{"id":2920,"date":"2017-03-30T23:16:12","date_gmt":"2017-03-31T04:16:12","guid":{"rendered":"http:\/\/commons.trincoll.edu\/amst-data-driven\/?p=2920"},"modified":"2017-03-31T11:53:46","modified_gmt":"2017-03-31T16:53:46","slug":"blm-convos-clusters-do-they-exist-are-they-united","status":"publish","type":"post","link":"http:\/\/commons.trincoll.edu\/amst-data-driven\/2017\/03\/30\/blm-convos-clusters-do-they-exist-are-they-united\/","title":{"rendered":"BLM Convos & Clusters– Do They Exist? Are They United?"},"content":{"rendered":"

Gephi is network analysis tool, which visualizes people and their interactions using weighted lines (edges) and circles (nodes). With the use of math algorithms and by formatting attributes, conversation groups (clusters) are visualized using distinct colors for each cluster. To begin learning how to use Gephi, I\u00a0performed a network analysis on Les Miserables. \u00a0The width of edges\u00a0connecting two nodes (characters) corresponds to the number of interactions between\u00a0the characters. There are 77 nodes and 254 edges. When the Les Mis file was opened in Gephi, I chose to use an undirected analysis, because I was more concerned with the number of interactions and less concerned with the direction of communication between characters. Below is the initial network analysis created by Gephi, when the data (to be analyzed) is opened with Gephi.<\/p>\n

\"les<\/p>\n

Looking at the list of nodes, we see that Valjean (ID = 11) and Fantine (ID = 23) are interacting with some of the same characters, and are interacting with almost as many different characters. Below is the network analysis that is generated after running algorithms and increasing node repulsion strength.<\/p>\n

\"Screen
Les Mis Network Analysis: Repulsion Strength 1000. Repulsion strength correlates to the distance between each node.<\/figcaption><\/figure>\n

The average path length, or average degrees of separation between characters, is 2.641, and the network diameter, or maximum degrees of separation between any two characters, is 5. From this data, we can guess that many of the characters are closely connected. However, we do not know if the characters are interacting in cluster. Also, with the current format, it is difficult to determine which character is interacting the most with other characters. The following network analysis correlates node diameter with the the total number of interactions a character has.<\/p>\n

\"Screen<\/p>\n

Les Mis Network Analysis: Betweenness Centrality: min = 10 max = 200<\/p>\n

Node size was set to correspond to Betweenness Centrality. This attribute results in the nodes with the most network connections to have the largest diameter, and the nodes with the fewest network connections to have the smallest diameter. As seen in the figure above, Valjean, Myriel, and Gavroche have the most interactions with other characters. However, this network analysis makes seeing clusters and each individual node difficult.<\/p>\n

\"LesMis_NetworkAnalysis\"
Les Mis Network Analysis: Final Analysis<\/figcaption><\/figure>\n

The final network analysis, shown above, reveals that while Gavroche, Marius, Fantine, and Myriel have the most interactions with other characters, they each belong to relatively distinct conversation clusters. \u00a0Valjean, the main character, is an exception and interacts with most clusters.<\/p>\n

——————————————————————————————————————————————<\/p>\n

Next, I applied the same methods to generate a network analysis, using \u00a0Gephi, for 3,000 #BlackLivesMatter tweets that were posted on February 26, 2017, the anniversary of Trayvon Martin’s death. When I performed a text analysis (see word cloud below) for the same set of tweets, I found that “trayvonmartin” was one of the most commonly used words. Therefore, I will look to see if each cluster was tweeting about Trayvon Martin on February 26, 2017.<\/p>\n

\"BLM-Word-Cloud-45\"
#BlackLivesMatter Twitter Word Cloud– Feb. 22-28, 2017<\/figcaption><\/figure>\n

For this network analysis, I chose not to differentiate between retweets and non-retweets and to include singletons. Consequently, I expect to see larger clusters. had I chosen to differentiate between tweet type, the clusters might also be separated by tweet type. For example, suppose only two users are posting new tweets, and 200 users are retweeting the pair of users’ tweets. The authors of the original tweets may appear in a separate cluster from those who are retweeting. Please note that this is a guess, not an observation.<\/p>\n

 <\/p>\n

\"Screen<\/p>\n

From the start, we can see (figure above) eight main clusters circled around the following users: texasinafrica, crystal1johnson, blklivesmatter, acluofga, khaledgeydoun, nmaahc, delo_taylor, and robertglasper. In the text analysis (see word cloud above), I filtered out “texasinafrica”, due to its frequent use. After a quick lookup in Twitter, I was able to determine that @texasinafrica is maintained by Laura Seay, a white woman from Texas, who lived and studied communities’ responses to “state weakness” in the Democratic Republic of Congo. Seay lived in the\u00a0Democratic Republic of Congo while she performed her studies. Since Seay is a white person from Texas, I expect to see that users in her cluster also use #woke or other language related to white participation in the Black Lives Matter movement. At a first glance, I am unsure as to why @acluofga, “American Civil Liberties Union of GA”, appears to be the user through which @texasinafrica and @blklivesmatter interact. Perhaps\u00a0@texasinafrica and @blklivesmatter interact directly, but this cannot be seen in the image below. For reference, @nmaahc stands for National Museum of African American History and Culture, and @crystal1johnson is maintained by a young black woman. Both @nmaahc and @crystal1johnson appear in clusters that are separate from the other clusters.<\/p>\n

\"Screen<\/p>\n

zooming in…<\/p>\n

\"Screen<\/p>\n

Searching through the tweets that were analyzed, I found that only two different tweets included @texasinafrica:<\/p>\n

    \n
  1. @ceasarishome @realJerryFlavor @texasinafrica Wrong. Go support your racist man in the White House. #TrayvonMartin #BlackLivesMatter<\/li>\n
  2. RT @texasinafrica: Trayvon Martin would have been 22 years old &amp; graduating from college this spring. #BlackLivesMatter https:\/\/t.co\/FW34Ec\u00e2\u20ac\u00a6<\/li>\n<\/ol>\n

    Moreover, neither of these tweets were created by @texasinafrica. While the first tweet seems to infer that @texasinafrica does not support the Black Lives Matter Movement, this is likely not the case since the second tweet only occurred in the data (3000 tweets) once. On the other hand, the second tweet was occurred 235 times in the data set. Thus my hypothesis that @texasinafrica would be connected to mention of Trayvon Martin was correct. The most common tweets to include @blklivesmatter also mention Trayvon Martin:<\/p>\n

      \n
    1. RT @Blklivesmatter: #DearTrayvonsMom: We will honor your son\u00e2\u20ac\u2122s memory each and every day. #RememberingTrayvon #BlackLivesMatter https:\/\/t.c\u00e2\u20ac\u00a6<\/li>\n
    2. RT @Blklivesmatter: #RememberingTrayvon #BlackLivesMatter https:\/\/t.co\/Yx76Z0mpx5<\/a><\/li>\n
    3. RT @BLMChi: #TrayvonTaughtMe #JusticeForTrayvon #TrayvonMartin #RememberingTrayvon #BlackLivesMatter @Blklivesmatter https:\/\/t.co\/ll4QRUZIkq<\/li>\n<\/ol>\n

      Again, the most common tweet to include @ACLUofGA references the memory of Trayvon Martin:<\/p>\n

        \n
      1. RT @ACLUofGA: Today we honor #TrayvonMartin by continuing to fight for racial justice. #BlackLivesMatter<\/li>\n<\/ol>\n

        So while there are distinct cluster of users using #BlackLivesMatter, on February 26, 2017 all of the clusters were tweeting about Trayvon Martin. I believe that a network analysis of a different day might have revealed more about why the different clusters exist.<\/p>\n

         <\/p>\n","protected":false},"excerpt":{"rendered":"

        Gephi is network analysis tool, which visualizes people and their interactions using weighted lines (edges) and circles (nodes). With the use of math algorithms and by formatting attributes, conversation groups (clusters) are visualized using distinct colors for each cluster. To begin learning how to use Gephi, I\u00a0performed a network analysis on Les Miserables. \u00a0The width…<\/p>\n","protected":false},"author":1383,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[6,1],"tags":[],"_links":{"self":[{"href":"http:\/\/commons.trincoll.edu\/amst-data-driven\/wp-json\/wp\/v2\/posts\/2920"}],"collection":[{"href":"http:\/\/commons.trincoll.edu\/amst-data-driven\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/commons.trincoll.edu\/amst-data-driven\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/commons.trincoll.edu\/amst-data-driven\/wp-json\/wp\/v2\/users\/1383"}],"replies":[{"embeddable":true,"href":"http:\/\/commons.trincoll.edu\/amst-data-driven\/wp-json\/wp\/v2\/comments?post=2920"}],"version-history":[{"count":23,"href":"http:\/\/commons.trincoll.edu\/amst-data-driven\/wp-json\/wp\/v2\/posts\/2920\/revisions"}],"predecessor-version":[{"id":3068,"href":"http:\/\/commons.trincoll.edu\/amst-data-driven\/wp-json\/wp\/v2\/posts\/2920\/revisions\/3068"}],"wp:attachment":[{"href":"http:\/\/commons.trincoll.edu\/amst-data-driven\/wp-json\/wp\/v2\/media?parent=2920"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/commons.trincoll.edu\/amst-data-driven\/wp-json\/wp\/v2\/categories?post=2920"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/commons.trincoll.edu\/amst-data-driven\/wp-json\/wp\/v2\/tags?post=2920"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}