**3.4 Visualization based on the severity**

374 New Research on Knowledge Management Models and Methods

steps above should be repeated and the periods are saved in Period0,1,2,,,,n. **Step 3.** To calculate the variance (standard deviation) of a histogram, it can be defined how

*Variance*

Then, the value in Variance is used for weighing of the target paper.

(1)Each Paper have a unique scores. and each cited have a unique score,too.

・ When X is quoted by other papers, X obtains scores, Variance1,,,,Variancen,

In other words, assuming that the total score of "out going" quotation for paper should be equal to the total score of "in coming", the total score should be considered to be the base of the paper. Then, paper should be considered to be more important as the score becomes higher. Thus, We intend to identify major papers in each field by applying the value in Variance to calculation of the "in coming" score for paper. In the conventional algorithm, if there are more than one "in coming" quotations, the score of each quotation was thought to be equal. On the contrary, in this research a quotation with a larger value in Variance should be thought to have a larger score. In this algorithm, the importance of paper can be

・ When X quoted other papers, X provides scores to others,O1,,,,Om.

The PageRank algorithm [7] is a technique to calculate quantitatively which page is most important if there are cross reference relations among pages such as the hyperlink structure. In this study, We calculate the importance of papers by utilizing this algorithm. The

saved in Variance.

calculation follows the steps below:

(2) Assuming there is paper (X);

*Variance1+…+ Variancen=P* 

・ The score of X is P,

**3.3 Calculation of the severity of each cited papers** 

Where, the statement below is assumed to be true:

calculated with the factor of quoted years reflected.

In addition, if there are two or more peaks in the histogram as shown in Fig. 3., the

long paper has been quoted by examining the variance (standard deviation) of the release years of the papers that referred to the target paper. The following shows a common way to obtain standard deviation; the obtained standard deviation is

<sup>2</sup> ( ) *x x*

*n*

Again, if the histogram is not normally-distributed (it has two or more peaks) as shown in Fig. 3., the variances (standard deviations) of different periods (Period0,1,,,n) is calculated and the average of them is to be saved in Variance.

*Period := (LastYear + 1) — StartYear* (2)

(3)

Fig. 4. shows a visualization of the quotation network based on the importance obtained in the previous section. Each nodes has the title of paper displayed. As described in the previous section, a node with a larger importance is displayed as a larger node. In addition, this tool is called SciHi (**Sci**ence **Hi**ghangle).

Fig. 4. Example of Academic Landscape based on weight of each cited papers
