A comparative analysis of algorithm for detecting concept-drift
Charusheela, Lalitha · AIP conference proceedings · 2024
In these day's social media has become more popular.Many people are using social networking sites like Twitter, LinkedIn , Google+, Facebook etc., to be connect throughout the world for communicate with the families, business use or both.Among this Social Medias Twitter is one of the most popular social media people are using these days.They all collect huge amount of data day by day and serve as the source of Big Data analysis.In the Social media like Twitter there is always a drastic change in the trending topics depending on many parameters this is called concept drift.There is a mechanism or model to identify this concept drift so that we can find out what is the span of the trending tweets.In order to come out with the solution for this, we have organized one work as objectives: Collection of tweets, identifying the trending tweets, Algorithms to identify concept drift are ADWIN and Page-Hinkley test is used in this work and Comparative analysis of algorithms is done.After doing comparative analysis we got the accuracy of 83.24% and 80.06% on ADWIN algorithm and Page-Hinkley test algorithm respectively on our twitter dataset. INTRODUCTOINThe social media sites creation was done from the 1997, from that time the popularity and the number of users in social media has been exponentially increasing.Today, in 21st century, according to the report generated on current number of active users there are almost 3.96 billion people are the social media users across the world.There are almost 58.11% of social media users who are eligible audience aged above 13 years in the world's population.With this huge number we can very well imagine the massive amount of data being shared online in the form of images, texts, audios, videos and other activity logs.The variation in popularity of all networks differs with demographics and they are still growing.On multi networking across a mean of 8 social networks and messaging apps users are spending a mean of two hours and twenty-four minutes per day on social media.Just few years ago TikTok was not launched, but now it has become popular among top 10 of the foremost popular social networks [8].When targeting specific audiences, we need to understand these dissimilarities in popularity, consumer behavior and preferences are important when utilizing various social networks.To remain connected with companions, brood(family), co-workers, frequenters, or clients we use social networking which is the Internet-based social media platforms.Platforms like Twitter, Facebook, LinkedIn, Instagram, Whatsapp, and Gmail among others social networking can have a communal(social) use, a trade use, or both.Social networks have transformed marketing and made easier and in the statistics of latest global social media research summary for 2021, the popularity of social media is still growing.When comparing the most popular social networks, it is a good idea to compare them based on the number of active usage accounts rather than the number of user accounts.In correlation, we see that 50.64 percent of the world's population, or 7.82 billion people, use social media.90.71 percent of the internet's 4.17 billion users are active social media users.This suggests future growth opportunities and helps to explain why social media users increased by nearly 13% within 2020 and 2021.Twitter, among other popular Online-Social-Media (OSM) sites like Facebook, Instagram, WhatsApp, etc. is a social networking service that is used for microblogging purposes.Here, the messages are called "tweets" and the Twitter users can post and interact with these tweets.As of 2020, Twitter has about 330 million monthly active users