Abstract: Compositionality is the tenet that the meaning of an expression is determined by the meanings of its immediate parts along with their method of combination. The semantics of artificial languages (such as programming languages or logics) are uniformly given compositionally, so that the notion doesn’t even arise in that literature. Linguistic theories, on the other hand, differ as to whether the relationship that they posit between the syntax and semantics of a natural language is structured in a compositional manner. Theories following the tradition of Richard Montague take compositionality to be a Good Thing, whereas theories in the transformational tradition eschew it.

I will look at what compositionality is and isn’t, why it seems desirable, why it seems problematic, and whether its advantages can’t be provided by other means. In particular, I argue that synchronous semantics can provide many of the advantages of compositionality, whether it is itself properly viewed as a compositional method, as well as having interesting practical applications.

Abstract: Linkage of health-related data in Australia dates back to the late 1960’s with the first inspiration coming from the United Kingdom. Since then computers have developed at a barely believable rate, and technical considerations still exist but do not pose any serious problems. Progress has been slowed by the increasing need for better privacy and confidentiality. Further complications have resulted from living in a large and diverse country ruled by several highly parochial states as well as the federal government. This presentation tells the story from a viewpoint largely based in Perth, Western Australia. In 1984 this city had a population of less than a million, and the nearest city/town of more than 20,000 people was Adelaide, more than 1,650 miles away by road. In our context, this was a benefit as much as a hindrance, and Perth has been very much the epicentre of data linkage.

Bio: After an early career in marine zoology combined with computing, John Bass has been at the leading edge of health-related data linkage in Australia since 1984. Early work on infant mortality in Western Australia resulted in a linked dataset that became the cornerstone of the Telethon Institute for Child Health Research. He then implemented the Australian National Death Index in Canberra before returning to Perth as the founding manager of the Western Australian linked health data project — the first of its kind in the country. He designed and implemented the technical system of this group, which is widely recognised as the foremost data linkage unit in Australia. John stepped aside from his position in 2000 but has continued a close relationship with the project, designing and overseeing the implementation of genealogical links and then spending several years working with state and federal government to implement the first large-scale linkage of national pharmaceutical and general practice information. This involved the development of new best-practice privacy protocols that are now widely adopted across Australia. He was a core participant in developing a detailed plan for the implementation of a second state-based data linkage unit involving New South Wales and the Australian Capital Territory. In 2008 John moved to Tasmania, where he spent four years planning and paving the way for the implementation of a state-wide data linkage unit. He is now semi-retired, but still working on new developments in data linkage technology.

ABSTRACT:
Weighted automata model quantitative aspects of systems like the consumption of resources during executions. Traditionally, the weights are assumed to form the algebraic structure of a semiring, but recently also other weight computations like average have been considered. Here, we investigate quantitative context-free languages over very general weight structures incorporating all semirings, average computations, lattices. In our main result, we derive the Chomsky-Schutzenberger Theorem for such quantitative context-free languages, showing that each arises as the image of a Dyck language and a recognizable language under a suitable morphism.

This is joint work with Manfred Droste (University of Leipzig)

BIOGRAPHY:
Prof. Dr.-Ing-habil. Heiko Vogler received the degree of Doktor in De Technische Wetenschappen at the Technische Hogeschool Twente, The Netherlands in 1986. He achieved the Habilitation in Computer Science at the RWTH Aachen in 1990, was associate professor at the University of Ulm from 1991-1994, and since 1994 he is full professor at the TU Dresden. He received the degree of Doktor honoris causa from the University of Szeged, Hungary in November 2013. His research interests are weighted tree automata and formal models for statistical machine translation of natural languages.