Hashtags of twitter
Twitter uses proprietary algorithms to surface trending topics, which may not be represented by hashtags. Despite not divulging their methodology for identifying trending topics, Twitter has stated that key metrics for the algorithms include novelty as well as popularity (Twitter)This means that the longer a topic has been trending, the higher a bar it must clear with respect to popularity. Additionally, there is a spatial component to trending topics that is hierarchical in nature: Twitter’s awareness of many of its users’ approximate geographical locations allows it to categorize trending topics (if available) on a granular level. For example, topics that are trending from users within a city contained by a given country may be almost or completely disjoint from those trending from users countrywide.
Figure 2 shows a variety of trending topics from different locations, captured within minutes of each other on December 20th, 2012. One notices that the worldwide trends list is multilingual and otherwise differs slightly from the other lists; the United States and San Diego trends lists, on the other hand, happen to be identical. In this example, each of the three lists coincidentally contains five hashtags out of the nine entries within, and on each list the hashtags are (again coincidentally) listed before the other topics.
Compare this to this author’s personalized trends list as shown in Figure 3,automatically curated by Twitter This personalized trends list, captured in the same time frame as those in Figure 2 shows that the ordering of hashtags within a trending topic list is not a given. Additionally, the ratio of hashtags to overall trends is not constant.
Both stipulations are seen in Figure 4 : To help gauge the relationship of hashtags to trends, over the span of a few hours I ran a script that collected 488 trends from 49 locations derived from the geographic and/or political boundaries of the United States; after duplicates had been discarded, 42 hashtags were identified out of 91 trends, meaning that approximately 46% of the available trends were found to be in the form of hashtags. The extent to which this percentage is a function of Twitter’s trending algorithm is currently unknown.