英文论文审稿意见汇总（3）

武夷山

2011年7月8 日

1.      Since one method divides the journals evenly into 4 parts, while the other method identifies 4 groups according to certain clustering approach, then it is too natural that the two results would not be the same.

2.      As a result, it seems that the authors could focus their efforts to another direction: to argue  why their fuzzy classification is a better approach for ranking journals. For instance, they may ask a panel of experts to rank the journals in Artificial Intelligence independently and then compare their results with the result of fuzzy classification and that of quartile ranking.

3.      The next question is why should we use these 7 indicators? The more indicators, the better? Should we use indicators with high correlation among them, or use indicators which are relatively independent of each other? For instance, I guess that if we delete Scimago indicator from 7 indicators, the resulting classification would be quite different; but if we get rid of 5-year IF, the result won’t change much.

4.      One point that is worth discussion is the possible impact upon the journal ranking introduced by the original classification of journals by Thomson Reuters. In fact, their journal classification is based on some kind of clustering already, which will definitely impact the “destiny” of some individual journal at the boundary—if it happens to be attributed to another category, its quartile position may jump onto a higher one. However, if Thomson Reuters journal classification were based on, say, the Decimal Classification System, then we would see a different quartile ranking list and maybe in this case, the quartile ranking result and the fuzzy classification result will be closer to each other.