Aspect Extraction on fresh Laptop or Restaurant Reviews

lixin4ever / BERT-E2E-ABSA

[EMNLP 2019 Workshop] Exploiting BERT for End-to-End Aspect-based Sentiment Analysis

https://arxiv.org/abs/1910.00883

Apache License 2.0

397 stars 89 forks source link

Aspect Extraction on fresh Laptop or Restaurant Reviews #9

Open savvyyy opened 4 years ago

savvyyy commented 4 years ago

Hi @lixin4ever Does this model supports the functionality to EXTRACT ASPECT TERMS over fresh Restaurant or Laptop Review ? If Yes, Please guide through this.

lixin4ever commented 4 years ago

Sorry for the late reply.

Since Aspect Term Extraction (ATE) is a natural sequence tagging tasks, you can simply change this function to adapt this E2E-ABSA project for the ATE tasks. Note that there is no span label (e.g., POS, NEG and NEU ) in the ATE task and set boundary tag should be enough. Considering you adopt BIO tagging scheme, then the function get_labels should return [''O', 'EQ', B', 'I'] as the valid tag set.