Why cannot long-term cascade be predicted? Exploring temporal dynamics in information diffusion processes

Abstract

Predicting information cascade plays a crucial role in various applications such as advertising campaigns, emergency management and infodemic controlling. However, predicting the scale of an information cascade in the long-term could be difficult. In this study, we take Weibo, a Twitter-like online social platform, as an example, exhaustively extract predictive features from the data, and use a conventional machine learning algorithm to predict the information cascade scales. Specifically, we compare the predictive power (and the loss of it) of different categories of features in short-term and long-term prediction tasks. Among the features that describe the user following network, retweeting network, tweet content and early diffusion dynamics, we find that early diffusion dynamics are the most predictive ones in short-term prediction tasks but lose most of their predictive power in long-term tasks. In-depth analyses reveal two possible causes of such failure: the bursty nature of information diffusion and feature temporal drift over time. Our findings further enhance the comprehension of the information diffusion process and may assist in the control of such a process.

Applications of Artificial Intelligence for Information Diffusion Prediction: Regression-based Key Features Models

“…Sina Weibo is a popular social network site like Twitter. We used a large dataset from Sina Weibo [7] for our research. The dataset provides by DataCastle.…”

“…The tweets were posted between January 1, 2015 and May 20, 2016. This dataset was developed by authors in [7]. It is publicly available online.…”

“…The constructs a network (G_F) based on the relationships among users who follow each other. Five network structural features are derived from measuring user influence in this network, including out_degree_F, in_degree_F, all_degree_F, bi_degree_F, and pagerank_F, as described in Table I [7].…”

“…The builds a retweeting network (G_R) using retweet data from the first 60 minutes. Similar to the following network, by extracts five network structural features to quantify user influence in the retweeting network: out_degree_R, in_degree_R, all_degree_R, bi_degree_R, and pagerank_R, as described in Table I [7].…”

“…In this study, the authors proposed Machine learning and deep learning models to improve our forecasts using a publicly available dataset provided by authors in [7]. This dataset was collected from Weibo, a social network like Twitter.…”