding Different encodings: Unicode, UTF-8, CP-1256, romanized formsnot too bad, definitely not as bad as Hindi;-) Needed to convert, e.g. training and testing data in different encodings Not all conversion are loss-less Used romanized form for processing Converted all data using ‘Darwish’ tr