Content-based Modeling and Prediction of Information Dissemination

Abstract

Social and communication networks across the world generate vast amounts of graph-like data each day. The modeling and prediction of how these communication structures evolve can be highly useful for many applications. Previous research in this area has focused largely on using past graph structure to predict future links. However, a useful observation is that many graph datasets have additional information associated with them beyond just their graph structure. In particular, communication graphs (such as email, twitter, blog graphs, etc.) have information content associated with their graph edges. In this paper we examine the link between information content and graph structure, proposing a new graph modeling approach, GC-Model, which combines both. We then apply this model to multiple real world communication graphs, demonstrating that the built models can be used effectively to predict future graph structure and information flow. On average, GC-Model's top predictions covered 19% more of the actual future graph communication structure when compared to other previously introduced algorithms, far outperforming multiple link prediction methods and several naive approaches.

Reachability Analysis and Modeling of Dynamic Event Networks

“…From the plots, we can see that as Δ increases from 10 minutes to 6 hours, more nodes gather on the right of the plot. Similar curves were found for other dynamic networks, as well (Figures 9 and 11 display additional histogram results for the Enron Email dataset [15] consisting of 87K unique email users and 360K directed email edges, as well as a crawled Twitter dataset [26] containing 8K Twitter users and 663K directed "reply to" and "retweet" edges). Again, this shift across time is largely reflected in the plots, and is echoed in the decreasing slope of the power-law curve loosely fit to the data.…”

“…In contrast, temporal graphs have been an active topic of research in only the last few years and the research has largely concentrated on dynamic "state" networks and ignored dynamic "event" networks [2,26,32].…”

“…The analysis and mining of these networks has been an active and important area of research, leading to both newly discovered fundamental network properties, as well as interesting and useful new knowledge and applications, from graph clustering for gene function discovery to network modeling for link or structure prediction [24,25,26].…”

Answering top-k representative queries on graph databases

“…Such graphs can be used to model a wide range of scientific data such as chemical compounds [20,23], system call graphs [8], communication graphs [17], social networks [7] and gene interaction networks [21], as demonstrated in Table 1. Example 1 in Table 1 formalizes the problem for molecular libraries discussed earlier.…”

Classification-Based Prediction on the Retweet Actions over Microblog Dataset

“…The number of times of possible-view of a tweet is defined as the sum of all possible-view numbers of re-tweet actions. Existing prediction methods are mainly focused on content [1,2] and link structure [3]. But based on the characteristics of the data, existing methods are not suitable for this task.…”