Call for Contributions
Information Sciences
Special Issue on Processing and Mining Complex Stream Data

The paper submission deadline March 18, 2013

Data mining and machine learning have shown tremendous methodological development and have been applied to various real-world problems. Nevertheless, many of current approaches assume processing static and simple (usually tabular) forms of data. On the other hand, modern applications and the rapid grows of information technologies give access to massive, complex and dynamic data.

Massive data are generated every day in many fields. These are often machine-generared data produced, e.g., by sensor or monitoring systems. Enormous amount of data overhelm current computer systems with respect to storing, processing and analysing them under acceptable space and time constraints. Moreover these data are no longer of a standard form but they could be represented in more complex structures offering richer descriptions of real-world objects. Both massive and complex data characteristics require new scalable algorithmic solutions allowing for data summarizing, sampling and approximating. New architectures for efficient managing such data and quering them are also necessary as well.

This need is particularly relevant in the emerging data stream mining domain, where large volumes of data records are generated continuously. The amounts of data arriving at a high rate, often with dynamically changing characteristics, require real-time or near-real-time analysis and introduce constraints over the available amount of memory. Another important aspect of mining data streams refers to changes in the data distributions and target concepts over time. Detecting theses changes and adapting classifiers to concept drifts becomes one of the challenges for new scalable algorithms.

• Scalability in processing massive data volumes
• Handling machine-generated data streams
• Approximate processing and approximate queries
• Near-real-time analytics of massive and stream data
• Discovering complex patterns in data, including multi-labeled classification and structured, complex decisions
• Classification, clustering and frequent patterns from data streams
• Detecting and adapting to changes and concept drifts in evolving data streams
• Ensemble learning in changing environments
• Efficient algorithms for mining data streams in ubiquitous environments
• Handling uncertainty in mining stream data
• Cleaning algorithms for data stream mining
• Adaptive, complex learning from rare and imbalanced data
• Architectures of data repositories for learning in complex and dynamic environments
• Data stream mining and processing over cloud infrastructures
• Applications requiring mining massive, complex and stream data

Last changes: Feb. 21, 2013 - J.Stefanowski