Direct Models for Word Alignment and Machine Translation - Arabic to English
Center for Language & Speech Processing(CLSP), JHU via YouTube
Overview
Explore the advancements in word-alignment algorithms and machine translation techniques in this hour-long lecture by IBM researcher Abe Ittycheriah. Delve into the application of direct models for Arabic to English translation, examining supervised training methods and their impact on alignment accuracy. Gain insights into the evolution of machine translation systems, including modifications necessitated by improved alignments. Learn about the direct translation model and its implementation in government projects. Understand the speaker's background in natural language processing, question answering, and speech recognition. Follow the comprehensive syllabus covering topics such as statistical machine translation, maximum entropy models, feature engineering, and translation analysis.
Syllabus
Intro
Setting the stage
Statistical Machine Translation
Problems with Translation
Alignment
Maximum entropy
Transition distribution
Observation model
Features
dynamic features
segmentation
other features
Translation
Source Words
Block Types
Histogram
Example
Review
Analysis
Taught by
Center for Language & Speech Processing(CLSP), JHU