Pantip, Kasetsart’s MikeLab team up on Auto Tag service
Joint venture aims to help users find the right tag for their stories easily
The largest Thai web-board, pantip.com, has joined hands with Kasetsart University’s Massive Information and Knowledge Engineering Laboratory, or MikeLab, to launch the Auto Tag service to help its users easily find the right tag for their stories.
Pantip.com’s 35 board-rooms see 4,000-5,000 posts, 4.2 million visitors and 16 million page-views per day. It makes available 10,000 tags for its 3 million members to use to navigate readers to their stories.
Apisilp Trunganont, co-founder and chief technology officer of pantip.com, said last week that before, hundreds of stories used the wrong tag, which was not related to the stories.
“Even though we provided choices of our 10,000 tags for users to get to fit their stories, hundreds of posts went with the wrong tag.
“That wasted web-board administrators time to correct it and caused the stories to get lost among the 4,000-5,000 posts created in a single day,” he said.
Auto Tag is the latest move of the 20-year-old web-board to embed innovation into its services.
The large online content platform handling huge data needs to have a tool to get the platform to be intelligent and instantly respond to users’ needs.
Auto Tag is the fruit of Kasetsart University’s research to analyse and learn the Thai language.
Arnon Rungsawang, leader of MikeLab, said big data analytics and machine learning are the two key technologies behind Auto Tag.
It is a challenge to do this kind of research on massive data and to do it especially in Thai, a unique and difficult language.
It is also the department’s mission to commercialise research and development inventions.
“Auto Tag is a result of not only data analytics processing but also machine learning of 4,000 to 5,000 new posts created in a single day at pantip.com.
“This is very challenging because we need to handle the huge amount of data in the dynamic Thai language,” he said.
Big data analytics and machine leaning aim to analyse and learn a user’s behaviour in creating a post as well as how the creator uses a tag for his stories.
Auto Tag shortens the time for users to get the right tags for their stories. It automatically suggests relevant tags for users in a few seconds. That helps visitors to access the stories quickly.
Auto Tag was launched last Tuesday.
Apisilp insists that pantip.com is committed to introducing enhancements its web-board experience for Thais.
“Only being the best in large (user-generated) content is not enough (for us).
“We want to be the best in innovation applied to our service.
“All our moves and efforts are geared to making pantip.com the platform for good story-sharing and to getting users to learn, share and have fun on the platform,” he said.
This is the first time that pantip.com has applied machine leaning to improve its platform.
Machine learning will help automate some tasks, which will also help reduce the difficulty for users. It will also add value to the platform.
“Our mission is to be a showcase for the collaboration between business and educational research and development to bring R&D to be used in commercial applications.
“This is a win-win collaboration. Pantip.com gains from having advanced and intelligent tools to enhance services, while the university gains the ability to challenge their R&D with a real case of huge and difficult data (from pantip.com).
“Meanwhile, users benefit from having better service,” he said.
