Darja Lescovec – Techcrunch Stories on Twitter

I examined how news from TechCrunch is shared on Twitter. The aim of this project was to understand what kinds of news are most frequently shared by people interested in technology and the internet.

The data was scraped from Twitter using Import from Twitter Search Network feature in NodeXL. I searched for tweets containing a phrase ‘techcrunch’ and a URL pointing to the TechCrunch website. Data was collected on a half-hourly basis over three days in April 2011. 27,000 tweets and pertaining data were collected altogether. After substantial cleaning of the data a 2-mode network of Twitter users and TechCrunch stories they had shared was extracted. This network consisted of 11,472 users and 3,577 unique stories connected through 23,083 edges (a link between user and story). This implies that each story was shared by many users.

From this affiliation network a 1-mode network of the 100 most-shared stories was extracted. Here, the nodes are stories and the edge between two stories is created if they are shared by the same user. There are 3,018 such edges in the network of 100 top stories.

The most popular stories were Spotify, AppStore and TC_Giveaway (see Appendix 1 for links to the examined stories) each sharing 86 unique edges. The most strongly connected pair, however, was Spotify and Yobongo; 288 users shared them both. To analyze which stories are most commonly shared together Wakita-Tsurumi clustering algorithm was used. There were 7 distinct clusters of stories as seen in Figure 1 (the darker the edge, the larger its weight).

The biggest cluster (in red colour) connects stories talking about big tech companies such are Microsoft, Google and Apple. What the also have in common is discussing companies’ services and products. For instance, Spotify discusses changes in their free access policy, while Goo.gl and Yobongo talk about products’ new features. On the other hand, stories in the pink cluster talk about companies’ performance (Janrain, AOL, GetGlue), their financial matters (ZipCar, Instructure, SugarInc) and smart phone applications (TekTrak, VodPod, Taptu). In the purple cluster the main topics are local business (LocalFocal, Hertz) and new start-ups (TastyLAbs, Friendslist, Fenning&Parker), while the green cluster connects editor’s comments (Google_ditchesYT, Motorola_Huawei) and news about web platforms’ social features (FB_50kSites, Dailyplaces, Flock_Down).

Leave a Reply