Specification Mining over Temporal Data^†

Round 1

Reviewer 1 Report

Clarity of Problem Statement: The problem you're addressing in the paper is not immediately clear. Start by providing a succinct and clear problem statement in the introduction. Explain why specification mining for temporal data is challenging and the relevance of your approach. Motivation and Importance: Provide a stronger motivation for why the problem you're addressing is important. How is the field of "Verified Artificial Intelligence" impacted by your work? What are the practical applications of specification mining over temporal data? Terminology and Notations: Clearly define any specialized terminology and notations you're using in your paper. This will help readers understand your concepts more easily. Comparison to Existing Approaches: Give a more detailed comparison of your proposed algorithm, Bolt2, with existing algorithms like Bolt and others in the field. How does Bolt2 address the limitations of previous algorithms? What are its key advantages? Algorithm Description: Provide a step-by-step breakdown of your Bolt2 algorithm. This should include the key heuristics, lattice search, and any other components that make up your approach. Diagrams or pseudo-code can greatly enhance the understanding of the algorithm. Experiments and Results: Your experiments are important for validating your approach. However, the presentation of the experimental results could be clearer. Use tables and graphs to present the performance metrics in a more organized manner. Additionally, consider providing explanations for any unexpected or interesting findings. Real-world Datasets: Explain the selection criteria and characteristics of the real-world datasets you've used. How representative are they of real-world scenarios? Address any limitations or biases that might affect the generalizability of your results. Discussion of Results: Your discussions of the experimental results need to be more detailed. For example, explain why Bolt2 consistently outperforms other algorithms in terms of mining time. Provide insights into the trade-offs between accuracy and efficiency. Future Work and Implications: Your discussion of future works is brief. Elaborate on potential directions for further research and how the limitations you've identified could be addressed. Conclusion: Summarize your findings and contributions more explicitly in the conclusion. Emphasize the significance of your approach and how it advances the field of specification mining for temporal data.

Minor

Author Response

Please see the attached document.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper proposes an algorithm to mine temporal data, namely being adequate for log data. It is very extensive and at many points not really easy to follow. In general it is coherent and well organised. The methods are adequate. The results make sense and there is a fair comparison to the state of the art. Below are more specific comments.

  - There is a note right below the affiliations informing the paper is an extended version of a previous manuscript which seems to be reference 5. Reference 5 is apparently where the authors propose Bolt, a previous version of the algorithm, which is now called Bolt2. That seems clear from the paper, but perhaps that could be clarified in more detail along the paper and the note removed from the affiliations section.
 
  - All the paper is very verbose. Some sections could possibly be summarised for succinctness and clearness. But as long as there is no page limit that is just a minor issue.
 
  - Section 1 has a very lengthy introduction and then a single subsection 1.1. It could be reorganised so that there exist at least two subsections or no subsections at all - 1.1 without 1.2 does not make sense.
 
  - The paper needs careful proof reading, to correct English but also technical mistakes. There are numerous typos, awkward sentences, mismatched capitals (including title of 4.3.4), mismatched spaces and missing symbols. Part of it may be a problem in PDF generation, but the PDF that I received has many missing symbols/words/references and sometimes there may be semantic errors. Example of Line 218 "According to [what?], there exists..."
 
  - Check table 1 - it also has formatting issues, namely one or more extra columns.
 
  - At line 321 there may be a missing section heading
 
  - There could be a reference to dataset BPIC2019, it is mentioned but no reference or link is given.
 
  - Besides the missing symbols/words, and awkward sentences, at line 1115 there is a line break and it is ambiguous which algorithm(s) the authors refer to - even though it is more or less clear from the context of the paper, that should be revised.
 

Requires careful proof reading - there are many typos and awkward sentences.

Author Response

Please refer to the attached document.

Author Response File: Author Response.pdf