Hi,
We are a group of researchers interested in your work and have been reading your paper, NodLink. We are currently attempting to reproduce the results mentioned using the source code and datasets provided in your repository. However, we have encountered a challenge regarding the calculation of precision and recall and would like to seek your guidance on this matter.
We have summarized our concerns below:
1. Aggregation of Metrics Across Datasets
Your paper reports precision and recall for the In-Lab Arena dataset. However, we noticed that your repository contains three datasets: SimulatedWin10, SimulatedWin12, and SimulatedUbuntu, which seem to correspond to the In-Lab Arena dataset mentioned in the paper.
Could you clarify the method of aggregation used to calculate precision and recall? Specifically:
- Were true positives, false positives, true negatives, and false negatives computed separately for each of the three datasets and then summed before calculating overall precision and recall?
- Or was precision and recall calculated individually for each dataset and then averaged in some way?
2. Results Obtained Using Summation-Based Aggregation
We attempted to aggregate the metrics by summing them across all three datasets and then computing precision and recall accordingly. The total values we obtained are:
- True positives: 177
- False positives: 1131
- True negatives: 2776
- False negatives: 24
Based on these totals, we calculated precision as approximately 0.14 and recall as approximately 0.88.
Could you please confirm if this approach aligns with how precision and recall were computed in your paper?
Your assistance would be greatly appreciated, as it will help ensure accurate reproduction of your results.
Thank you very much for your time and support. We look forward to your response!
Best regards,
Abdullah
Hi,
We are a group of researchers interested in your work and have been reading your paper, NodLink. We are currently attempting to reproduce the results mentioned using the source code and datasets provided in your repository. However, we have encountered a challenge regarding the calculation of precision and recall and would like to seek your guidance on this matter.
We have summarized our concerns below:
1. Aggregation of Metrics Across Datasets
Your paper reports precision and recall for the In-Lab Arena dataset. However, we noticed that your repository contains three datasets: SimulatedWin10, SimulatedWin12, and SimulatedUbuntu, which seem to correspond to the In-Lab Arena dataset mentioned in the paper.
Could you clarify the method of aggregation used to calculate precision and recall? Specifically:
2. Results Obtained Using Summation-Based Aggregation
We attempted to aggregate the metrics by summing them across all three datasets and then computing precision and recall accordingly. The total values we obtained are:
Based on these totals, we calculated precision as approximately 0.14 and recall as approximately 0.88.
Could you please confirm if this approach aligns with how precision and recall were computed in your paper?
Your assistance would be greatly appreciated, as it will help ensure accurate reproduction of your results.
Thank you very much for your time and support. We look forward to your response!
Best regards,
Abdullah