The European Conference on Object Oriented Programming (ECOOP) was held in Lancaster, 25-29 July. This year, the conference featured the novelty of a Research Project Symposium, providing an opportunity for the dissemination of integrated project visions as well as discussions aiming to seed new collaborations and future research projects. With half-day sessions from three current European research projects, the symposium provided an interesting overview of European ICT research ranging from applications to the design of large-scale machine-translation systems and marine information systems to foundational research in the specification and verification of adaptable systems.
These projects should be of particular interest to JOT readers as they show the application of concepts from object-orientation and beyond (e.g., ontologies and components) in a variety of contexts, which is why a brief overview of the projects is given below. Full materials as well as slides and video capture of the presentations are available from the ECOOP website at 2011.ecoop.org. The text below is based on contributions from each of the projects.
HATS – Highly Adaptable Trustworthy Software
Many software projects make strong demands on the adaptability and trustworthiness of the developed software, while the contradictory goal of ever faster time-to-market is a constant drum beat. Current development practices do not make it possible to produce highly adaptable and trustworthy software in a large-scale and cost-efficient manner. Adaptability and trustworthiness are not easily reconciled: unanticipated change, in particular, requires freedom to add and replace components, subsystems, communication media, and functionality with as few constraints regarding behavioural preservation as possible. Trustworthiness, on the other hand, requires that behaviour is carefully constrained; preferably through rigorous models and property specifications since informal or semi-formal notations lack the means to describe precisely the behavioural aspects of software systems: concurrency, modularity, integrity, security, resource consumption, etc.
The HATS project develops a formal method for the design, analysis, and implementation of highly adaptable software systems that are at the same time characterized by a high demand on trustworthiness. The core of the method is an object-oriented, executable modelling language for adaptable, concurrent software components: the Abstract Behavioural Specification (ABS) language. Its design goal is to permit formal specification of concurrent, component-based systems on a level that abstracts away from implementation details, but retains essential behavioural properties. HATS is an Integrated Project supported by the 7th Framework Programme of the EC within the FET (Future and Emerging Technologies) scheme.
PRESEMT – Pattern Recognition-Based Statistically Enhanced Machine Translation
The objective of the PRESEMT project is to develop a flexible and adaptable Machine Translation (MT) system from source to target language, based on a method which is easily portable to new language pairs. This method attempts to overcome well-known problems of other MT approaches, e.g. the need to compile bilingual corpora or create new rules per language pair. PRESEMT is intended to result in a language-independent machine-learning-based methodology. To that end, a cross-disciplinary approach is adopted, combining linguistic information with pattern recognition techniques towards the development of a language-independent analysis.
PRESEMT is intended to be easily customisable to new language pairs. Consequently, relatively inexpensive, readily available language tools and internet-sourced resources are used (a large monolingual corpus in the source language and a small parallel corpus in the source and target languages), while the platform can handle the input of different linguistic tools, in order to support extensibility to new language pairs and user requirements. The translation context is modelled on phrases that are produced via an automatic and language-independent process, removing the need for specific, compatible NLP tools per language pair. Furthermore, system optimisation and personalisation is implemented via meta-heuristics (such as genetic algorithms or swarm intelligence).
PRESEMT is funded by the European Union under its Framework 7 Programme.
NETMAR – Open Service Network for Marine Environmental Data
NETMAR aims to develop a pilot European Marine Information System (EUMIS) for searching, downloading and integrating satellite, in situ, and model data from ocean and coastal areas. It will be a user-configurable system offering standards-based, flexible service discovery, access and chaining facilities. It will use a semantic framework coupled with ontologies for identifying and accessing distributed data, such as near-real time, forecast and historical data. EUMIS will also enable further processing of such data to generate composite products and statistics suitable for decision-making in diverse marine application domains.
The project aims to support operational “Smart Discovery” based on ontologies. This is the process by which users of the system are able to locate and therefore utilise datasets using search terms that are different but semantically linked to dataset labels such as keywords. A frequently quoted example is the location of datasets labelled ‘rainfall’ using the search term ‘precipitation’. In a pan-European context the issue of language arises, meaning that operational “Smart Discovery” also needs to be able to link datasets labelled in one European language to search terms supplied in another. The creation of “Smart Discovery” demonstrators is straightforward using a small ontology containing carefully selected dataset labels linked to a few search terms that are well known to the demonstrator. However, taking this to the operational scale is much more difficult because there is no foreknowledge of the search terms that will be used. Consequently, domain coverage of the underlying ontologies needs to be as close to complete as possible.
One possible approach to the development of operational ontologies might be to bring together groups of domain experts and knowledge engineers in a series of knowledge capture semantic workshops. However, this approach fails to take into account existing computerised knowledge resources making it extremely inefficient. A much more productive approach is to identify pre-existing controlled vocabularies, thesauri and ontologies that may be brought together into a single semantic resource to underpin the EUMIS.
NETMAR is funded by the European Union under its Framework 7 Programme.