Publication

Path-Based Attention Neural Model for Fine-Grained Entity Typing

The Architecture of PAN

Date

    06/2017 - 09/2017

Authors

    Denghui Zhang, Manling Li, Pengshan Cai, Yantao Jia, Yuanzhuo Wang, Xueqi Cheng

Paper

Problem

    Fine-grained entity typing aims to assign entity mentions in the free text with types (e.g., person, politician, etc.), which are arranged in a hierarchical structure. Traditional distant supervision based methods employ a structured data source as a weak supervision and do not need hand-labeled data, but they neglect the label noise in the automatically labeled training corpus. Although recent studies prune wrong data ahead of training, they suffer from error propagation and bring much parameter and computation complexity.

Solution

    Proposed an end-to-end typing model, called the Path-based Attention Neural model (PAN), by leveraging the hierarchical structure of types to learn a noise-robust performance.

My Work

  • Assisted in designing PAN to dynamically reduce the weights of wrong labeled sentences for each type.
  • Specifically, propose to formulate path as semantic composition of all the types on the path by different composition operators.
  • Co-authored paper was accepted as a poster by AAAI 2018.