The Role of Interpretable Patterns in Deep Learning for Morphology
This research provides insights into the shared and distinct character patterns used by different NLP tasks, which is beneficial for researchers working on multilingual morphological processing.
This paper investigates the importance of character patterns in deep learning models for morphological analysis, lemmatization, and copy tasks across 12 languages. It uses a sequence-to-sequence model with a pattern matching encoder to identify significant subwords for output generation.
We examine the role of character patterns in three tasks: morphological analysis, lemmatization and copy. We use a modified version of the standard sequence-to-sequence model, where the encoder is a pattern matching network. Each pattern scores all possible N character long subwords (substrings) on the source side, and the highest scoring subword's score is used to initialize the decoder as well as the input to the attention mechanism. This method allows learning which subwords of the input are important for generating the output. By training the models on the same source but different target, we can compare what subwords are important for different tasks and how they relate to each other. We define a similarity metric, a generalized form of the Jaccard similarity, and assign a similarity score to each pair of the three tasks that work on the same source but may differ in target. We examine how these three tasks are related to each other in 12 languages. Our code is publicly available.