sjebbara / clwe-ote

Improving Opinion-Target Extraction with Character-Level Word Embeddings
11 stars 2 forks source link

Improving Opinion-Traget Extraction with Character-Level Word Embeddings

Research code for the paper "Improving Opinion-Traget Extraction with Character-Level Word Embeddings", to be published at the workshop on "Subword and Character LEvel Models in NLP" at the EMNLP 2017 (https://sites.google.com/view/sclem2017/home)


Abstract

Fine-grained sentiment analysis is receiving increasing attention in recent years. Extracting opinion target expressions (OTE) in reviews is often an important step in fine-grained, aspect-based sentiment analysis. Retrieving this information from user-generated text, however, can be difficult. Customer reviews, for instance, are prone to contain misspelled words and are difficult to process due to their domain-specific language. In this work, we investigate whether character-level models can improve the performance for the identification of opinion target expressions. We integrate information about the character structure of a word into a sequence labeling system using character-level word embeddings and show their positive impact on the system's performance. Specifically, we obtain an increase by 3.3 points F1-score with respect to our baseline model. In further experiments, we reveal encoded character patterns of the learned embeddings and give a nuanced view of the performance differences of both models.

Paper

The Paper can be found here: https://pub.uni-bielefeld.de/publication/2913711 and https://arxiv.org/abs/1709.06317


Bibtex:

@InProceedings{jebbara-cimiano:2017:SCLeM,
  author    = {Jebbara, Soufian  and  Cimiano, Philipp},
  title     = {Improving Opinion-Target Extraction with Character-Level Word Embeddings},
  booktitle = {Proceedings of the First Workshop on Subword and Character Level Models in NLP},
  month     = {September},
  year      = {2017},
  address   = {Copenhagen, Denmark},
  publisher = {Association for Computational Linguistics},
  pages     = {159--167},
  url       = {http://www.aclweb.org/anthology/W17-4124},
  doi       = {10.18653/v1/W17-4124}
}

The Code

The code is still in rough shape. I try to clean it soon. If you have any qeustion, send me a message on Github or an email.

Dependencies