Ms. Schauber, what exactly is Watson?
Schauber: Watson is a technology platform that can be categorized as a cognitive system. On the one hand, it draws on traditional search algorithms, machine learning or established analysis applications - but on the other hand, it also works with a new generation of algorithms, including in the areas of neural networks, text analysis and speech recognition.

What is possible with such cognitive systems?
Watson can process structured and unstructured data in equal measure, recognize patterns, establish correlations and hidden connections and thus - after appropriate training - also develop its own understanding of topics or facts. Cognitive systems such as Watson work with probability hypotheses and suggest different options. One example: When you ask the system a question, it doesn't just kick into action as a search engine looking for the right answer, but the system actually understands the language and the meaning behind it. To do this, it breaks down the question into its individual semantic fragments, searches for the deeper meaning and then offers a selection of possible answers based on various criteria, graded according to probability. This is also new. This is a departure from the deterministic systems of the past.

What role will the technology platform play in the fourth industrial revolution?
Following the principle of cognitive systems, the development of cognitive factories that think for themselves is also conceivable. Such a 'cognitive factory' would help to gain direct insights into the production systems through dynamically provided information and thus obtain an up-to-date, centralized status picture of the plant at all times. It could also provide assistance in the assessment of incidents and problems. For example, whether it is actually necessary to shut down a system, replace a module or take measures for predictive maintenance. Above all, learning systems could also be integrated into the processes of modern production, allowing continuous improvement and best practice management.

How does Watson communicate with the outside world?
APIs are an indispensable part of the technology platform. These are interfaces through which Watson is supplied and trained with specialist knowledge, for example on healthcare, financial topics or technical knowledge. This allows software developers to easily test Watson technology and integrate it into their own applications. Around 30 APIs are currently available - and the number is constantly growing. IBM provides its own developer platform, IBM Bluemix, or the Watson Developer Cloud in Bluemix, for this purpose. Important APIs are, for example: Natural Language Processing(1) APIs are available for dialog in natural language, which allow questions to be put into the context of a specific discipline or use case with appropriate training. The Machine Learning Watson APIs in turn automate data processing, continuously learning from every interaction with the data and prioritizing it. These APIs can be used for all types of data from different sources that are generated within a specific scenario. The Video/Image/Audio Analytics Watson APIs process unstructured data, including videos, images and audio, such as video footage from surveillance cameras at industrial facilities or engine sounds.

Are the services only available via the cloud? Or can companies also install the software on their PC in the traditional way?
Both are possible. It depends on what Watson is supposed to do for the company. Intelligent search and cognitive analysis functions in structured and unstructured data, for example in texts such as workshop reports or technical documentation, are available as software. Companies can install this variant, our so-called 'Watson Explorer Advanced Edition', as software on their own systems. This is the IBM platform available 'on premise', so to speak, which supports up to 20 languages and allows basic cognitive functions, as well as access to the Watson Developer Cloud.

However, when it comes to advanced cognitive computing technologies that are adapted to specific customer situations, it is advisable to use them via the cloud. This enables rapid development and high scalability.

If a company wants to set up a private platform, the individual training required to prepare the system for certain tasks must be carried out on site to a greater extent. If, on the other hand, individual capabilities are obtained from the cloud, quick access to new functionalities is also easier. In addition, around 100 business partners are already marketing applications from the cloud commercially. This makes the technology interesting for companies of different sizes.

IBM Bluemix: In this eco-system, software developers, system integrators, small and large companies can work together with IBM experts to further develop the Watson platform and equip it with ever new capabilities.

© IBM

At the end of January, Windriver announced a collaboration with IBM and its Watson technology. What do you hope to gain from this?
The question of how and where data from smart devices, machines and systems equipped with sensors can best and most effectively be collected and analyzed is becoming increasingly important. The cloud, now a well-established location for such tasks, is of course the obvious choice. However, it is not always sensible or necessary to send data to the cloud first if there is also the option of not only capturing it directly at the point of action, but also processing and analyzing it at the same time. This is where edge computing comes into play - the relocation of computing power, applications and network services directly to the source of data generation, to the logical edge(2) of the network. This could be wind turbines, turbines, punching machines or painting robots, but also vehicles, 3D printers or ship engines.

Wind River offers software that simplifies the exchange of data between the edge of the network, i.e. the smart devices, and the IBM Watson IoT Cloud with specific 'edge-to-cloud' solutions. It also enables access to IBM Bluemix Cloud Services and Analytics, which developers can use to develop IoT-enabled smart machines more easily and quickly.

Why did IBM choose Munich of all places as the Watson IoT capital?
Munich is the ideal location for us for several reasons: firstly, it has two excellent universities that train first-class technical talent. Secondly, there is a strong economic environment with companies that are in the process of actually implementing Industry 4.0 concepts in their organizations. We have also opted for a location in Europe, which we can use as a bridgehead to the Asian market.

What about data security?
We know from many projects that company data managed via our clouds is usually more secure than on our customers' servers because we constantly monitor current security threats and condition our systems to detect threats before they cause damage. No matter which cloud variant is chosen, the data we manage is not public. Our customers decide what happens to their data.

IBM at the Hannover Messe 2016

IBM will be demonstrating how Industry 4.0 projects can be implemented step by step with various solutions and concepts at the Hannover Messe in Hall 7, Stand D18. The spectrum ranges from solutions for collecting and evaluating sensor data, consolidating and analyzing it, to intelligent software tools for predictive maintenance and the use of cognitive systems.

For example, IBM will be demonstrating how the worker on site can receive immediate assistance in solving problems that suddenly arise during the production process in dialog with the IBM Watson learning system. The pilot project from John Deere and the platform solution from Icosense show how Industry 4.0 projects can be implemented on both a large and small scale.

IBM Watson - a concrete example from manufacturing

© Fotolia, Rainer Plendl

The 'moment of truth' in automotive production usually comes at the 'marriage' of the body with the engine or a complete powertrain. This is because during final assembly, structured data in the form of error codes almost always meets unstructured data in the form of handwritten documentation reports, which are written by the fitters in prose, often with different wording, in the course of post-processing. This leaves room for interpretation; systematic evaluation, also in the sense of passing on findings to other factories, has hardly been possible to date.

This is precisely where a self-learning, cognitive system comes into its own. It is able to establish the connection between error codes and the documentation for reworking quickly and as accurately as possible. In this specific case, a major German car manufacturer decided to use IBM Watson Explorer. The system can create correlation analyses between fault codes and documentation reports, for example. This ability to combine semantic and numerical analyses helps to identify errors more quickly, improve quality within the production processes and meet tight production deadlines.