Extracting Trending Keywords from OpenChat Messages
Key point
To boost user engagement on OpenChat, a technique was developed to extract trending keywords from message text.
Details
Currently, the OpenChat main screen focuses on chat room recommendations and rankings, which means users have to enter a new chat room directly to understand its content—an inconvenience that limits users' incentive to keep coming back.
To solve this, instead of showing the chat rooms themselves, a method was proposed to display the message content exchanged within chat rooms directly on the main screen. The goal is to encourage post-centric browsing, similar to a microblogging service, in order to increase user inflow and ad revenue.
However, individual messages often contain typos or incomplete sentences, making them difficult to use as standalone content. Therefore, the approach adopted is to extract key keywords from messages and group together messages on similar topics.
When defining trending keywords, simply picking high-frequency words is not enough. To exclude everyday words such as greetings or exclamations, the method extracts words whose frequency growth rate has sharply increased compared to their frequency a week earlier.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.