ARTIFICIAL INTELLIGENCE FOR THE DETECTION OF TRADE-BASED MONEY LAUNDERING, SMUGGLING, BRAND AND PRODUCT PIRACY

Demo

ATTENTION! (ArTificial inTelligENce for the deTectIon of trade-based mOney lauNdering!) was an international 3-year project that started in July 2022 and ended in June 2025, enabled by ITEA and funded by the German Federal Ministry for Economic Affairs and Energy (BMWE) and Enterprise Singapore.

The Project | Web Crawler | Data Model & Red Flags | Company Networks | Trade Data | Key Results | Contact
These demonstrations show exemplary and partially anonymised results. This website is not optimised for mobile browsers.

The Project
1 / 3
ATTENTION! - ArTificial inTelligENce for the deTectIon of trade-based mOney lauNdering!
2 / 3
Project Components
3 / 3
Project Partners
❮ ❯

Prototype
1 / 2
Example screenshot of our prototype. The shown data has been anonymised.
2 / 2
First example video of our prototype. The shown data has been anonymised.
2 / 2
Second example video of our prototype. The shown data has been anonymised.
❮ ❯

Web Crawler
Example output of the web crawler searching for a set of websites about companies offering trailers.
All company names and URLs are anonymised.

Goods are traded on the web, and thus, we find relevant company and product information on the web. For example, web data can help to identify vendors of potentially illicit products by their online offers and gathers and enriches data about the online presence of said vendors.

The goal of ATTENTION!'s Web Crawler is to identify vendors of potentially illicit products using their online offers as well as to gather web data on these actors since several red flags and patterns of illicit trades require information not present in company databases.

This includes information about actual products offered by the vendor (brands, product descriptions, prices, images), company metadata such as the (claimed) company location, and behavioral / relational data such as where potentially illicit vendors advertise or which websites are linking to theirs.

Data Model & Red Flags
1 / 2
Data Model as an RDF Ontology
2 / 2

Example Query 1

The company is located in a country that is known as a tax haven.

PREFIX att: <https://attention-project.eu/rdf>
PREFIX schema: <https://schema.org/>

SELECT DISTINCT ?organization
WHERE {
  ?organization a schema:Organization ;
    schema:address ?address .
  ?address a schema:PostalAddress ;
    schema:addressCountry ?country .
  ?country a schema:Country ;
    att:isTaxHaven true .
}

Example Query 2

The weight of traded good mismatches the average for similar transactions.

PREFIX att: <https://attention-project.eu/rdf/>
PREFIX schema: <https://schema.org/>

SELECT ?company ?product ?weight
WHERE {
  ?tradeaction a schema:TradeAction ;
    schema:seller ?company ;
    att:tradedProduct ?product .
  ?company a schema:Organization .
  ?product schema:weight ?weight ;
    att:productCategory ?category .
  ?category att:hsCode ?hsCode ;
    att:avgWeight ?averageWeight .
  FILTER (?weight != ?averageWeight)
}
Example Red Flags as SPARQL Queries
❮ ❯

Company Networks
Example company network where certified retailers are marked in green and known suspicious companys are marked in red.
All company names, addresses and person names are anonymised. This demonstration is based on an early version of our prototype.

To understand patterns behind complex, large scale illicit trade and its perpetrators, it is relevant to understand company networks and to investigate entities that could be behind complex trade-based money laundering structures.

Company networks describes companies, their attributes (e.g., revenue and address) and relations (e.g., subsidiaries).

Trade Data
Example vessel movement close to a port

Trade data is about import/export transactions (example of a trade transaction: a good of category "Furniture" that is delivered from Hamburg to Singapore on May 6, 2023), and it is relevant to iunderstand the flow of goods between companies.

By connecting trade data with vessel positions, we aim at the detection of spatio-temporal patterns and anomalies in global trade.

Key Results
  • Analysis of the largest trade database of imports and exports together with extensive web content and metadata
  • Machine Learning models to understand and detect patterns of illicit trade activity
  • MVP that empowers companies and authorities to better detect illicit trade

Publications
  • 2025
    • Stefan Schestakov, Simon Gottschalk. Trajectory Representation Learning on Grids and Road Networks with Spatio-Temporal Dynamics. ACM Transactions on Intelligent Systems and Technology (TIST) 2025.
    • Tin Kuculo, Sara Abdollahi, Simon Gottschalk. Transformer-Based Architectures versus Large Language Models in Semantic Event Extraction: Evaluating Strengths and Limitations. Semantic Web Journal (SWJ) 2025.
    • Simon Gottschalk, Sergej Wildemann and Eleni Ilkou. Research Institute Knowledge Graph for Internal Organisation and Collaboration. Extended Semantic Web Conference (ESWC) 2025.
  • 2024
    • Amirabbas Afzali, Borna Khodabandeh, Ali Rasekh, Mahyar JafariNodeh, Sepehr Kazemi Ranjbar, Simon Gottschalk. Aligning Visual Contrastive learning models via Preference Optimization. The Thirteenth International Conference on Learning Representations (ICLR) 2025.
    • Michalis Mitsios, Dharmen Punjani, Sara Abdollahi, Simon Gottschalk, Eleni Tsalapati, Elena Demidova, and Manolis Koubarakis. Generating a Question Answering Dataset about Geographic Changes in a Knowledge Graph. 24th International Conference on Knowledge Engineering and Knowledge Management (EKAW). 2024.
    • Ashutosh Sao, and Simon Gottschalk. Spatially Constrained Transformer with Efficient Global Relation Modelling for Spatio-Temporal Prediction. 27th European Conference on Artificial Intelligence Ereignis (ECAI) (2024).
    • Stefan Schestakov, Simon Gottschalk, Nicolas Tempelmeier, Thorben Funke and Elena Demidova. Transferring Traffic Predictions to Urban Regions without Target Data. IEEE International Conference on Intelligent Transportation Systems (ITSC) (2024).
    • Marco Markwald and Elena Demidova. REFUEL: Rule Extraction for Imbalanced Neural Node Classification. Springer Machine Learning Journal.
    • Rajjat Dadwal, Ran Yu, Elena Demidova. A Multimodal and Multitask Approach for Adaptive Geospatial Region Embeddings. To appear in PAKDD 2024.
    • Sara Abdollahi, Tin Kuculo, and Simon Gottschalk. Event-specific Document Ranking through Multi-stage Query Expansion using an Event Knowledge Graph. The 46th European Conference on Information Retrieval (ECIR).
    • Stefan Schestakov, Simon Gottschalk, Thorben Funke, Elena Demidova. RE-Trace: Re-Identification of Modified GPS Trajectories. ACM Transactions on Spatial Algorithms and Systems (TSAS).
  • 2023
    • Alishiba Dsouza, Moritz Windoffer, Ran Yu, Elena Demidova. Iterative Geographic Entity Alignment with Cross-Attention.
    • Genivika Mann, Alishiba Dsouza, Ran Yu, Elena Demidova. Spatial Link Prediction with Spatial and Semantic Embeddings.
    • Gounoue, Steve, Ran Yu, and Elena Demidova. SCANNER: A Spatio-temporal Correlation and Neighborhood-based Feature Enrichment for Traffic Prediction. Proceedings of the 31st ACM International Conference on Advances in Geographic Information Systems. 2023.
    • Schestakov, Stefan, Paul Heinemeyer, and Elena Demidova. Road Network Representation Learning with Vehicle Trajectories. Pacific-Asia Conference on Knowledge Discovery and Data Mining. Cham: Springer Nature Switzerland, 2023.
    • Riaz, Aniqa, Sara Abdollahi, and Simon Gottschalk. Entity Typing with Triples Using Language Models. European Semantic Web Conference. Cham: Springer Nature Switzerland, 2023.

Contact