Twitter might not be having the best of time, but still, it is a platform many can’t do without. The platform not only serves as a source of news, but it is also a great help due to its ability to predict serious incidents long before the police can detect them, claim researchers at Cardiff University, who analyzed Twitter data from before the 2011 London riots.
Analyzing Twitter data from the London riots
The victims of the London riots were witnesses to various crimes, such as windows being broken, vehicles being set on fire and other potentially dangerous crimes. Thus, the analysis of the London riots data led researchers to conclude that computer systems were capable of automatically scanning through Twitter data and identifying a possibility of such incidents taking place even before the Metropolitan Police Service got a clue about them.
Twitter is a platform where rumors and information spread pretty fast; thus, the system was able to detect the location where riots were most likely to take place and offer real-time information about the areas where people were gathering. The researchers proved that it was possible for their system to pick up information from Twitter data regarding the disorder approximately one hour and 23 minutes earlier than the first reports made to the police.
Blend of machine learning and Twitter data
The findings of the study were published last month in a report entitled “Can We Predict a Riot? Disruptive Event Detection Using Twitter.” Researchers Nasser Alsaedi, Pete Burnap and Omer Rana used a machine learning algorithm across five steps for the study: data collection, pre-processing, classification, online clustering and summarization.
They were also able to derive contextual information with the help of machine learning owing to Twitter’s underlying features, such as time and timeframes for related tweets and the context of the text itself. All them were then applied to the detection systems and the tweets posted during the 2011 riots. The researchers concluded that in some cases, Twitter was able to detect crimes in a timely manner.
According to Burnap, to better understand online deviance, including antagonistic narratives and cyber hate, they used a combination of machine learning and natural language processing on a blend of Twitter data.
Massive potential for such systems
Burnap noted that slowly and gradually, social media and Internet-based communications are becoming more ingrained in our daily life, thus making it possible for systems similar to the one created by Cardiff University to drastically bring down crime rates due to their ability to detect crimes before the police do.
“This research could augment existing intelligence gathering and draw on new technologies to support more established policing methods,” Burnap said.
Detecting events in real-time can be very beneficial; hence, scientists are continually scrutinizing data from various popular social networks, such as YouTube, Facebook and Twitter. In all, there are about 2.5 billion non-unique users on social networking sites, and according to Hindustan Times, the data they produce have been used to predict the outcomes of elections, movie revenues and the epicenter of earthquakes.
