**2. Related studies**

Academic Landscape is kind of network analyses. Network analyses have long history. Network analysis is based on graph theory [1] by Leonhard Euler in the 18th century. There are a various analysis types to Network analysis. For example, "Complex Networks", "Network Optimization", "Small World Phenomenon", "Analysis of the degree distribution", "Clustering". Currently, the most important technology is "Clustering" in the Academic Landscape.

 Clustering is the method to divide data into clusters. Clustering can be a simplify the structure of the vast amounts of data by having a common feature of the each clusters. The initial clustering algorithm was focused on the central link, and the common method was disconnect the center link to the first, then disconnect the around links to the second. On the other hand, Girvan and Newman [2-4] proposed a new algorithm that uses the modularity as an Evaluation function. Modularity is focused on links that mediate between the best cluster and disconnect the link from the link-mediated higher.

Academic Landscape Based on Network Analysis

publication databases.

each target papers.

87,399.

algorithm with the weight above.

**3.1 Publication databases and search query** 

Follow the steps 1) to 3) below to weight paper:

quotation is achieved by the formula below:

MaxYear.

Considering Analysis of Variation in the Years of Lucubration Publishing 373

1. To narrow down the number of papers by keyword (query) search in the

3. To analyze the variance of release years of the papers in the list above to weight

4. To calculate the importance of each target papers by applying the page rank

This study used SCOPUS for publication databases. As a result of narrowing down the number of papers by "clustering" as query, the number of papers for this study became

**Step 1.** To extract the maximum value in the histogram, the year with the maximum

**Step 2.** To identify the quotation period, find out the maximum year by examining the

number of quotations is extracted by using the function below and save it in

release years chronologically; the year with the number of quotations exceeding 10% of the maximum year's quotations for the first time should be the start year and it is saved in StartYear. Then, the year with the number of quotations getting below 10% should be the end year and it is saved in LastYear. The period of

*MaxYear=max{y(x)|y(x):= Number of times the references in year Y }* (1)

Fig. 2. Procedure of Academic Landscape based on network analysis

2. To make a list of quotations (papers) for a target papers.

5. To visualize the result based on the importance (node & edge)

**3.2 Publication year analysis of variance and weight of each cited papers** 

Then, Bibliometrics is a method developed by Garfield and Price. Bibliometrics is a method to support the resarch activity. It does analysis of academic papers or patents, Then, It can be understand that: "What is hot research topics?", "Which a large number of cited papers?", "Which are important papers?", "Which is related to that area?", "Who is important researcher?", "Which is the important research institute?"

There are three methods of Bibliometrics analysis (Fig. 1.).

The first is the "Direct Citation". It is regarded there is a link between paper A and paper B, If paper A is cited in the Paper C. In this case, there are 2 nodes and 1 link in the network. Papers are regarded as having any link between the papers themselves has been cited, when using a"Direct Citation".

The second is the " Co-Citation". It is the method proposed by Small [5]. Then it is regarded there is a link between paper A and paper B, If paper A and paper B is cited by paper C. In this case, there are 2 nodes and 1 link in the network. Be considered that there are a links between a pairs of those papers that all papers are listed in the bibliographies.

The third is the " Bibliographic Coupling". It is the method proposed by Kessler [6]. Then it i regarded there is a link between paper D and paper E, If paper C is cited in the Paper D and paper E. In this case, there are 2 nodes and 1 link in the network. Be considered that there are links for all pairs of papers that cite.

Fig. 1. The kinds of citation analysis

It became possible to analyse the academic landscape by combining these studies. Specifically, Each phase of the "Build of citation networks", "Getting the largest connected components", "Clustering" and "Visualization" will be processed automatically. But each area cannot automatically interpret. It is currently being analysed by the experts (Fig. 2).
