Fraud and Spam Detection

When on the web, how can we trust content generated by other users? As the web has become an increasingly integral part of our daily lives, from work to shopping to socializing, it has become a focus of spammers attempting to make money off Internet users, even if it takes dubious means. Fraudsters can greatly influence the success of a restaurant through its Yelp reviews or the recommendations to a user on Facebook. For people to use and trust web services, it is crucial that we remove and prevent spam and fraud.

While spam and fraud occur in a wide variety of services and take many different forms, detecting such fraud can often be framed as a graph analysis problem. On Facebook, purchased Page Likes are attempting to manipulate the bipartite graph between users and Pages by adding edges. Likewise on Twitter, purchased followers try to add many incoming edges to certain users in the social follower graph. And even on Yelp, Amazon, or Netflix, ratings attempting to manipulate recommendations can be viewed as edges added to the bipartite graph between users and items (restaurants, products, or movies). In each case, we focus on detecting surprising graph patterns that do not occur naturally. By framing spam detection as a graph analysis problem, our algorithms are applicable to a wide variety of situations.

In working to prevent spam and fraud from effecting online services, we take a variety of approaches and have a few goals. In particular, we focus on:

• Pattern Mining – Find patterns that distinguish natural user behavior from fraudulent behavior.
• Adversarial analysis – Design algorithms that bound the amount of spam that can be added by fraudsters without being detected.
• Robust recommendation – Spam often is used to influence recommendation algorithms. Therefore, we focus on building recommendation algorithms that detect or limit the impact of spam.

Spam and fraud detection is an increasingly important and evolving field, with new issues arising each year. While we hope to build long term solutions to detect spam, we also plan to keep abreast of the changing dynamics of web services and fraud markets to detect new types of attacks.

People

• Alex Beutel
• Neil Shah
• Meng Jiang
• Kijung Shin
• Stephan Günnemann
• Polo Chau
• Christos Faloutsos

Publications

• K. Shin, B. Hooi, J. Kim, and C. Faloutsos, "D-cube: Dense-block detection in terabyte-scale tensors," in Proceedings of the Tenth ACM International Conference on Web Search and Data Mining, 2017, pp. 681-689. [PDF] [BIBTEX]
  @inproceedings{shin2017dcube, title = {D-cube: Dense-block detection in terabyte-scale tensors},
  author = {Shin, Kijung and Hooi, Bryan and Kim, Jisu and Faloutsos, Christos},
  booktitle = {Proceedings of the Tenth ACM International Conference on Web Search and Data Mining},
  pages = {681--689},
  year = {2017},
  url = {http://www.cs.cmu.edu/~kijungs/papers/dcubeWSDM2017.pdf},
 }
• K. Shin, B. Hooi, J. Kim, and C. Faloutsos, "DenseAlert: Incremental Dense-Subtensor Detection in Tensor Streams," in Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2017. [PDF] [BIBTEX]
  @inproceedings{shin2017densealert, title = {DenseAlert: Incremental Dense-Subtensor Detection in Tensor Streams},
  author = {Shin, Kijung and Hooi, Bryan and Kim, Jisu and Faloutsos, Christos},
  booktitle = {Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining},
  year = {2017},
  url = {http://www.cs.cmu.edu/~kijungs/papers/alertKDD2017.pdf},
 }
• K. Shin, T. Eliassi-Rad, and C. Faloutsos, "CoreScope: Graph Mining Using k-Core Analysis - Patterns, Anomalies and Algorithms," in ICDM, 2016. [PDF] [BIBTEX]
  @inproceedings{shin2016corescope,
  author = {Kijung Shin and Tina Eliassi-Rad and Christos Faloutsos},
  title = {CoreScope: Graph Mining Using k-Core Analysis - Patterns, Anomalies and Algorithms},
  booktitle = {ICDM},
  year = {2016},
  url = {http://www.cs.cmu.edu/~kijungs/papers/kcoreICDM2016.pdf},
 }
• K. Shin, B. Hooi, and C. Faloutsos, "M-Zoom: Fast Dense-Block Detection in Tensors with Quality Guarantees," in ECML/PKDD, 2016, pp. 264-280. [PDF] [BIBTEX]
  @inproceedings{shin2016mzoom,
  author = {Kijung Shin and Bryan Hooi and Christos Faloutsos},
  title = {M-Zoom: Fast Dense-Block Detection in Tensors with Quality Guarantees},
  booktitle = {ECML/PKDD},
  pages = {264--280},
  year = {2016},
  url = {http://www.cs.cmu.edu/~kijungs/papers/mzoomPKDD2016.pdf},
 }
• M. Giatsoglou, D. Chatzakou, N. Shah, A. Beutel, C. Faloutsos, and A. Vakali, "ND-Sync: Detecting Synchronized Fraud Activities," in Advances in Knowledge Discovery and Data Mining - 19th Pacific-Asia Conference, PAKDD 2015, Ho Chi Minh City, Vietnam, May 19-22, 2015, Proceedings, Part II, 2015, pp. 201-214. [PDF] [BIBTEX]
  @inproceedings{NDSync,
  author = {Maria Giatsoglou and Despoina Chatzakou and Neil Shah and Alex Beutel and Christos Faloutsos and Athena Vakali},
  title = {ND-Sync: Detecting Synchronized Fraud Activities},
  booktitle = {Advances in Knowledge Discovery and Data Mining - 19th Pacific-Asia Conference, {PAKDD} 2015, Ho Chi Minh City, Vietnam, May 19-22, 2015, Proceedings, Part {II}},
  pages = {201--214},
  year = {2015},
  crossref = {DBLP:conf/pakdd/2015-2},
  url = {http://dx.doi.org/10.1007/978-3-319-18032-8_16},
  doi = {10.1007/978-3-319-18032-8_16},
  timestamp = {Sun, 11 Oct 2015 19:15:09 +0200},
 }
• M. Jiang, A. Beutel, P. Cui, B. Hooi, S. Yang, and C. Faloutsos, "A General Suspiciousness Metric for Dense Blocks in Multimodal Data," in ICDM, 2015. [BIBTEX]
  @inproceedings{crossspot, title={A General Suspiciousness Metric for Dense Blocks in Multimodal Data},
  author={Meng Jiang and Alex Beutel and Peng Cui and Bryan Hooi and Shiqiang Yang and Christos Faloutsos},
  booktitle={ICDM},
  year={2015},
  organization={IEEE},
  ee = {http://alexbeutel.com/papers/crossspot-icdm15-paper.pdf},
 }
• A. Beutel, K. Murray, C. Faloutsos, and A. J. Smola, "CoBaFi: Collaborative Bayesian Filtering," in WWW, 2014, pp. 97-108. [BIBTEX]
  @INPROCEEDINGS{Coabfi,
  author = {Alex Beutel and Kenton Murray and Christos Faloutsos and Alexander J. Smola},
  title = {CoBaFi: Collaborative Bayesian Filtering},
  booktitle = {WWW},
  year = {2014},
  pages = {97-108},
  ee = {https://doi.acm.org/10.1145/2566486.2568040},
 }
• N. Günnemann, S. Günnemann, and C. Faloutsos, "Robust Multivariate Autoregression for Anomaly Detection in Dynamic Product Ratings," in WWW, 2014, pp. 361-372. [BIBTEX]
  @inproceedings{GunnemannGF14,
  author = {Nikou G{\"u}nnemann and Stephan G{\"u}nnemann and Christos Faloutsos},
  title = {Robust Multivariate Autoregression for Anomaly Detection in Dynamic Product Ratings},
  booktitle = {WWW},
  year = {2014},
  pages = {361-372},
  ee = {https://doi.acm.org/10.1145/2566486.2568008},
  crossref = {DBLP:conf/www/2014},
  bibsource = {DBLP, http://dblp.uni-trier.de},
 }
• S. Günnemann, N. Günnemann, and C. Faloutsos, "Detecting Anomalies in Dynamic Rating Data: A Robust Probabilistic Model for Rating Evolution," in KDD, 2014. [BIBTEX]
  @inproceedings{GunnemannGF14,
  author = {Stephan G{\"u}nnemann and Nikou G{\"u}nnemann and Christos Faloutsos},
  title = {Detecting Anomalies in Dynamic Rating Data: A Robust Probabilistic Model for Rating Evolution},
  booktitle = {KDD},
  year = {2014},
 }
• M. Jiang, P. Cui, A. Beutel, C. Faloutsos, and S. Yang, "Inferring Strange Behavior from Connectivity Pattern in Social Networks," in PAKDD, 2014. [BIBTEX]
  @INPROCEEDINGS{Jiang:PAKDD2014,
  author = {Meng Jiang and Peng Cui and Alex Beutel and Christos Faloutsos and Shiqiang Yang},
  title = {Inferring Strange Behavior from Connectivity Pattern in Social Networks},
  booktitle = {PAKDD},
  year = {2014},
  ee = {http://alexbeutel.com/papers/pakdd2014.getthescoop.pdf},
 }
• M. Jiang, P. Cui, A. Beutel, C. Faloutsos, and S. Yang, "CatchSync: Catching Synchronized Behavior in Large Directed Graphs," in Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining, 2014, pp. 941-950. [BIBTEX]
  @inproceedings{jiang2014catchsync, title={CatchSync: Catching Synchronized Behavior in Large Directed Graphs},
  author={Jiang, Meng and Cui, Peng and Beutel, Alex and Faloutsos, Christos and Yang, Shiqiang},
  booktitle={Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining},
  pages={941--950},
  year={2014},
  organization={ACM},
  ee = {http://alexbeutel.com/papers/kdd2014.catchsync.pdf},
 }
• N. Shah, A. Beutel, B. Gallagher, and C. Faloutsos, "Spotting Suspicious Link Behavior with fBox: An Adversarial Perspective," in ICDM, 2014. [BIBTEX]
  @inproceedings{shah2014fbox, title={Spotting Suspicious Link Behavior with fBox: An Adversarial Perspective},
  author={Shah, Neil and Beutel, Alex and Gallagher, Brian and Faloutsos, Christos},
  booktitle={ICDM},
  year={2014},
  organization={IEEE},
  ee = {http://alexbeutel.com/papers/icdm2014_fbox.pdf},
 }
• A. Beutel, W. Xu, V. Guruswami, C. Palow, and C. Faloutsos, "CopyCatch: stopping group attacks by spotting lockstep behavior in social networks," in WWW, 2013, pp. 119-130. [BIBTEX]
  @INPROCEEDINGS{Beutel2013,
  author = {Alex Beutel and Wanhong Xu and Venkatesan Guruswami and Christopher Palow and Christos Faloutsos},
  title = {CopyCatch: stopping group attacks by spotting lockstep behavior in social networks},
  booktitle = {WWW},
  year = {2013},
  pages = {119-130},
  bdsk-url-1 = {http://dl.acm.org/citation.cfm?id=2488400},
  bibsource = {DBLP, http://dblp.uni-trier.de},
  crossref = {DBLP:conf/www/2013},
  date-added = {2014-02-08 23:07:44 -0500},
  date-modified = {2014-02-08 23:07:44 -0500},
 }