Social Graph Visualizer Courtesy of Salmon.com Juan Zepeda Faculty - - PowerPoint PPT Presentation

social graph visualizer
SMART_READER_LITE
LIVE PREVIEW

Social Graph Visualizer Courtesy of Salmon.com Juan Zepeda Faculty - - PowerPoint PPT Presentation

Social Graph Visualizer Courtesy of Salmon.com Juan Zepeda Faculty Advisor: Xifeng Yan Lab Mentor: Yinghui Wu Computer Science UCSB Computer Science Santa Barbara City College Department Funding: Army Research Lab, NS-CTA Big Data Problem:


slide-1
SLIDE 1

Social Graph Visualizer

Faculty Advisor: Xifeng Yan Lab Mentor: Yinghui Wu UCSB Computer Science Department Juan Zepeda Computer Science Santa Barbara City College

Funding: Army Research Lab, NS-CTA

Courtesy of Salmon.com

slide-2
SLIDE 2

Big Data

Problem: Social Graphs are big. How do we analyze a

social graph to find patterns without overwhelming the user with such a complex network.

Goal: Give the user the ability to use a social graph

visualizer to view, manipulate, query and match the part of a social graph that is only important with a intuitive user interface.

2

slide-3
SLIDE 3

Project Social Viz

Software that analyzes a social graph to create relations between people and activities by answering queries.

What is a social graph?

Nodes - People Edges - Relationships

3

slide-4
SLIDE 4

Real Life Applications:

  • Improve Target

Advertising

An example of how Facebook users are interconnected Courtesy of Facebook

4

slide-5
SLIDE 5
  • Analyze

Terrorist Network

Hijacker’s Network Neighborhood Courtesy of Valdis Krebs

5

slide-6
SLIDE 6

Software Objectives

1. Interactive Graph GUI 2. Layout Algorithms 3. Node/Edge Filtering Gephi, Cytoscape, Graphviz and Other Visualization Software Project SocialViz

4. Server Side Interchangeable Database 5. Graph Queries 6. Compare Multiple Queries Results 7. Social Graph Focused

✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✗ ✗ ✗ ✗

6

Able to swap in different data sets to analyze more social networks Visualize the results

  • f the queries and not

the whole graph Able to analyze and compare results Can be applied to

  • ther fields.
slide-7
SLIDE 7

Server – Client Model

Server Client

GUI - Java Universal Network/ Graph Framework Graph Data Base and Query Algorithms

7

Query Query Results

slide-8
SLIDE 8

Application Side

8

Query Graph

Query Results Query Results Query Results

Parse GML files into nodes and edges Graph structure in memory Use JUNG to Visualize graph

slide-9
SLIDE 9

Subgraph Isomorphism Example

  • Accesses to LinkedIn Network
  • You are a boss at a company

Let’s Setup a scenario…

  • Designer
  • Engineer
  • Tester
  • Marketer

9

  • Make the perfect product
slide-10
SLIDE 10

Subgraph Isomorphism Example

Designer Engineer Marketing Tester

10

  • Experience
  • Recommendations
  • Industries
  • Education
slide-11
SLIDE 11

Query LinkedIn

Subgraph Isomorphism Example

Designer Engineer Marketing Tester Person A Person C Person B Person D

Results: Not just one query graph but many query graphs based on how relaxed the algorithm is

11

slide-12
SLIDE 12

12

Designer Engineer Marketing Tester John Mary James Linda Maria Barbara Michael Robert David Susan Jennifer William 1. 2. 3.

Interchangeable database View multiple query graph results View qualifications

  • f each person

Query submitted Give all possible results Show Statistics

slide-13
SLIDE 13

Future Plans

  • Add more query algorithms
  • Have the ability to relax the algorithms
  • Add a web interface
  • Drill-down, roll-up mechanism
  • Make it an open-source project
  • Possibly apply idea to other networks i.e.

Chemical , Biological, etc.

  • Integrate more data sets

13

slide-14
SLIDE 14

YouTube Videos

14

slide-15
SLIDE 15

Terrorist Networks

15

slide-16
SLIDE 16

Project Social Viz

This summer I just created a social graph Visualizer that can analyze social networks at UCSB with the INSET program. I learned a lot from my principle investigator and mentor about

  • research. I also was grateful for the people that pushed and lead

me here.  Nick Arnold Jens-Uwe Kuhn Virginia Estella Marilynn Spaventa Xifeng Yan Yinghui Wu & INSET

16

slide-17
SLIDE 17

Questions?

17

slide-18
SLIDE 18

Drill-down, Roll-up Mechanisim

18

Node Information

  • Name
  • Age
  • URL
slide-19
SLIDE 19

Why Java?

  • Strong Object Oriented Language
  • Portable
  • A lot Third-Party Graph Software Packages
  • APIs for Networking, GUI, Database Access

19

slide-20
SLIDE 20

20

slide-21
SLIDE 21

21

LinkedIn Facebook YouTube Drug Network Twitter