Hierarchical Character Embeddings: Learning Phonological and Semantic Representations in Languages of Logographic Origin using Recursive Neural Networks
This work addresses the challenge of improving embeddings for logographic languages, which could benefit downstream NLP tasks, but it is incremental as it builds on existing recursive neural network methods.
The paper tackled the problem of learning better embeddings for logographic characters (like Chinese) by exploiting their recursive hierarchical structures, using a treeLSTM neural network, and showed that these hierarchical embeddings outperform baseline methods in tasks such as predicting Cantonese pronunciation and language modeling, with diagnostic analysis indicating improved robustness on complex characters.
Logographs (Chinese characters) have recursive structures (i.e. hierarchies of sub-units in logographs) that contain phonological and semantic information, as developmental psychology literature suggests that native speakers leverage on the structures to learn how to read. Exploiting these structures could potentially lead to better embeddings that can benefit many downstream tasks. We propose building hierarchical logograph (character) embeddings from logograph recursive structures using treeLSTM, a recursive neural network. Using recursive neural network imposes a prior on the mapping from logographs to embeddings since the network must read in the sub-units in logographs according to the order specified by the recursive structures. Based on human behavior in language learning and reading, we hypothesize that modeling logographs' structures using recursive neural network should be beneficial. To verify this claim, we consider two tasks (1) predicting logographs' Cantonese pronunciation from logographic structures and (2) language modeling. Empirical results show that the proposed hierarchical embeddings outperform baseline approaches. Diagnostic analysis suggests that hierarchical embeddings constructed using treeLSTM is less sensitive to distractors, thus is more robust, especially on complex logographs.