INTRODUCTION

As a fan of Marvel Universe, I almost watched every important movie in their main storylines. However, I am curious about the rest of the heros which I have not to know maybe because they only show in the comic books or they appear rarely. Therefore, I used Gephi to visualize all the heroes and explore where does each hero belong to, how is the connection between each of them, and also the connection go across movie series.

MATERIALS, SOFTWARE, AND DATASETS

In this project, I used Gephi to visualize the data package about heroes in Marvel universal. Gephi is an open-source network analysis and visualization software package written in Java on the NetBeans platform, and it is available to display data from the individual level to the upper group level in an interactive way. Therefore, it is appropriate to analyze and visualize the big amount of heroes in all kinds of Marvel movie series.

PROCESS

In this project, Gephi was used to visualized data for three purposes: modularize hero movies, rank hero connections, and explore the relationship between degree and modularity class. The data was narrated from individual hero’s level to the higher group connection, and it tells the story about where does each hero belong to, how is the connection between each of them, and also the connection between each movie series.

-Modularity Class is used to classify heroes based on movie series, and the classification is already marked in the raw data package so that Gephi could colorize heroes differently. In this function, heroes are grouped in bunches and colorized by the movie series which we belong to. In this process, I can change the color of each bunch based on the features of different movie characteristics or the final effect of color combination.

-Degree Ranking is used to visualize the density of the connections during each hero, and Filters could simply the data by the range. In this function, audiences can find how different heros associate with other characters in the whole Marvel universal. In this process, I can use the filter to isolate the characters who are the most important character by having the strongest relationship network with others.

-MDS layout is a plugin function to cooperate two variables by dimensions. In this project, I determined Degree and Modularity Class as two dimensions. Degree represented the connection of individual hero, and Modularity Class shows heroes in different movies. Therefore, MDS layout divides them into 2 dimensions and display the cross-film collaboration of each hero.

In the first export file, I found the final image is too small because of some far distant data.

Therefore, I increase the gravity to the strongest level, but I found it is too concentrated to see the differences.

After a few times attempts, I found that gravity in 250 is the most appropriate level for this data because it concentrates the far distance data but also shows the individual differences.

RESULTS

In this section, I am going to show the final effect of Modularity Class, Degree Range, and MDS layout:

-Modularity Class

The different color modularities show different Marvel movies:

The boundaries of each modular are blurry so that I ran Prevent Overlap and DIssuade Hubs to distinguish each bunch:

I also used filter to isolate each modular to see the whole heroes in the same movie series:

-Degree Range

After I applied Ranking to Degree, I filtered data by Degree Range Settings. In this function, I can isolate the character networks based on their Edges:

Also, I can see the extreme situation by filter: the hero who has weakest connections with others and heros who have very important proportion in Marvel Universal:

-MDS layout

This result reveals how different heroes associate others across movies and the connections among each movie series:

DESIGN CRITIQUE & REFLECTION

In this project, I explored Gephi to discover the connections during each hero, the modularity of different hero, and the connections between Degree and Modularity Class. The visual results for each purpose are different, however, the exported file is not clear enough to see the details of each result. The quality of the exported file is limited by the data arrangement, such as Modularity Class, too dispersive data arrangement may cause the major part of data is too concentrated and small to read. Therefore, I think adjusting the data and design based on exported results is also an important factor which needs to be considered in the design process.