Solenix

The European Space Agency commissions bespoke software that is developed by a wide range of industry partners from different countries, with the added value that any resulting system becomes licensable to any organisation in an ESA member state. One very important aspect is to ensure licensability, i.e. the produced software does not infringe any copyright or third-party rights. Besides other roles, the European Space Operations Centre (ESOC) is responsible for the Intellectual Property Management services within ESA. This means that ESOC is responsible for ensuring that any software delivered to the Agency, and afterwards further distributed to the industry, is compliant with intellectual property rights and licensing matters. This is not an easy task and is becoming increasingly difficult with the widespread use of open source software and accompanying third party licenses. For this reason, ESOC has already several tools in place that support the analysis of source code and identifies snippets that potentially contain licensing issues. However, the analyses are often not conclusive without any human intervention to interpret their results, e.g. discarding false positives, deciding on the most likely origin, etc.

In mid-2018 we were awarded a new and very challenging project in the area of data analytics and machine learning. This project aimed at the development of a prototype, known as ADOS (Automatic Determination of genuine Origin of source code Snippets), that could support the Agency in its Intellectual Property Rights Management activities.

In the first phase of this project Solenix developed a prototype, which supported many of the manual steps involved in the analysis of reports. This prototype combines classical software engineering methods (e.g. string processing, pattern matching, source code syntax understanding) and artificial intelligence approaches (natural language processing, decision trees, neural networks) to support software license analysis and automate some of the daily work of analysts. The outputs of this first phase showed that such a prototype could not only help ESOC in potentially reducing the post-processing time, but would also be very beneficial for developers and maintainers, who implement changes as a result of the identified issues. In summary, the prototype can increase ESOC’s efficiency and reliability in licensing matters.

As a consequence, a follow-up project was commissioned in 2019 to continue our work. The prototype will be brought up closer to an operational status by integrating it with other tools, extending the use of AI, providing more concise licensing summaries and improving its usability and graphical user interface.

We are very happy to support the Agency with our expertise in Artificial Intelligence and Machine Learning in such a complex and important topic as intellectual property rights.

Image Credit: Gerd Altmann