XTRACTIS FOR homeland security

Temporal Identification of Criminal Profiles & Action Phases from Communications Metadata during Surveillance Investigations

Benchmark vs. Logistic Regression, Random Forests, Boosted Trees & Neural Networks

Design an AI-based decision system that:

  • Accuately identifies risky behavior linked to criminal activities by analyzing communication metadata from surveillance investigations, without accessing the content of telephone calls.
  • Rationally predicts dangerous Homeland Security situations.
Play Video
Goals & benefits

Identify specific metadata characterizing different criminal activities and enhance expert knowledge by helping intelligence specialists understand the causal relationships between the communication profiles and the roles inside criminal organizations.

Help intelligence services detect attacks as early as possible and understand the underlying strategy of the criminals in order to consider measures to thwart future attacks.

Avoid a large number of false alarms.

  • The top-model is a decision system composed of 12 gradual rules without chaining, aggregated into 10 disjonctive fuzzy rules.
  • Each rule uses only some of the 24 variables that XTRACTIS identified as significant (out of the 321 potential predictors characterizing the communications metadata).
  • Only a few rules are triggered at a time to compute the decision.

It has a good Real Performance for all the 6 External Test scenarios.

It computes real-time predictions up to 70,000 decisions/second, offline or online (API).

Benchmark Scores
LoR=Logistic Regression
RFo=Random Forests
BT=Boosted Trees
NN=Neural Networks

Detailed results and explanations in full document

Use Case 2024/02 (v5.0)

Powered by XTRACTIS® REVEAL12.2.44349 (2023/01)


  1. Problem Definition
  2. XTRACTIS-induced Decision System
  3. XTRACTIS Process
  4. Top-Model Induction
  5. Explained Predictions for 2 unkown cases
  6. Top-Models Benchmark
  7. Appendices