A Photoplethysmogram Dataset for Emotional Analysis
, Erkinov Habibilloh 1, Ye-Seul Jang 1, Taejun An 2, Donghyun Jo 2, Saron Park 1 and Won-Du Chang 1,*
Round 1
Reviewer 1 Report
General:
The authors developed the dataset of photoplethysmogram (PPG) signals while participants watched emotional video clips. Good emotion classification accuracies with the dataset were reported using various artificial neural networks.
This issue warrants investigation, and the new PPG dataset overcoming the limitation of stimulus length in the previous dataset (p. 2) will have a significant impact on the literature. Artificial neural network analysis is valid and presented clearly. The manuscript is written clearly.
However, there are some concerns in the current manuscript.
- In the Materials and Methods section, there is no evidence that video clips can evoke target valence/arousal states (p. 4). This issue is relevant because previous studies suggest that some sad films can induce high arousal experiences in participants. I recommend that the authors provide the dimensional rating data for the video clips in different participants.
- Methodology of film presentation and ratings is not presented clearly and in enough detail to allow replication. Was stimulus order counterbalanced? Were there any different effects between the first and second watch? What were the “the degree of emotional stimulus” (p. 9) ratings?
- In the Results and Discussion section, the authors frequently interpreted results without statistical evidence. I recommend that the authors thoroughly conduct statistical tests and report statistics.
Minor:
- Abbreviations (e.g., low frequency) should be defined when first mentioned and should not be switched into a written-out form.
Author Response
Please see the attached file.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors present research in the field of emotion classification based on physiological signals. Their particular interest is on the use of PPG data because there are only a few studies and datasets available in comparison to other biosignals such as ECG and EEG. The main contribution of their study is the introduction of a new public dataset of PPG recordings for the evaluation of emotion recognition algorithms.
I would accept this resubmitted manuscript because my previous comments were taken into account and the authors have enlarged the dataset of PPG recordings.
Author Response
Thank you very much for your previous comments on our manuscript. They certainly allowed us to improve our paper.
Author Response File: Author Response.pdf
Reviewer 3 Report
The paper is very fluent and easy to read.
I have found only few small mistakes:
1. "and high-frequency (HF, 0.15~0.4 Hz). 143 The frequency bands were defined as listed in Table 4" - I do not see that table
2. figure 2 is not centered in the paper
Author Response
Please see the attached file.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Great improvement has been made. The authors addressed all the issues I raised.
