The research teams led by Vijay Erramilli from Telefonica Research in Spain create a derivative Twitter named Social Network Write Generator (SONG) to conduct research on human interaction.
To generate the data, SONG walk with the help of 16 Pentium computers connected to each other. Computer 16 is then worked simulate the network in ‘twinning’ Twitter is.
Reported by New Scientist and quoted on Wednesday (02/09/2011), in his research, the research team collected 12 million tweets written 2.4 million people on 25 November to 4 December 2008.
“After analyzing them, it was found that the level of tweet generally formed during the day, then going down at night. But the emergence of this tweet too often up and down in ways that can be predicted from a matter of seconds to hours,” said Erramilli.
In this experiment also found that by gradually increasing the number of tweets per second, the CPU is overloaded because it causes intermittent network at 100 tweets per second.
‘Twitter’ is then the total dead at 150 tweets per second. So you can imagine how many units of data processing required for the site for Twitter.
Erramilli mention, SONG can be used as a model for the analysis of Twitter. The reason, this mini version of the Twitter model easier than examining research using original Twitter tweets that contain activity within a very massive scale.
“The results of analysis can be used to help the company advertisers, website maker, and other businesses that intend to use Twitter,” lid Erramilli.