|
--- |
|
task_categories: |
|
- text-classification |
|
language: |
|
- vi |
|
tags: |
|
- code |
|
pretty_name: Vietnamese Open-Domain Complaint Detection in E-commerce Websites |
|
size_categories: |
|
- 1K<n<10K |
|
--- |
|
# Vietnamese Open-Domain Complaint Detection in E-commerce Websites |
|
This is the official repository for the ViOCD dataset from the paper [Vietnamese Open-Domain Complaint Detection in E-commerce Websites](https://arxiv.org/pdf/2103.10069.pdf), which was accepted at the [SoMeT 2021](https://dblp.org/db/conf/somet/somet2021.html). |
|
|
|
# Citation Information |
|
The provided dataset is only used for research purposes! |
|
``` |
|
@misc{nguyen2021vietnamese, |
|
title={Vietnamese Complaint Detection on E-Commerce Websites}, |
|
author={Nhung Thi-Hong Nguyen and Phuong Phan-Dieu Ha and Luan Thanh Nguyen and Kiet Van Nguyen and Ngan Luu-Thuy Nguyen}, |
|
year={2021}, |
|
eprint={2104.11969}, |
|
archivePrefix={arXiv}, |
|
primaryClass={cs.CL} |
|
} |
|
``` |
|
|
|
## Abstract |
|
Customer product reviews play a role in improving the quality of products and services for business organizations or their brands. Complaining is an attitude that expresses dissatisfaction with an event or a product not meeting customer expectations. In this paper, we build a Open-domain Complaint Detection dataset (ViOCD), including 5,485 human-annotated reviews on four categories about product reviews on e-commerce sites. After the data collection phase, we proceed to the annotation task and achieve the inter-annotator agreement Am of 87%. Then, we present an extensive methodology for the research purposes and achieve 92.16% by F1-score for identifying complaints. With the results, in the future, we aim to build a system for open-domain complaint detection in E-commerce websites. |
|
## Dataset |
|
The ViOCD dataset is consist of 5,485 reviews on four categories about product reviews on e-commerce sites. |
|
|
|
The dataset is divided into three parts as below: |
|
1. Train set: 4.39K reviews |
|
2. Valid set: 548 reviews |
|
3. Test set: 549 reviews |
|
|
|
## Contact |
|
Please feel free to contact us by email [email protected] if you have any further information! |
