Paper Title:  "Non-homogeneous Generalization in Privacy Preserving Data Publishing" by  Wai Kit Wong,  Nikos Mamoulis,  David Wai Lok Cheung.
Primary Reviewer: Jong Wook Kim

1. Machine used in the RWE evaluation
   - Operating System (OS): Window XP
   - Class of hardware (HW): Desktop
   - CPU type: Intel Core2 Quad CPU Q6600 @ 2.40GHz
   - # Cores per CPU: 4
   - # CPUs per machine: 1
   - Main memory (RAM) size per machine: 2GB
   - Storage/disk system: single 230 GB desktop disk @ 5400 RPM
   - Free disk space per machine: 182 GB
   
2. Repeatability of the experiments in the paper

2.1 Steps taken to re-run the experiments in the paper

To test repeatability of the original experiments described in the accepted papers, the reviewer ran the batch files that were submitted by the authors. The batch file re-ran all experiments described in the paper, and automatically generated excel file.  During reviewing period, the authors solved all issued that arouse while running the code. The authors also provided the excel sheet that they used in the original experiments, which helped to easily understand the experiment results.

2.2	Repeatability Results
All original experiments in the accepted paper were repeated by the reviewer. 

3. Workability 

3.1 Steps taken to test workability
For testing the workability of experiment with different parameters, based on the Table 6 in the paper and the instruction by the authors, the parameters were varied such as

- Size of table n: 100K, 200K, 300K, 400K, and 500K,
- Dimensionality d: 2, 3, 4, 5, 6, and 7
- Privacy parameter k: 10, 20, 40, 60, 80, and 100

Note that the original experiments run only default cases (n=100k, d=3, k=10). The batch files used for repeatability test were re-used for testing workability of the algorithms in the paper with different parameters. The batch file generated the output file, and then the excel sheets were created.

3.2 Workability Results
Workability with different parameters was mostly confirmed by the reviews. Figure 9 (only time graph) and Figure 10 (only time graph), however, cannot be repeated with different parameters.

In Figure 9 (k=10, d=3) of the paper, the graph indicates that HS outperforms NH in the execution times. However, in the repeated experiment with different parameters (k=100, d=7), NH is slower than HS. Furthermore, similar results were obtained for Figure 10 (See the attachment for details).