Abstract To accommodate the proliferation of heterogeneous network models and protocols, the use of semantic technologies to enable an abstract treatment of networks is proposed. Network adapters are employed to lift network specific data into a semantic representation. Semantic reasoning integrates the disparate network models and protocols into a common data model by making intelligent inferences from low‐level network and device details. Automatic discovery of new devices, monitoring of device state, and invocation of device actions in a generic fashion that is agnostic of network types is enabled. A prototype system called SNoMAC is described that employs the proposed approach operating over UPnP, TR‐069, and heterogeneous sensors. These sensors are integrated by means of a sensor middleware named SIXTH that augments the capabilities of SNoMAC to allow for intelligent management and configuration of a wide variety of sensor devices. A major benefit of this approach is that the addition of new models, protocols, or sensor types merely involves the development of a new network adapter based on an ontology. Additionally, the semantic representation of the network and associated data allows for a variety of client interfaces to facilitate human input to the management and monitoring of the system.

Semantic technologies are increasingly being employed to integrate, relate and classify heterogeneous data from various problem domains. To date, however, little empirical analysis has been carried out to help identify the benefits and limitations of different semantic approaches on specific data integration and classification problems. This paper evaluates three alternative semantic techniques for performing classification over data derived from the telecommunications domain. The problem of interest involves inferring the "health" status of network nodes (femtocells) from synthesized performance management (PM) instance data based on the operational PM schema. The semantic approaches used in the comparison include OWL2 axioms, SPARQL queries and SWRL rules. Empirical tests were performed across a range of data set sizes, using Pellet for axioms and rules and ARQ for queries. The experimental results provide (mostly) quantitative and (some) qualitative indication of the relative merits of each approach. Key among these findings is confirmation of the clear superiority of queries over rules and axioms in terms of raw performance and scalability.

Integrating and relating heterogeneous data using inference is one of the cornerstones of semantic technologies and there are a variety of ways in which this may be achieved. Cross source relationships can be automatically translated or inferred using the axioms of RDFS/OWL, via user generated rules, or as the result of SPARQL query result transformations. For a given problem it is not always obvious which approach (or combination of approaches) will be the most effective and few guidelines exist for making this choice. This paper discusses these three approaches and demonstrates them using an "acquaintance" relationship drawn from data residing in common RDF information sources such as FOAF and DBLP data stores. The implementation of each approach is described along with practical considerations for their use. Quantitative and qualitative evaluation results of each approach are presented and the paper concludes with initial suggestions for guiding principles to help in selecting an appropriate approach for integrating heterogeneous semantic data sources.

A key challenge presented by the increasing volume and complexity of information collectable from battlespace situations is the development of intelligent applications able to automatically analyze the information and identify critical enemy activities in a timely manner. What makes this information fusion problem particularly difficult is the strong contextual dependency of the interpretation of complex battlespace information. This paper examines the dimensions of intelligence information, establishes a framework for exploring the effects of contextual information and provides an illustration of its use. The paper concludes with a discussion of the ramifications of context on the development of intelligent applications intended to aid in the interpretation of complex battlespace intelligence

Event correlation is a widely accepted technology for managing the complexity of modern telecommunication and data networks. This paper outlines the directions for future advancement of the technology and practice of event correlation systems. It shows that event correlation solutions are an integral part of the next generation of network management systems, which will be based on distributed service (component) architecture. We present this architecture, and discuss the importance of the unification of data and knowledge presentation formalisms, and the use of advanced technologies such as CORBA, Java, and XML. Finally, we demonstrate how these technological solutions have been implemented in the distributed event correlation system called GRACE /spl lambda/obal Real-Time Advanced Correlation Environment.

The analysis of network traffic and customer usage patterns is critical for the network operation, capacity planning and targeted marketing of the cellular industry. This data analysis task presents both an opportunity and a challenge for data warehousing technology because of the huge amount of wireless calling data and the dynamic nature of the usage reports. We have developed a data-warehousing system to address the issues of both the performance and the flexibility of the usage analysis reports. Our system collects traffic data from a wireless network and provides the following functionalities: basic reporting, dynamic modeling, customer profiling and usage forecasting. This paper focuses on the usage analysis reporting for the traffic by an individual customer or a group of customers on a subset of a network or the network as a whole. Based on the traffic usage, it describes three different usage forecasting models for capacity planning. A combined application of data warehousing and statistical analysis to a customer calling plan selection is also presented.

Knowledge-discovery systems face challenging problems from real-world databases, which tend to be dynamic, incomplete, redundant, noisy, sparse, and very large. These problems are addressed and some techniques for handling them are described. A model of an idealized knowledge-discovery system is presented as a reference for studying and designing new systems. This model is used in the comparison of three systems: CoverStory, EXPLORA, and the Knowledge Discovery Workbench. The deficiencies of existing systems relative to the model reveal several open problems for future research.>

We describe the Knowledge Discovery Workbench, an interactive system for database exploration. We then illustrate KDW capabilities in data clustering, summarization, classification, and discovery of changes. We also examine extracting dependencies from data and using them to order the multitude of data patterns. © 1992 John Wiley & Sons, Inc.

This paper presents an overview and analysis of teaming in artificial neural systems (ANSs). It begins with a general introduction to neural networks and connectionist approaches to information processing. The basis for learning in ANSs is then described and compared with classical machine learning. While similar in some ways, ANS learning deviates from tradition in its dependence on the modification of individual weights to bring about changes in a knowledge representation distributed across connections in a network. This unique form of learning is analyzed from two aspects: the selection of an appropriate network architecture for representing the problem, and the choice of a suitable learning rule capable of reproducing the desired function within the given network. The various network architectures are classified, and then identified with explicit restrictions on the types of functions they are capable of representing. The learning rules, i.e., algorithms that specify how the network weights are modified, are similarly taxonomized and, where possible, the limitations inherent to specific classes of rules are outlined.