Enhancing SPARQL Query Generation for Knowledge Base Question Answering Systems by Learning to Correct Triplets

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 

This paper presents a model named TSET (Triplet Structure Enhanced T5) to enhance SPARQL query generation for Knowledge Base Question Answering (KBQA) Systems. The statement problem is clear. The introduction explains the problem statement to be covered and issues to be addressed in the field of semantic web. The related work section presents a State-of-the-Art, however, the majority of the references are proceedings. An Evaluation is presented where the main contributions and results are presented.  The conclusions section is good, and the authors have properly commented on all outcomes

 Some comments are suggested to improve the quality of the paper:

·         The description (title) of Figure 1 is very long.

·     The authors need to justify why LC-QuAD,2.0 , QALD-9 plus , and QALD-10 were used for the experiment Why not other Linked Open Datasets?

·         It is necessary to present some examples of the SPARQL statements to justify the results of Table 1.

·         Likewise, the authors need to present some examples of the SPARQL statements carried out on QALD-9 plus and QALD-10 to justify the results of Table 2.

·        The quality of the references must be improved and updated with high-quality references, preferably in JCR-indexed journals.

 

 

Comments on the Quality of English Language

Overall, the paper is well written.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The paper introduces TSET (Triplet Structure Enhanced T5), a model 5 with a novel pretraining stage positioned between the initial T5 pretraining and the fine-tuning for the 6 Text-to-SPARQL tasks.  Experimental 12 results demonstrate that our proposed TSET outperforms existing methods on three well-established 13 KBQA datasets. The paper is well-written; however, I have the following comments;

1.     The abstract may quantitatively describe the achieved performance improvement. 

2.     It is better to explicitly present experimental setting. It is not sufficient to just write We align the experimental 295 settings with [8,41] line 295.

3.     Are the datasets used are Publically available? If yes authors may provide the link to access datasets? If how they get these datasets?

4.     The authors may justify why they use this evaluation metric, considering the fact that several existing works do not consider QM (Table 1) as performance indicator.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf