Similarity between staff can be defined by degree of possessing keywords selected in common. Two staffs who share more keywords show higher similarity and seem to be in a similar research field. The position of each staff can be plotted in the two dimensional map based on the keyword similarity by means of the multi dimentional scaling (MDS) method. This multidimensional data analyzing method can visualize similarity and help to recognize the general view of a data set.
You can see a full name by focusing a mouse cursor over the each plot.
Clicking the name display a page of staff’s information.
Colors of the names shows the course a staff belongs to.
Make a matrix composed of staff’s names as rows and keywords as columns. Input value 1 into the matrix cells corresponding keywords selected by each staff. Input value 0 into the rest of the cells.
Each staff has a set of values made up of 0 and 1 in a row.
This row denotes a position of a multidimensional coordinate. For example, if there are three keywords, a row having (1, 0, 0) corresponds to the position at that the x coordinate is 1, the y is 0 and the z is 0.
Next, calculate the distance between two positions in all pairs.
Convert these distances in the multidimensional space into the distances on the two dimensional map. This process, which is called by dimensionality reduction, plots the map with keeping a correlation in the multidimensional space as much as possible.
The map was made by using the statistical package R. The distances were calculated with the dist function and the dimensionality reduction was carried out by the cmdscale function.