Arne John Glenstrup

BTA Algorithms to Ensure Termination of Off-Line Partial Evaluation

A partial evaluator, given a program and a known “static” part of its input data, outputs a residual program in which computations depending only on the static data have been precomputed [3]. The ideal is a “black box” which discovers and performs nontrivial static computations whenever possible, and never fails to terminate. Practical partial evaluators fall short of this goal: they sometimes loop (typical of functional programing partial evaluation), or terminate but are excessively conservative (typical in partial deduction 1). This paper presents efficient algorithms (being implemented) for binding-time analysis for off-line specialisers. They ensure that the specialiser performs many nontrivial static computations, and are at the same time guaranteed to terminate.

Matching of Bigraphs

We analyze the matching problem for bigraphs. In particular, we present a sound and complete inductive characterization of matching of binding bigraphs. Our results pave the way for a provably correct matching algorithm, as needed for an implementation of bigraphical reactive systems.

Termination analysis and specialization-point insertion in offline partial evaluation

Recent research suggests that the goal of fully automatic and reliable program generation for a broad range of applications is coming nearer to feasibility. However, several interesting and challenging problems remain to be solved before it becomes a reality. Solving them is also necessary, if we hope ever to elevate software engineering from its current state (a highly developed handiwork) into a successful branch of engineering, capable of solving a wide range of new problems by systematic, well-automated and well-founded methods.A key problem in all program generation is termination of the generation process. This article focuses on off-line partial evaluation and describes recent progress towards automatically solving the termination problem, first for individual programs, and then for specializers and “generating extensions,” the program generators that most offline partial evaluators produce.The technique is based on size-change graphs that approximate the changes in parameter sizes at function calls. We formulate a criterion, bounded anchoring, for detecting parameters known to be bounded during specialization: a bounded parameter can act as an anchor for other parameters. Specialization points necessary for termination are computed by adding a parameter that tracks call depth, and then selecting a specialization point in every call loop where it is unanchored. By generalizing all unbounded parameters, we compute a binding-time division which together with the set of specialization points guarantees termination.Contributions of this article include a proof, based on the operational semantics of partial evaluation with memoization, that the analysis guarantees termination; and an in-depth description of safety of the increasing size approximation operator required for termination analysis in partial evaluation.Initial experiments with a prototype shows that the analysis overall yields binding-time divisions that can achieve a high degree of specialization, while still guaranteeing termination.The article ends with a list of challenging problems whose solution would bring the community closer to the goal of broad-spectrum, fully automatic and reliable program generation.

Formalizing higher-order mobile embedded business processes with binding bigraphs

We propose and formalize HomeBPEL, a higher-order WSBPEL-like business process execution language where processes are firstclass values that can be stored in variables, passed as messages, and activated as embedded sub-instances. A sub-instance is similar to a WSBPEL scope, except that it can be dynamically frozen and stored as a process in a variable, and then subsequently be thawed when reactivated as a sub-instance. We motivate HomeBPEL by an example of pervasive health care where treatment guidelines are dynamically deployed as sub processes that may be delegated dynamically to other workflow engines and in particular stay available for disconnected operation on mobile devices. We provide a formal semantics based on binding bigraphical reactive systems implemented in the BPL Tool as part of the Bigraphical Programming Languages project at ITU. The semantics is an extension of a semantics given previously for a simplified subset of WS-BPEL and exploits the close correspondence between bigraphs and XML to provide a formalized run-time format very close to standard WS-BPEL syntax, which also constitutes the representation of frozen sub-instances.

Gaze-guided viewing of interactive movies

Abstract The idea of gaze-interactive moviesisillustrated by a simple example movie that unfolds non-deterministically via an analysis of the interest of the viewer measured by theinterpretedinput from an eye tracker. We demonstrate how the amount of relative attention paid to key subjects of narrativeimportance may guide the outcome of a narrative branching. An experiment was conducted to test the operation of gaze guided film. The experiment involved 11 subjects influencing a two-minute film clip by gaze in two scenarios. In the first case subjects were aware that gaze could be used to control the narrative, and in the second case the subjects were unaware of this control. The outcome was found to be quite uniform across subjects, and it was not influenced by repetitions or by knowledge about the controloption. Comments from the aware users indicated that they were looking for confirmation of gaze selections from the system. Thus, non-intrusive feedback seems to be fundamental for a successful gaze-interactive media. We suggest a range of discrete audio and visual effects that may serve this purpose and present some narrative control principles.

Goal-driven context-aware service composition

Two important aspects are associated with service composition. One is to understand the needs and constraints for a new added-value composite service, and otherwise it would lead to an ad-hoc effort for service composition. The second is to reflect the changes of computing environment to the service composition to catch up the on-demand of users. This paper introduces a goal-driven approach to specify the user requirements and demands that guides the service composition, and proposes context awareness to adapt to a dynamically changing environment. Computing contexts, including physical context, user profile and computed results, are gathered by various services, and imported into an ontology based a context repository. A Goal Description Language, Context Condition/Effect are designed to describe the dynamic semantics of goal requirements and service capability. A planner is designed and implemented to dynamically compose services based on the current contexts, and a service runner is designed and implemented to invoke proper services based on the contexts and interactions with users.

An inductive characterization of matching in binding bigraphs

We analyze the matching problem for bigraphs. In particular, we present a sound and complete inductive characterization of matching in bigraphs with binding. Our results yield a specification for a provably correct matching algorithm, as needed by our prototype tool implementing bigraphical reactive systems.

Rough Sets Based Context-Aware Service Discovery Framework

Service discovery is one of the key tasks in SOA, locating a particular or a series of services by matching a user query with service advertisement in service repository. It’s crucial to the automation and intelligence of service invocation, achieving better user-experience of QoS. However, key-word based syntactic matchmaker currently fails to provide a satisfied service discovery mechanism. Therefore, this paper proposes a RS-CASD (Rough Set-based Context-Aware Service Discovery) Framework that explores context information and ontology model to semantically enhance the user query and service advertisement for more fulfilled service discovery. Rough set theory is applied to deal with inexact or uncertain functional and contextual properties and extraction of context rules from historical records.

Collecting location-based voice messages on a TalkingBadge

This paper presents three experiments to explore the feasibility of location-based voice messaging. We first compared three methods for collecting synthetic speech messages located in a room, namely PIN code entry, barcode scanning and automatic detection with a Bluetooth antenna. In addition to being very reliable, Bluetooth detection was significantly faster than PIN code entry and barcode scanning. We then examined detection times and errors in an open five floor building with antennas located densely. This confirmed that Bluetooth is fast enough to catch people walking through a zone and specific enough to distinguish between zones located just 20 meters apart. Finally, we played digitized voice messages to 11 participants walking into a zone. They received most of the messages well, but a majority of their comments were negative, expressing concerns for the potential infringement of privacy. We conclude that location specific audio messaging works from a technical perspective, but requires careful consideration of social comfort.

Building a context world for dynamic service composition

Dynamic service composition requires responding and adapting to changes in the computing environment when orchestrating existing services into one or more new services that fit better to a composite application. This paper abstracts the changes of the environment as a context world to store the physical contexts of the computing environment, user profiles and computed results of services as well. We use ontology techniques to model the domain concepts of application contexts. Context Condition/Effect Description Language is designed to describe the dynamic semantics of the requirements and capabilities of goals and services in a concise and editable manner. Goal-driven and planning techniques are used to dynamically implement the service composition according to the domain knowledge and facts in the context world.