Labels are not necessary: Assessing peer-review helpfulness using domain adaptation based on self-training.
Chengyuan LiuDivyang DoshiMuskaan BhargavaRuixuan ShangJialin CuiDongkuan XuEdward F. GehringerPublished in: BEA@ACL (2023)
Keyphrases
- domain adaptation
- co training
- semi supervised learning
- semi supervised
- training set
- labeled data
- training examples
- unlabeled data
- training data
- weakly labeled data
- class labels
- supervised learning
- pairwise
- multiple sources
- multi view
- text classification
- multi label
- test data
- label information
- sentiment classification
- cross domain
- transfer learning
- active learning
- document classification
- covariate shift
- target domain
- single view
- manifold alignment
- named entities
- learning algorithm
- data mining
- machine learning
- data sets
- unsupervised learning