Why Attention? Analyzing and Remedying BiLSTM Deficiency in Modeling Cross-Context for NER
This addresses a specific deficiency in BiLSTM models for NER, offering incremental improvements for researchers and practitioners in natural language processing.
The paper tackled the limitation of BiLSTM in modeling cross-context patterns for Named Entity Recognition (NER) by showing that simple cross-structures like self-attention and Cross-BiLSTM effectively remedy this deficiency, achieving improvements of up to 8.7% on multi-token mentions in datasets such as OntoNotes 5.0 and WNUT 2017.
State-of-the-art approaches of NER have used sequence-labeling BiLSTM as a core module. This paper formally shows the limitation of BiLSTM in modeling cross-context patterns. Two types of simple cross-structures -- self-attention and Cross-BiLSTM -- are shown to effectively remedy the problem. On both OntoNotes 5.0 and WNUT 2017, clear and consistent improvements are achieved over bare-bone models, up to 8.7% on some of the multi-token mentions. In-depth analyses across several aspects of the improvements, especially the identification of multi-token mentions, are further given.