Engineering contextual knowledge for autonomic pervasive services

The changing role of pervasive middleware: From discovery and orchestration to recommendation and planning

“…Also in this case, though, the amount of information available about the context can become so large, and account for so many useless (from the service/application viewpoint) facets of the context to be not informative per se. This may force applications and services to engage in complex information selection and reasoning to get the needed clue out of it [Cas08]. Accordingly, the middleware should take charge of digesting such large amount of contextual information, elaborating and aggregating them in order to generate comprehensive and sound view of specific situations.…”

“…In particular, novel data mining algorithms to deal with such heterogeneous and large-scale data coming from distributed devices should be created. Moreover, data mining algorithms and partial results should be integrated in a network of knowledge allowing them to take advantage of each other [Cas08].…”

“…For instance, a large portion of pervasive middleware research so far has focuses on issues such as: (i) providing effective, possibly semantically-enriched, discovery services for resource lookup, specifically tuned to the physically-situated characteristics of pervasive services [Ton08]; (ii) defining models and tools to promote context-awareness and adaptation to context by services [Cas08]; (iii) service orchestration for enabling resources sharing and composition [Lag10]; (iv) addressing non-functional issues such as security, privacy, and energy consumption [Maj10]. Without minimizing the relevance and impact of such large body of work, our opinion is that a number of additional very relevant issues arise when considering how the pervasive computing scenario is evolving.…”

A Self-organizing Approach for Building and Maintaining Knowledge Networks

“…Pattern-matching operations work rather differently from the traditional tuple space model and may exploit differentiated mechanisms for the various W4 fields. In [7] we provide several examples of knowledge representation and knowledge generation using the W4 Data Model. Figure 1 depicts the overall architecture of a W4 system.…”

“…Accordingly, a real challenge for future pervasive services is the investigation of principles, algorithms, and middleware infrastructures, via which this growing amount of distributed information can be properly represented, organized, aggregated, and made more meaningful, so as to facilitate the successful retrieval by pervasive services [7,3]. Many approaches [10,4] are currently going in the direction of adopting specific middleware layer, i.e., a knowledge network layer embedding data and knowledge network algorithms and providing effective access to such data by services.…”