Multi-context systems (MCS) represent contextual information flow. We show that the semantics of an MCS is completely determined by the information that is obtained when simulating the MCS, in such a way that a minimal amount of information is deduced at each step of the simulation. In MCS, the acquisition of new information is based on the presence of other information only. We give a generalized account to model situations in which information can be obtained as a result of the absence of other information as well. 1

We have now reached a stage in the lifecycle of semantic technologies, where a major integrative effort is needed, so as to produce the infrastructure that the community requires. A large-scale effort is needed to tackle the `holistic' problem of an effective integration of technologies and methods. This effort would make the acquisition, design, development and maintenance of large, heterogeneous semantic applications, which are now, in principle, feasible, more cost-effective.

Summary. This chapter explores three aspects of learning in document analysis: (1) field classification, (2) interactive recognition, and (3) portable and networked applications. Context in document classification conventionally refers to language context, i.e., deterministic or statistical constraints on the sequence of letters in syllables or words, and on the sequence of words in phrases or sentences. We show how to exploit other types of statistical dependence, specifically the dependence between the shape features of several patterns due to the common source of the patterns within a field or a document. This type of dependence leads to field classification, where the features of some patterns may reveal useful information about the features of other patterns from the same source but not necessarily from the same class. We explore the relationship between field classification and the older concepts of unsupervised learning and adaptation. Human interaction is often more effective interspersed with algorithmic processes than only before or after the automated parts of the process. We develop a taxonomy for interaction during training and testing, and show how either human-initiated and machine-initiated interaction can lead to human and machine learning. In a section on new technologies, we discuss how new cameras and displays, web-wide access, interoperability, and essentially unlimited storage provide fertile new approaches to document analysis. 1

Context modeling is critical for the unambiguous and effective delivery of data and knowledge on the Web, and in particular, the Semantic Web. However, Contexts are often embedded in the application programs or are implied by the application- or community-specific agreements. In this paper, we propose a framework for contexts modeling that can provide formal and explicit representations for contexts on the Semantic Web. The core component of the framework is a new formalism of contexts based on the notion of jurisdictions of knowledge statements. We also discuss the semantic assumption (modeled as institutions) and provenance aspects of contexts. 1.