Know center

Analysing encrypted data without exposing it

Know center describes a set of cryptographic methods it helped develop for analysing data while that data stays encrypted. The idea is that an organisation can run calculations on information that the other parties cannot read, so sensitive records can be put to use without being handed over in plain form. Two examples from its portfolio are private set intersection, which compares the database entries of several companies without any side sharing its full database, and joint evaluations, which produce shared results from pooled data without revealing the figures underneath. Know center also lists quantum-safe applications, meaning security designed to hold up if future quantum computers threaten today's encryption, together with research into symmetric cryptography and privacy-friendly analytics built on quantum information.

Machine learning for trends, anomalies and contracts

On the analytics side, Know center trains models on a company's existing data so that they can detect trends and anomalies, and it names condition monitoring as one purpose those models serve. Keeping such models working is treated as a service in its own right: the operation of the models is automated, their performance is monitored continuously, and they are retrained when needed. A separate strand covers contract and patent management, where semantic search finds documents by meaning and extracts the facts inside them. The same semantic methods support individual summaries and the analytical search of text by its features. Know center also links this work to digitising supply chains.

Connecting and preparing company data

For the data itself, Know center helps a business identify the data it holds and keep that data in good order for analysis. Different types of source are combined into one fuller overview, and automated integration pipelines clean and prepare the data for analysis. Access controls are built so that sensitive information does not leave the organisation. Visual and guided tools let people explore the combined data, and its time series work includes searching very large databases for similar past patterns so that they can be compared and learned from.

Media monitoring and semantic text models

Know center uses artificial intelligence to scan a range of online platforms and extract useful information from them. Its monitoring platform lets users explore the patterns, trends and relationships that appear in the collected material. Its modelling tools manage the semantic models behind AI text analysis and take in feedback and annotations so that those models can be refined over time.

Transparent, fair and certifiable AI

Know center sets out to make AI decisions understandable to the people who depend on them. Domain experts can pass their knowledge to the system and check its decisions, and a user-aware AI guides people through the steps of an analysis according to their goals. Bias detection asks whether particular customer or user groups are treated unfairly, and bias mitigation looks at how the effect of existing distortions can be reduced. Its work also covers checking whether datasets and algorithms meet the EU AI Act, and a fairness check that reports where an application currently stands. Recommender systems for personalised suggestions and decision support sit alongside these tools.

Co-design and workplace learning

Know center's wider practice covers the design of trustworthy AI applications with the people who will use them, and socio-technical co-design that fits new technology to existing work and life contexts. Its staff development offer structures company training, drafts learning goal catalogues and shows learning progress, and it builds conversational agents that act as learning companions on the job. It also tests data-driven, sustainable and circular business models with companies.