Context Sensitive Systems

Requirements

none

Description

Examples of context-sensitive systems—so-called cyber-physical systems—in industrial applications (Industry 4.0), context-aware smartphone apps such as Google Now, location-aware advertising, the smart home, or ERP systems that optimize decisions using real-world information and thus implicitly interact with people and the environment.

What they all have in common is that they reduce the discrepancy between the real world and IT systems through the massive, automated analysis of time series and sensor data. By utilizing context in human-to-human interactions, as well as in machine-to-machine and human-to-machine interactions—and vice versa—explicit communication can be significantly optimized.

This lecture aims to provide insight into current research and development as well as an overview of the technology stack required for implementation. The lecture is based on methods of machine learning and data analysis, but the focus is on the system and its concrete application. The target of the lecture is to convey the necessary approach to the development and design of context-aware applications. To this end, industry examples from the Smart Data Innovation Lab (www.sdil.de) will be used. In addition to essential fundamentals, the course will therefore provide, in particular, a cross-domain perspective on methods, procedures, and frameworks.

Accompanying the lecture is the practicum of the same name, which provides a practical deep dive into the course content through the application of technologies from the divisions of machine learning algorithms, predictive analytics, and smart/big data technologies, as well as datasets from real-world applications. Concurrent participation in the practicum is recommended.

References

Earley, Seth. "Analytics, Machine Learning, and the Internet of Things." IT Professional 1 (2015): 10–13. (

Schilit, Bill, Norman Adams, and Roy Want. “Context-aware computing applications.” Mobile Computing Systems and Applications, 1994. WMCSA 1994. First Workshop on. IEEE, 1994.

Abowd, Gregory D., et al. "Towards a better understanding of context and context-awareness." Handheld and Ubiquitous Computing. Karlsruhe, 1999.

Course Content

Context-awareness is the property of an application to behave in a manner appropriate to the situation. Examples of current context-aware systems include mobile apps that adapt their output to environmental conditions based on usage history, location, and built-in sensors.

Context (as in interpersonal communication) forms the basis for more efficient interaction between computer systems and their users, ideally without explicit input. Context recognition also supports complex decision-making in a wide variety of systems through predictions based on large amounts of data. The various facets of the concept of context—such as sensory, application, and user context—that are necessary for understanding context-sensitive systems will be explained in the lecture, and a general design approach for context processing will be derived.

A system obtains knowledge about the current and anticipated context by continuously preprocessing time series and sensor data streams and classifying them through predictive analyses. A wide variety of machine learning methods for creating suitable models will be presented in the lecture. The lecture focuses on the design, implementation, and integration of a complete, efficient, and distributed processing chain based on appropriate “big data” approaches. Appropriate technical solutions for large datasets, real-time processing, various data types, sensitive data, and data quality are discussed in relation to the field of application. The lecture also covers knowledge and methods in the divisions of sensor technology, sensor-based information processing, knowledge-based systems and pattern recognition, and intelligent, reactive systems.

Requirements for Admission

None

Brief Description

This lecture is an advanced course in the division of data analytics that focuses on the detection, processing, and utilization of context (machine, user, and environmental information, etc.) in software systems.

Workload

The total workload for this course is approximately 120 hours (4.0 credits).

Activity

Workload

1. attendance: Attending the lecture

15 x 90 min

22 h 30 min

Preparation and follow-up for the lecture

15 × 90 min

22 h 30 min

Reviewing the reading material

14 × 45 min

10 h 30 min

Independent exercises

14 x 45 min

10 h 30 min

Review the slide set twice

2 x 12 hours

24 h 00 min

Prepare for the exam

30 h 00 min

TOTAL

120 h 00 min

Workload for the learning unit "Context-Sensitive Systems"

Target Audience

Students in a Diplom or Master’s degree program in Computer Science

Students in a diploma or master’s degree program in information management

Target

The target of the lecture is to impart knowledge of the fundamentals as well as advanced methods and techniques related to context-sensitive systems.

Upon completion of the lecture, students will be able to

• discuss the concept of context and list various contexts relevant to information processing by humans and computers
• categorize and distinguish context-sensitive systems based on various criteria
• Derive concrete technical implementations using existing components from methods for context acquisition, networking, feature extraction, classification, and adaptation based on a reference architecture
• evaluate and compare the performance of specific context-sensitive systems using experimentally determined metrics
• Address problems related to the scalability of data analysis methods in practical applications through the use of big data architectures
• Design new context-sensitive systems based on given requirements using existing “sensor,” “machine learning,” and “big data” components.

Exam

Assessment criteria are explained in the module description.