WikiSplit++: Easy Data Refinement for Split and Rephrase
This work addresses data quality issues in NLP for tasks like readability enhancement and downstream applications, but it is incremental as it refines an existing dataset rather than introducing a new method.
The paper tackled the problem of hallucinations and under-splitting in the Split and Rephrase task by creating WikiSplit++, a refined dataset that removes instances where complex sentences do not entail simpler sentences and reverses sentence order. Experimental results showed that training with WikiSplit++ improved performance over the original WikiSplit, with significant gains in the number of splits and entailment ratio, even with fewer training instances.
The task of Split and Rephrase, which splits a complex sentence into multiple simple sentences with the same meaning, improves readability and enhances the performance of downstream tasks in natural language processing (NLP). However, while Split and Rephrase can be improved using a text-to-text generation approach that applies encoder-decoder models fine-tuned with a large-scale dataset, it still suffers from hallucinations and under-splitting. To address these issues, this paper presents a simple and strong data refinement approach. Here, we create WikiSplit++ by removing instances in WikiSplit where complex sentences do not entail at least one of the simpler sentences and reversing the order of reference simple sentences. Experimental results show that training with WikiSplit++ leads to better performance than training with WikiSplit, even with fewer training instances. In particular, our approach yields significant gains in the number of splits and the entailment ratio, a proxy for measuring hallucinations.