CrossSpot

This repository contains the code package for the ICDM'15 / TKDE'16 paper:

A General Suspiciousness Metric for Dense Blocks in Multimodal Data.

Spotting Suspicious Behaviors in Multimodal Data: A General Metric and Algorithms.

Meng Jiang, Alex Beutel, Peng Cui, Bryan Hooi, Shiqiang Yang, Christos Faloutsos.

Usage

crossspot.py

1. set [c_local]: count of injected, foreground block, 1000 as default
2. generate_data(): generate random data and inject the block. See global variables for data information:

• [MAX_NUM_SEED]: number of seeds in the algorithm
• [k_data]: number of modes
• [vec_n_local]: size vector of block
• [vec_n_global]: size vector of data
• [c_global]: capital C for count of the data

3. load_data(): load from file (data.csv) to

• data: entry list + value
• item2lineno: [k_data] maps, each map is {item:no. entry in [data] (lineno)}

4. CrossSpot Algorithm

• Output: screen output with best accuracy performance (maximum F1 score) with precision, recall, and F1 (average F1 score).

crossspot-less-dense.py

• change [c_local] from 3000 down to 400, from denser block to less dense block: generate data, run CrossSpot algorithm
• Output: in report.csv, best accuracy performance, avarage F1 score

Citation

If you find this repository useful in your research, please cite our paper:

@inproceedings{jiang2015general,
  title={A general suspiciousness metric for dense blocks in multimodal data},
  author={Jiang, Meng and Beutel, Alex and Cui, Peng and Hooi, Bryan and Yang, Shiqiang and Faloutsos, Christos},
  booktitle={2015 IEEE International Conference on Data Mining},
  pages={781--786},
  year={2015},
  organization={IEEE}
}

@article{jiang2016spotting,
  title={Spotting suspicious behaviors in multimodal data: A general metric and algorithms},
  author={Jiang, Meng and Beutel, Alex and Cui, Peng and Hooi, Bryan and Yang, Shiqiang and Faloutsos, Christos},
  journal={IEEE Transactions on Knowledge and Data Engineering},
  volume={28},
  number={8},
  pages={2187--2200},
  year={2016},
  publisher={IEEE}
}