The task of Fine-grained Entity Type Classification (FETC) consists of assigning types from a hierarchy to entity mentions in text. The state-of-the-art relies on distant supervision and is susceptible to noisy labels that can be out-of-context or overly-specific relative to the training example. Previous methods that attempt to address this issue do so with hand- crafted features. Also, previous work solve FETC a multi-label classification followed by post-processing. Instead, we propose an end-to-end solution with a neural network model that uses a variant of cross-entropy loss function to handle out-of-context labels, and hierarchical loss normalization to cope with overly-specific ones. We show experimentally that our approach is robust against noise and consistently outperforms the state-of-the-art on established benchmarks for the task.
View on arXiv