tailieunhanh - Báo cáo khoa học: "A Comparative Study on Reordering Constraints in Statistical Machine Translation"

In statistical machine translation, the generation of a translation hypothesis is computationally expensive. If arbitrary wordreorderings are permitted, the search problem is NP-hard. On the other hand, if we restrict the possible word-reorderings in an appropriate way, we obtain a polynomial-time search algorithm. In this paper, we compare two different reordering constraints, namely the ITG constraints and the IBM constraints. | A Comparative Study on Reordering Constraints in Statistical Machine Translation Richard Zens and Hermann Ney Chair of Computer Science VI RWTH Aachen - University of Technology zens ney @ Abstract In statistical machine translation the generation of a translation hypothesis is computationally expensive. If arbitrary wordreorderings are permitted the search problem is NP-hard. On the other hand if we restrict the possible word-reorderings in an appropriate way we obtain a polynomial-time search algorithm. In this paper we compare two different reordering constraints namely the ITG constraints and the IBM constraints. This comparison includes a theoretical discussion on the permitted number of reorderings for each of these constraints. We show a connection between the ITG constraints and the since 1870 known Schroder numbers. We evaluate these constraints on two tasks the Verbmobil task and the Canadian Hansards task. The evaluation consists of two parts First we check how many of the Viterbi alignments of the training corpus satisfy each of these constraints. Second we restrict the search to each of these constraints and compare the resulting translation hypotheses. The experiments will show that the baseline ITG constraints are not sufficient on the Canadian Hansards task. Therefore we present an extension to the ITG constraints. These extended ITG constraints increase the alignment coverage from about 87 to 96 . 1 Introduction In statistical machine translation we are given a source language French sentence fJ fl. fj . fj which is to be translated into a target language English sentence el Cl. .ei. .ei. Among all possible target language sentences we will choose the sentence with the highest probability e argmax Pr e fJ 1 el argmax PrC Pr fJ e1 2 eI e1 The decomposition into two knowledge sources in Eq. 2 is the so-called source-channel approach to statistical machine translation Brown et al. 1990 . It allows an independent modeling of target .

• Báo cáo khoa học
• A Comparative Study
• Reordering Constraints
• Statistical Machine Translation
• báo cáo khoa học
• mô hình ngôn ngữ
• xử lý ngôn ngữ tự nhiên
• BMC Musculoskeletal Disorders
• Total knee arthroplasty
• Comparative study
• Posterior cruciate retaining
• Flat surface
• Medial pivot
• Target Dependency Structures
• báo cáo ngôn ngữ
• ngôn ngữ tự nhiên
• Comparative Study on Generalization
• Semantic Roles in FrameNet
• Yuichiroh Matsubayashi
• Hypothesis Alignment and its Improvement
• Machine Translation System Combination
• A Comparative Study of Parameter Estimation Methods
• Statistical Natural Language Processing
• Jianfeng Gao
• Strategies for internationalization
• Two industries
• A Comparative study of Thai and Vietnamese
• The Electronics
• Computer parts
• Situation of Vietnam and Thailand
• Reinforcement Learning Techniques
• Dialogue Management
• scientific reports
• model language
• process natural language
• Nghiên cứu khoa học
• Báo cáo nghiên cứu khoa học
• Directness in conversations
• American english and vietnamese
• Directness in English
• VNU Journal of Science
• Foreign language in Vietnam
• Vietnamese English contrastive linguistics
• Experiential domain of language
• Important experiential domain
• Making heritage in Vietnamese and Asian contexts
• Vietnamese and Asian
• The Asian context
• The State Party and local community
• Nursery rhymes
• Systemic functional multimodal
• Siscourse analysis perspective
• Analysis perspective
• Tạp chí Khoa học
• Comparative study of interpersonal meaning
• Hillary’s and Trump’s speeches
• The electoral speeches
• Modality and Personal pronouns system
• International Journal of Computer Networks and Communications Security
• Packet sniffer – A comparative study
• Network monitoring tools
• The network data
• Characteristic behaviour and quantitative parameter
• Impact of stock market liquidity on herding behaviour
• Stock market liquidity on herding behaviour
• Comparative study of conglomerate
• Consumer goods sectors
• Stock market liquidity
• Nigerian stock market
• Average monthly returns
• Broadband internet development and economic growth
• A comparative study of two asian countries
• Broadband internet development
• Broadband Internet competitive advantag
• Investigates various factors
• Mushroom compost
• Different carriers for shelflife
• Comparative study of different carriers
• Rhizobium spp as bioinoculants
• Bioinoculants development
• Hepatitis B
• Rapid tests
• Comparative study of screening
• Different immunochromatographic methods
• Patients attending
• Synchronisation of oestrous
• Stall fed Malnad gidda cattle
• Malnad gidda
• Repeat breeder
• Middle adulthood
• Sibling relationship
• Sibling relationships during middle adulthood
• Helical tube of steam generator
• Comparative study of corrosion resistance
• Nickel base alloys as the primary materials
• The harsh environment of reactor
• Chemical composition of passive film
• Cutting parameters on surface roughness
• Cast and rra treated AA7075 A comparative study
• Cast and rra treated AA7075
• Perceived enjoyment in people’s attitude
• Accept toy and equipment renting for children
• A comparative study between people
• The service versus
• The rental service
• Access based consumption
• E readiness
• E learning readiness
• Higher agriculture education
• Agricultural universities
• Comparative study of e learning readiness
• Rheumatoid arthritis
• Comparative effectiveness
• Randomized controlled trials
• TNF α inhibitor