Active Learning to Remove Source Instances for Domain Adaptation for Word Sense Disambiguation.
Hiroyuki ShinnouYoshiyuki OnoderaMinoru SasakiKanako KomiyaPublished in: PACLING (2015)
Keyphrases
- word sense disambiguation
- domain adaptation
- active learning
- target domain
- transfer learning
- labeled data
- multiple sources
- semi supervised
- unlabeled data
- wordnet
- semi supervised learning
- natural language processing
- cross domain
- information extraction
- machine translation
- supervised learning
- machine learning
- part of speech
- co occurrence
- learning tasks
- co training
- test data
- training data
- text classification
- feature set
- relevance feedback
- training set
- sentiment classification
- document classification
- knowledge transfer
- label information
- semantic similarity
- semantic information
- training samples
- knowledge representation
- feature space
- feature extraction
- decision trees
- learning algorithm