Browse Prior Art Database

A System and Method to Model User's Behavior in Online Social Networks with MapReduce

IP.com Disclosure Number: IPCOM000236556D
Publication Date: 2014-May-02
Document File: 6 page(s) / 253K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed are a method and system to model user’s behavior in Online Social Networks (OSNs) with MapReduce that scales with regard to the social network size, allows real-time updates on the graph representation of the social network, allows updates through streaming data, and hence allows adaptive modeling and evolution of the models.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 41% of the total text.

Page 01 of 6

A System and Method to Model User '

MapReduce

Online Social Networks (OSNs) are a widely used tool of information diffusion , a process for widely spreading a new idea or action through communication channels . Large OSNs are useful for studying information diffusion as topic propagation in blog space, linking patterns in a blog graph, favorite photo marking in a social photo sharing service, among many other domains.

In this context, modeling user's behavior on these OSNs becomes an interesting problem due to the variety, complexity, and abundance of data within online media .

Agent-Based Simulation (ABS) is one type of modeling that is consistent with the sciences of complexity. Agents are autonomous entities capable of acting without direct external intervention. Multi-agent systems can handle the complexity of solutions through decomposing, modeling, and organizing the interrelationships between components. Hence, methods can model each user behavior independently based on the available OSN data.

Big data plays a very important role when modeling OSN. For instance, a single social network can generate millions of posts per day. A method is needed to build agent-based models using a massive amount of data and very large networks .

The novel contribution is a MapReduce*-based solution for building agent-based models with massive amounts of data.

Since its first seminar paper about MapReduce, there are many other works proposing the use of MapReduce framework for a wide different type of problems . Common to all of these works, the main applications is to scale out data analytics workloads in cheap clusters.

The proposed solution is different from existing works because it scales multi -agent based simulation of large social networks. This method is the first to address the problem of building agent-based information diffusion models on large OSN using the MapReduce framework.

The core idea of is a method and system to model user's behavior in Online Social Networks (OSNs) with MapReduce that scales with regard to the social network size, allows real-time updates on the graph representation of the social network , allows updates through streaming data, and hence allows adaptive modeling and evolution of the models. The method is composed of three jobs :


1. User's states computation

2. User's transition computation
3. The computation of the pattern recognition of the transitions

With these jobs, it is possible to separately store the user's states, the transitions, and the patterns. Therefore, any changes and evolution on the user's states and transitions

1

'''s Behavior in Online Social Networks with

s Behavior in Online Social Networks with


Page 02 of 6

can be recomputed for the user and the associated network , rather than to the full network. Moreover, big networks can be computed in a distributed environment .

Figure 1: Overview

Figure 1 shows the MapReduce-based proposed solution from a macro perspective . For each user, the...