Contact

References

Scharlaken Koord: Online prostitution charted with OSINT and Analytics

How do you help sex workers in a world which is moving more and more towards a digital twilight zone? Scharlaken Koord is since recently using a specially developed solution for data analytics, based on web crawling and Open Source Intelligence (OSINT). With this, the foundation has a quick and efficient insight on online prostitution and can act quickly on signs of abuse.

The number of brothels in the Netherlands is decreasing, and also window prostitution is decreasing. This is due to more and more sex workers offering their services online. This development puts them out of the reach of Scharlaken Koord, a foundation aimed at helping prostitutes in the region Amsterdam, Flevoland, and Haarlem. “We make sex workers stronger, and offer social support”, Marijn Goodijk explains, project leader digital support at Scharlaken Koord. “Prostitution seems easy money, but along the line, sex workers are often confronted with abuse or black mail.”

Scharlaken Koord therefore offers prevention, for example by distributing information on schools about sexuality, relations, or the dangers of lover boys. Goodijk: “But we also want that every sex worker knows how to find us in case they have a negative experience. Because of the move of prostitution to the internet, this became increasingly difficult in recent years.”

Manual scanning sex websites
In 2016, Goodijk, as an intern, researched the size of internet prostitution. “The results were shocking. We ended up with 8,500 unique telephone numbers per day, and we did not know how many persons were behind this”, he says. It was decided to run a pilot to try to reach online sex workers, with Goodijk as project leader. At first, the pilot was directed at the region Flevoland. Half a year later, this was expanded to the region Haarlem.

Goodijk: “We started scanning the two largest Dutch sex websites twice a week. Manually, we looked up information about online prostitutes, such as name, age, gender, services, location, telephone number, availability, price, and level of activity. We also examined advertisements for signs of human trafficking. We copied all this data to Excel. We then sent all persons a text message about our support services, in English when we were in doubt of the language. We got a lot of response. Questions, but also people who wanted help leaving the business."

Brainstorming with analytics expert
Scharlaken Koord wanted to intensify the online scanning and collection of Open Source Intelligence, but wanted it automated. “It took too long; every scan took four hours. Moreover, we were structurally missing the people who were not advertising on the days of our scans and could not draw proper conclusions from the data. We had to increase the frequency to do so”, according to Goodijk. From the police, with whom Scharlaken Koord cooperates, Goodijk heard about DataExpert. “Our contact person advised us to first brainstorm with IBM-partner DataExpert, a specialist in analytics they were also working with. During the first conversation, it became clear DataExpert knows our problems well.”

“We wanted to intensify online scanning, but we wanted it automated”

       - Marijn Goodijk, project leader digital support, Scharlaken Koord

Building database completely automatically
DataExpert, in close cooperation with Scharlaken Koord, developed a customised solution to automate scanning completely. The solution makes use of special software for web crawling, in combination with IBM i2 iBase Designer/User and IBM i2 Analyst’s Notebook. The web crawler searches the two target websites, fetches the desired data, and imports it with a few mouse clicks in the right format in the IBM-database. Goodijk: “We record entities, such as person, telephone, place name, URL, and advertisements text. We further specify the person with attributes such as age, gender, etc. Connections are made between the entities, such as between a person and a place or telephone number. In this way, we are automatically building a database”.

From weeks to seconds
Scanning now takes place five days per week. The region Amsterdam, previously too large, is also included. The process is started by pushing a button and saves Scharlaken Koord many hours of manual scanning. Aggregating the data, which took three weeks before, is now done in seconds. Scharlaken Koord can search the data on details using queries, and approach sex workers with certain profiles. Goodijk: “The more our text messages are finetuned to the receivers, the better, they must have as few obstacles as possible”. The queries also help to detect suspicious activity. “If someone younger than 25 years is asking no more than 50 euro per hour, it could be a sign of human trafficking”, according to Goodijk.

Aggregating data is now done in seconds

Easily expand profiles
The queries make it easy for Scharlaken Koord to further analyse data with IBM i2 Analyst’s Notebook. Goodijk: “With a query, you can reduce the number of persons from several thousand to around ten. You can further expand these profiles in an analysis. You will see, for example, that one person is linked to three telephone numbers, or connections to other persons”. In high-risk cases, such as abuse, the analysis results are shared with the police or investigation services. Scharlaken Koord also uses the information to give municipalities insight into the nature and size of the prostitution in their area. Goodijk: “With the new system, this is much easier than it used to be. It makes us less dependable on subsidies of government and industry”.

Learned much about analytics and OSINT
DataExpert helped Scharlaken Koord to configure the web crawler, database, and analysis tool, and forged all components into one powerful solution. Goodijk: “I am still amazed by the enthusiasm with which DataExpert tackled the entire project. They are experts in their field, and obviously have more clients with similar questions as us. Not only the police, but also banks and insurance agencies who face organised crime such as fraud”.

Goodijk is very happy with the close cooperation and training. “I have built up the database with a colleague and a DataExpert consultant. We have learned a lot from the DataExpert approach. We also had training of several days in making queries and analysis. I have also followed an Open Source Intelligence course, to improve the information transfer to the police and investigation services.”

Directly exchange data with the police
Since a month and a half, Scharlaken Koord has completely switched to the new methodology. The expectations are high. “In the past year, a quarter of Scharlaken Koord’s care recipients turned out to have contacted us after having received a text message. Now we can send a lot more and directed text messages, we believe we will get even more response”, according to Goodijk. He also expects the cooperation with the police and investigation services to improve. “Particularly because they are working with the same IBM software. We are now talking about how we can optimise the information stream by directly exchanging information through our systems.” Since recently, Scharlaken Koord is performing web crawls for colleague organisations in the rest of the Netherlands. “In this way, we build an extensive network and can chart prostitution nationally”, according Goodijk. With this, Scharlaken Koord is going to help sex workers in the entire country using data analytics.