《ILLC-AR-2009.pdf》由会员分享，可在线阅读，更多相关《ILLC-AR-2009.pdf（19页珍藏版）》请在三一文库上搜索。
1、Annual Report 2009 Institute for Logic, Language and Computation ILLC Universiteit van Amsterdam Science Park 904 1098 XH Amsterdam Postal adress 109 P.O. Box 94242 0 GE Amsterdam The Netherlands E: illcuva.nl 31 20 525 6051 I: www.illc.uva.nl T: + 2 Common acronyms used throughout the annual report
2、 CWI: Centrum voor Wiskunde en Informatica (Centre for Mathematics and Computer Science) FGW: Faculteit der Geesteswetenschappen (Faculty of Humanities) FNWI: Faculteit der Natuurwetenschappen, Wiskunde en Informatica (Faculty of Science) FTE: Full Time Equivalent (1.0 FTE= 38 work hours) ILLC: Inst
3、itute for Logic, Language and Computation IvI: Informatics Institute LaCo: Language and Computation LoCo: Logic and Computation LoLa: Logic and Language NWO: Nederlandse Organisatie voor Wetenschappelijk Onderzoek (Netherlands Organisation for Scientific Research) Research FTE: Part of FTE officiall
4、y dedicated to research UvA: Universiteit van Amsterdam 3 Table of Contents Chapter 1 Annual Survey. 4 Chapter 2 Research . 6 2.1 Language and Computation 6 2.1 Logic and Computation . 9 2.2 Logic and Language . 12 Chapter 3 The ILLC and the Faculties 15 Chapter 4 Management 17 4.1 People, research
5、input 17 4.2 Publications, research output . 18 4.3 Communication. 18 4.4 Events 18 4.5 Finances. 18 4.6 Administration. 19 4 Chapter 1 Annual Survey The year 2009 has been a very interesting and turbulent year for the ILLC. There were a lot of personnel changes. Frank Veltman ended his term as ILLC
6、s director in December, to be followed by Jeroen Groenendijk, who in turn was succeeded by Leen Torenvliet in September. There were also other changes in personnel that had important consequences for the ILLC. The long foreseen change of generations, that will reach its maximum around 2014 has start
7、ed. Paul Vitnyi retired in the summer, and Peter van Emde Boas retired in October. Jouko Vnnen returned to Helsinki, and Bas Terwijn unfortunately ended his appointment with the ILLC early. Fortunately for him though, to accept a permanent position at the University of Nijmegen. The changes also imp
8、lied a rotation of functions on the level of programme leadership. Vnnen was succeeded as leader of the LoCo programme by Yde Venema. The LoCo programme also welcomed a new deputy programme leader, Ulle Endriss, who took over from Leen Torenvliet. Jeroen Groenendijk accepted leadership of the LoLa p
9、rogramme. On the positive side of the personnel balance, the ILLC welcomes Alessandra Palmigiano as a new UD in the LoCo programme. 2009 was also the year of the great move. ILLC people from the Plantage Muidergracht and the Binnengasthuis terrain packed up their things in May to move to the new and
10、 wonderful Science Park. Of course, there are some initial flaws in the new location. The temperature takes some time to settle to an ideal point, the sunshades are not yet optimally configured, and the building has a whole other noise level than we were used to in the old location. To this stand ma
11、ny advantages. Not only is (almost) all of the ILLC together now, which makes communication a whole lot easier, but also the new building has many nice facilities. Everybody is working hard to optimize this new building, which will soon become a real home to the ILLC. The advantages of having not on
12、ly the colleagues of the ILLC, but also other institutes that ILLC has close cooperation with, and service departments of the Science Faculty literally within walking distance, cannot be overrated. Notwithstanding the impact that the move had on research personnel and staff, the year 2009 has been a
13、 successful year research wise. Khalil Simaan and Jeroen Groenendijk won projects in NWOs open competition and the Gloriclass project, nearing its completion, produced no less than five PhD defences in this year. As things stand today it seems that this project will be a 100% success and that all ei
14、ght students who started their research in this project will have obtained their PhD by the end of 2010. The number of projects obtaining funding in 2009 was somewhat less than what is usual in the ILLC. This is a worry, but on the other hand, the matching system changed in 2009, which on the one ha
15、nd makes the financial situation regarding projects much clearer, but also makes it clear that not all projects can be taken offhand to be profitable, or even doable, for the institute. Conclusion: yes we need to regain the level of funding, but have to do so with care. The PhD defences in the LoLa
16、programme have also reached what seems an all time high. The system of return on investment regarding PhDs is different in the Faculties of Sciences and Humanities. The Faculty of Humanities still has the concept of directly funded PhD students, and the number of students granted to a research insti
17、tute depends on a multiyear average of PhD defences in the past. The fact that LoLa has been so successful in this system means more PhD students in the years 5 to come, which will of course be welcomed by the ILLC. This effect will however not immediately be visible in 2010. Next year, 2010, will a
18、lso be the year of the midterm review. A new scientific advisory board, consisting of Angelika Kratzer, Mark Steedman and Wolfgang Thomas will evaluate the ILLCs research on the basis of a new selfstudy report and the recommendations made by their predecessors in the 2006 evaluation of the period 20
19、002005. This review was originally planned to take place in 2009, but circumstances like the change of directors and the move to the science park resulted in the postponement of this event to May of 2010. The midterm will cover the years 2006 2008. As these lines are written, the report is already p
20、rinted. The facts and figures in the report tell us that the ILLC can look forward to the midterm review with confidence. The year 2009 was a year of great changes, and the end of these changes are not yet in sight. The period until 2014 will be a period of rejuvenation, in which the ILLC will pick
21、up new momentum and a new generation will take over from the founding fathers of the ILLC. Fortunately, the ILLC has the support of both faculties that it is in, and a strong body of new leaders who will shape the ILLC to meet with modern times built upon a strong foundation. We therefore look forwa
22、rd to this period. 6 Chapter 2 Research 2.1 Language and Computation Programme leaders: Remko Scha (until May)/Rens Bod (as of May) and Khalil Simaan (deputy) Research area This programme is concerned with computational models of natural language processing and music perception. Its main goal is to
23、further our understanding of the human mind by building realistic models of these cognitive processes. The ideas developed for this purpose are sometimes extended to deal with practically useful applications, such as Machine Translation. The programme tries to synthesize the insights from traditiona
24、l generative grammar and logical semantics with probabilistic and connectionist approaches. A major focus of the programme is the development of statistical models for natural language processing and cognition. Our work in this area deals with syntactic and morphological disambiguation, language acq
25、uisition and processing, and automatic translation. It builds on our experience with the DataOriented Parsing model, which was developed and refined in this programme in the course of the last fifteen years. Our current models work with monolingual and bilingual corpora which may consist of raw text
26、 or be (completely or partially) syntactically annotated, as well as with corpora containing reading time data. We tackle challenges such as grammar and structure induction, domain adaptation, first language acquisition, modelling reading time, and syntaxenriched machine translation. We also investi
27、gate the adequacy of the syntactic/morphological annotations in treebanks for languages with more complex morphology than English. Another important application area is information storage and retrieval. Here we pursue system centred as well as usercentred approaches, in ways that contribute to our
28、understanding of the cultural and societal context of information access. We concentrate on innovative techniques which exploit textual information in combination with additional data, such as document structure, Web link structure, and other metadata. In cooperation with the programme Logic and Lan
29、guage, we develop models of linguistic processes at the level of pragmatics and discourse. Here we employ the framework of Optimality Theory to articulate complex models as hierarchies of competing constraints. Our research on music cognition and computational musicology focuses on aspects of music
30、that are fundamental but illunderstood: the perception of temporal and melodic aspects of music. We develop computational models which implement mathematically articulated theories, and we validate these models through both psychological experiments with human listeners and on huge collections of mu
31、sical data using the DataOriented Parsing approach. The models developed here can be applied in algorithms for automatic transcription, automatic accompaniment, automatic style recognition and music generation. Language research and music research deal with significantly different domains; they cann
32、ot be expected to use exactly the same concepts, tools, and techniques. But language and music do have 7 important features in common: both are sign systems evolved in human society, which rely on the human ability to perceive complex hierarchical structures in linear sequences. We believe it is use
33、ful, therefore, to explore these two domains jointly. Some convergence can be observed already. Highlights Avi Arampatzis, Jaap Kamps, and Stephen Robertson. Where to stop reading a ranked list? Threshold optimization using truncated score distributions. Proceedings of the 32nd Annual International
34、ACM SIGIR Conference on Research and Development in Information Retrieval, pages 524531. ACM Press, New York NY, USA, 2009. Rens Bod, 2009. From Exemplar to Grammar: A Probabilistic Analogybased Model of Language Learning, Cognitive Science, 33(5), 752793. Gideon Borensztajn, Willem Zuidema and Rens
35、 Bod, 2009, Childrens grammars grow more abstract with age Evidence from an automatic procedure for identifying the productive units of language. TopiCS in Cognitive Science 1:175188. Best paper award on applied cognitive modeling. Shlomo Geva, Jaap Kamps, Carol Peters, Tetsuya Sakai, Andrew Trotman
36、, and Ellen Voorhees, editors. Proceedings of the SIGIR 2009 Workshop on the Future of IR Evaluation. IR Publications, Amsterdam, 2009. Caroline van Heijningen, Jos de Visser, Willem Zuidema and Carel ten Cate, Simple rules can explain discrimination of putative recursive syntactic structures by a s
37、ongbird species, Proceeding National Academy of Sciences (PNAS), December 1, 2009, vol. 106, no. 48, pp.2053820543 Henkjan Honing, 2009. Iedereen is muzikaal. Wat we weten over het luisteren naar muziek. Nieuw Amsterdam Uitgevers. Khalil Simaan, Granted (Euro 216k) project proposal in the Free Compe
38、tition of NWO Exact Sciences Board (11 proposals granted out of 58 submissions). Title: Machine Translation When Exact Pattern Match Fails. Winkler, I., Haden, G., Ladinig, O., Sziller, I., together with Hugh Woodin, Daisuke Ikegami managed a breakthrough by calculating the consistency strength of B
39、lADRR Ongoing collaboration with Marta Blkov led Alessandra Palmigiano and Yde Venema to a very satisfactory result in the theory of coalgebraic modal logic, namely, the introduction of a sound and complete, cutfree calculus for the finitary version of Moss coalgebraic logic. A paper on a novel appr
40、oach to abduction in description logics, based on the MSc Logic thesis of our former student Szymon Klarman, has been accepted for publication in the Journal of Automated Reasoning. Coauthors are Ulle Endriss and Stefan Schlobach. Many LoCo members were involved in the conference Logic, Rationality
41、and Interaction (LORIII, Chongqing, China, October 2009). On the initiative of Johan van Benthem, a very successful day was organized to foster interaction and cooperation between Western and Chinese logicians. Dissemination and valorisation Krzysztof Apt organized an internship at the CWI for 7 pup
42、ils of an Amsterdam high school. Peter van Emde Boas gave a five week course in Guangzhou (China) on Logic, Language and Computation to academics from other fields. Johan van Benthem initiated the publication of the book “Geestdrift. Wat cognitiewetenschappers bezielt”, a book of interviews with Dut
43、ch cognitive scientists written by science journalist B. Mols. Van Benthem also wrote the Preface to this book, he was one of the interviewed scientists, and he appeared in the radio program “Hoe? Zo!” to promote the book. 11 People Programme Position NAMETotal Logic and Computation Full professorAp
44、t0.20 Benthem1.00 Buhrman0.20 Emde Boas0.60 Vnnen0.92 Vitanyi0.10 Associate professorTorenvliet1.00 Venema1.00 Assistant professorEndriss1.00 Lwe1.00 Palmigiano1.00 Terwijn1.00 Emeritus Emde Boas0.20 Jongh1.00 Vitanyi0.20 Guest PhD studentBrit1.00 Erven1.00 Garca Soriano1.00 Kontinen1.00 KoolenWijks
45、tra1.00 Loff1.00 Minica1.00 Semmes0.40 VelazquezQuesada1.00 Lecturer Barmpalias0.50 Jager0.50 Linnebank0.00 Postdoc Airiau1.00 Barmpalias0.50 Grossi1.00 Porello1.00 Uckelman, S.1.00 PhD studentDgremont1.00 Fontaine1.00 Galliani1.00 Gheerbrant1.00 Gierasimczuk1.00 Grandi1.00 Ikegami1.00 Keskinen1.00
46、Khomskii1.00 Kurzen1.00 Leal Rodriguez1.00 Vosmaer1.00 Witzel0.58 Zvesper0.66 Uckelman, J.0.42 Uckelman, S.0.83 Logic and Computation Total 38.82 12 2.2 Logic and Language Programme leaders: Frank Veltman (until September) /Jeroen Groenendijk (as of September) and Paul Dekker (deputy) Research area
47、Logic and Language (LoLa) is a broad research programme in logic and the philosophy of language, crossing the borders with linguistics and cognitive science. Human reasoning and the interpretation of natural language are the major themes. Logical and philosophical analysis is the basic scientific me
48、thod. Empirical ratification of analytical work is the main touchstone for success. Binding force is the conviction that interpretation should be studied as a dynamic cognitive process that is embedded in both social practices and the external environment. Hence, the integration of semantics and pra
49、gmatics is a dominant longterm research aim. The various systematic investigations concentrate on empirical phenomena that are intrinsically related to the way in which information is structured in conversations. Our view on how logic and language connect has obvious philosophical roots in the writings of Aristotle, Leibniz, Frege, Wittgenstein, Montague, and Grice. Systematic and historical study of the works of these intellectual forebears forms a substantial part of the project, also to stimulate c