Reducing Confusion in Active Learning for Part-Of-Speech Tagging
This work addresses the challenge of minimizing annotation costs for building low-resource syntactic analyzers, such as POS taggers, by improving active learning strategies, though it is incremental as it builds on existing AL frameworks.
The paper tackled the problem of active learning for part-of-speech tagging by showing that existing heuristics based on uncertainty and representativeness are suboptimal, even in an oracle scenario, and proposed a new strategy focusing on reducing confusion between specific tag pairs, which outperformed other methods by a significant margin across six languages.
Active learning (AL) uses a data selection algorithm to select useful training samples to minimize annotation cost. This is now an essential tool for building low-resource syntactic analyzers such as part-of-speech (POS) taggers. Existing AL heuristics are generally designed on the principle of selecting uncertain yet representative training instances, where annotating these instances may reduce a large number of errors. However, in an empirical study across six typologically diverse languages (German, Swedish, Galician, North Sami, Persian, and Ukrainian), we found the surprising result that even in an oracle scenario where we know the true uncertainty of predictions, these current heuristics are far from optimal. Based on this analysis, we pose the problem of AL as selecting instances which maximally reduce the confusion between particular pairs of output tags. Extensive experimentation on the aforementioned languages shows that our proposed AL strategy outperforms other AL strategies by a significant margin. We also present auxiliary results demonstrating the importance of proper calibration of models, which we ensure through cross-view training, and analysis demonstrating how our proposed strategy selects examples that more closely follow the oracle data distribution.