"Low Frequency Tweeters Have More to Say!" A New Approach to Identify Importance of Tweets

2025-09-04

AI Summary

Information overload in Twitter feeds often obscures high-value tweets. To address this, we hypothesize that content from infrequent posters is inherently more important; empirical analysis confirms that users posting fewer than 10 tweets per week exhibit significantly higher average engagement rates. Leveraging this inverse correlation between posting frequency and content importance, we design six interpretable importance dimensions integrating engagement signals (retweets, likes, replies) and author-level social participation metrics. We then propose a lightweight, training-free scoring mechanism suitable for real-time feed re-ranking. Experiments demonstrate that our approach effectively enhances the visibility of high-value tweets, mitigating information overload while providing transparent, quantifiable criteria for identifying potentially influential users. The core contribution lies in the first systematic validation and operationalization of the negative relationship between posting frequency and content importance—enabling an efficient, explainable, and low-overhead ranking optimization without reliance on supervised learning.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Knowledge Representation and Reasoning: Preferences
• Search and Optimization: Algorithm Configuration

Application Category

• Web Mining and Content Analysis: Content-based information diffusion
• User Modeling, Personalization and Recommendation: Personalized content ranking and presentation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Twitter is one of the most popular social media platforms.With a large number of tweets, the activity feed of users becomes noisy, challenging to read, and most importantly tweets often get lost. We present a new approach to personalise the ranking of the tweets toward solving the problem of information overload which is achieved by analysing the relationship between the importance of tweets to the frequency at which the author tweets. The hypothesis tested is that "low-frequency tweeters have more to say", i.e. if a user who tweets infrequently actually goes to the effort of tweeting, then it is more likely to be of more importance or contain more "meaning" than a tweet by a user who tweets continuously. We propose six new measures to evaluate the importance of tweets based on the ability of the tweet to drive interaction among its readers, which is measured through metrics such as retweets, favourites, and comments, and the extent of the author's network interacting with the tweet. Our study shows that users who tweeted less than ten tweets per week were more likely to be perceived as important by their followers and have the most important messages. This identified tweet-frequency band could be used to reorder the activity feed of users and such reordering would ensure the messages of low-frequency tweeters do not get lost in the stream of tweets. This could also serve as a scoring index for Twitter users to identify users frequently tweeting important messages.

Problem

• Addressing information overload in Twitter feeds by personalizing tweet ranking
• Testing if low-frequency tweeters produce more meaningful content than frequent users
• Developing measures to evaluate tweet importance based on user engagement metrics

Innovation

• Tweet ranking based on author frequency
• Six new measures for tweet importance
• Low-frequency tweeters prioritized in feeds