Datasets:

ArXiv:
License:
The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.
Couldn't infer the same data file format for all splits. Got {NamedSplit('train'): ('csv', {'sep': '\t'}), NamedSplit('validation'): ('json', {}), NamedSplit('test'): ('json', {})}
Error code:   FileFormatMismatchBetweenSplitsError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

advABSA

An adversarial aspect-based sentiment analysis (ABSA) benchmark, dubbed advABSA for both aspect-based sentiment classification (SC) and opinion extraction (OE).

advABSA (Adversarial ABSA Benchmark)

In response to the concerning robustness issue of ABSA, Arts is proposed, which contains datasets crafted only for adversarial evaluaiton on SC but not for OE. So we additionally craft datasets for adversarial evaluaiton on OE following their track. These gathered datasets form advABSA. That is, advABSA can be decomposed to two parts, where the first part is Arts-[domain]-SC reused from Arts and the second part is Arts-[domain]-OE newly produced by us.

stdABSA (Standard ABSA Benchmark)

In addition, we also provide stdABSA containing datasets from SemEval14 for standard evaluation, namely Sem14-[domain]-SC and Sem14-[domain]-OE. So corresponding performance drops can be measured properly.

Citation

If you find advABSA useful, please kindly star this repositary and cite our paper as follows:

@inproceedings{ma-etal-2022-aspect, 
    title = "Aspect-specific Context Modeling for Aspect-based Sentiment Analysis", 
    author = "Ma, Fang and Zhang, Chen and Zhang, Bo and Song, Dawei",
    booktitle = "NLPCC", 
    month = "sep", year = "2022", 
    address = "Guilin, China", 
    url = "https://arxiv.org/pdf/2207.08099.pdf",
}

Credits

The benchmark is mainly processed by Fang Ma.

Downloads last month
67