1 / 12

John Johnson

Comparing the use of Network Structure and Profile Characteristics in Discovering Groups in Social Networks. John Johnson. Original Work. “Discovering Social Circles in Ego Networks” by Julian McAuely and Jure Leskovec from Stanford University

oya
Télécharger la présentation

John Johnson

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Comparing the use of Network Structure and Profile Characteristics in Discovering Groups in Social Networks John Johnson

  2. Original Work • “Discovering Social Circles in Ego Networks” by Julian McAuely and Jure Leskovecfrom Stanford University • Automatic classification into group of contacts from Social Networking site • Unsupervised ANNutilizing network structure and profile characteristics

  3. Motivation • Does the algorithm proposed in the original paper rely more heavily on network structure or profile characteristics in determining groups?

  4. Methodology • Establish a baseline by running McAuely’s and Leskovec’s code on their unmodified data. • Determine performance of algorithm using only network structure by removing profile characteristic data. • Determine performance of algorithm using only profile characteristics by removing network structure data.

  5. Results = 73.56 min = 29.35 min = 3.02 min

  6. Results

  7. Results • Both modifications were significantly faster in running time. • Using network structure took much longer than using profile characteristics. • Using network structure resulted in far more accurate results than profile characteristics. • Using profile characteristics resulted in very small groups compared to baseline.

  8. Conclusions • Using only network structure resulted in a longer running time than using profile characteristics, implying the algorithm spends more time on the network structure. • Using only network structure resulted in far more accurate results than using only profile characteristics, implying the algorithm uses network structure more than profile characteristics.

  9. Conclusions • Algorithm proposed in “Discovering Social Circles in Ego Networks” relies more heavily on network structure than on profile characteristics.

  10. Future Work • Expanding data set to include data from other social networking sites. • Twitter & Google+ contain different profile characteristics and are directed graphs instead of undirected graphs. • Comparison against other grouping methods that use only network structure to determine efficiency of this algorithm when profile characteristics are unavailable.

  11. References J. McAuley and J. Leskovec, “Discovering Social Circles in Ego Networks,” in Neural Information Processing Systems, 2012

  12. Questions

More Related