Estimating the Information Source under Decaying Diffusion Rates
Round 1
Reviewer 1 Report
Overall Comments:
The paper needs a check for English grammar. A discussion section is needed to discuss the overall results and any discussion around them. The paper needs to have a section to discuss the limitations and possible future studies. The conclusion is too short and can be expanded to summarize the findings and results of the study and highlight its contribution clearly at the end.Detailed Comments:
Please remove the citation from the abstract. Figure 6a needs to be modified to avoid the overlap of the legend and the graph.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 2 Report
The paper address the problem of searching the source of information spreading over a network. The topic is not only of interest for networks of electronic devices but also for social networks.
The authors consider different diffusion rates, with exponential decay and generalized exponential decay. They test the MLE estimator over trees, finding a closed-form. This result is relevant since even random graphs can be approximated by them and, in general, one can find for any connected network a maximum/minimum weighted generated tree.
Some points to be addressed:
In which examples these decay rates are good approximations? It is clear than in a majority of cases the trees generated by BFS can be a good approximation of and SI expansion. But what about these results if the spreading like DFS algorithm? Concerning the trees, it is unclear to me if they are considered as rooted trees, and how the root is defined. What about finite graphs? As time goes by, it will be more unclear the predictions to be carried on, but in the end, in real applications graphs are finite, despite they can be very large.Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
All comments have been addressed. What I meant about the finite case is about the performance of the model between the starting point of the infection and the final moment when all nodes are infected. For instance, when the performance arrives to the maximum point before decaying.