cms

Composition of Web services in a component-wise fashion is becoming the major paradigm for distributed computing. The two main facets of composite Web services are orchestration: the internal pattern of composition (implementation) and choreography: the external interaction model (specification).

In this talk I will investigate the former by presenting a language which describes the composition of Web services using workflow patterns, with control flow and data flow. This language allows the description of composite Web services in a component oriented fashion with a powerful data flow model. The language pulls ideas from several places, in particular WS-BPEL, and includes support for compensable transactions, as well as many other features. It is also loosely linked to Kleene algebra.

The behavioural semantics are given using an abstract relative timed process calculus which I will also describe. The use of this calculus will enable various types of model verification and provide a formal link between choreography and orchestration.

[ Slides ]

Most public image retrieval engines utilise free-text search mechanisms, which often return inaccurate matches as they in principle rely on statistical analysis of query keyword recurrence in the image annotation or surrounding text. In this talk I will present our work on building a semantically-enabled image annotation and retrieval engine that relies on methodically structured ontologies for image annotation, thus allowing for more intelligent reasoning about the image content and subsequently obtaining a more accurate set of results and a richer set of alternatives matchmaking the original query. Our semantic retrieval technology is designed to satisfy the requirements of the commercial image collections market in terms of both accuracy and efficiency of the retrieval process.

GP (for Graph Programs) is a rule-based programming language for solving graph problems at a high level of abstraction, freeing programmers from dealing with low-level data structures. The core of GP consists of just three constructs: single-step application of a set of conditional graph-transformation rules, application of a rule set as long as possible, and sequential composition of programs. The addition of a few powerful branching and iterating constructs make the language much easier to use whilst preserving the simple semantics.

As the implementation of GP nears maturity, we will discuss the overall structure of the components and the issues which have arisen in the development of this highly non-deterministic graph-transformation based system.

There is strong evidence that modal logic is *the* logic of systems just as equational logic is the logic of (universal) algebra. By systems we mean generalisations of transition systems known as coalgebras, which covers amongst others: probabilistic and weighted automata and also combination of these features e.g. combining nondeterminism and probability in various ways. Recently a PSPACE algorithm has been constructed which is capable of deciding local satisfiability in these general structures.

I will introduce a first implementation of this generic PSPACE algorithm in Haskell. This covers 8 example logics: K, KD, Monotone, Henessey-Milner, Coalition, Majority, Graded and Probabilitic Logics. Adding logics is completely modular, one only needs to provide an instance of a particular class. Moreover there is an additional level of genericity where one can combine these logics in a semantically meaningful way, crucially this inherits PSPACE decidability from its component logics.

Context-awareness within mobile learning has become an increasingly important field because of the dynamic and continually changing learning settings in the learners? mobile learning environment giving rise to many different learning contexts. Therefore it is crucial to capture these contexts in order to enhance the students? learning process whilst they are using mobile devices in mobile environments for learning. We have proposed a framework named Context-aware and Adaptive Learning Schedule (CALS) which aims to improve the process of students? studying/learning whilst they are using a mobile device (such as a Pocket PC) by taking into the contextual features in the mobile environment that they are situated in at that moment, as well as their user preferences and available time for learning at that point. For the evaluation of our framework, we intend to use Java programming learning materials as our learning objects and the target audience will be first year computer scientists. Our aim is to have an increased understanding of pedagogic knowledge in how different contexts and their factors can affect the effectiveness of mobile learning systems; produce sets of guidelines, or a context taxonomy for how to enhance the effectiveness of students? learning/studying in different locations such as library, restaurant, station, public transport, home and campus.

All computer programs execute effects: they print characters to a terminal; fork threads; or mutate values stored in memory. While we have a solid understanding of "pure" programs, i.e., those that do not perform such effects, it can be extremely difficult to debug impure code, let alone prove its correctness. To address this, I have been working on pure, denotational specifications of effects. Using such specifications, we can test, debug, and reason about programs that perform I/O as if they were pure. In this talk, I want to outline our approach, discuss the work we have done so far in Haskell, and show how a richer type theory can be used to give more precise specifications.