Paul B. Thistlewaite

Australian National University

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

David Hawking

Australian National University

Nick Craswell

Microsoft (United States)

Michael A. McRobbie

Indiana University

Robert K. Meyer

Deutsches Zentrum für Luft- und Raumfahrt e. V. (DLR)

Donna Harman

National Institute of Standards and Technology

Helen Ashman

University of South Australia

Steve Ball

Australian National University

Peter Bailey

Microsoft (United States)

Jason Haines

Colorado State University

Chris Johnson Nick Craswell

Cooperative Institutions

Australian National University

Microsoft Research (United Kingdom)

Commonwealth Scientific and Industrial Research Organisation

Data61

Microsoft (United States)

University of South Australia

University of Nottingham

Information Technology University

Eindhoven University of Technology

Defence Science and Technology Group

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

The KRIPKE automated theorem proving system

Lecture notes in computer science (1986)

Paul B. Thistlewaite Michael A. McRobbie Robert K. Meyer

Kripke structure

10.1007/3-540-16780-3_147

Cite

Citations (7)

ACSys TREC-7 Experiments.

David Hawking Nick Craswell Paul B. Thistlewaite

Experiments relating to TREC-7 Ad Hoc, HP and VLC tasks are described and results reported. Minor remnants of last year’s Ad Hoc methods do not appear to have resulted in worthwhile improvements in performance. However, larger benefit were gained from automatic feedback than last year and concept scoring was very beneficial in the Manual AdHoc category. In the Automatic Ad Hoc category title-only performance seems to have suffered more severely than long-topic from an umber of lexical scanning shortcomings and from an excessive stop word list. The HP track was used to validate the usability of the combination of PADRE and the Quokk a GUI. In the VLC track, the 100 gigabyte collection was indexed in under eight hours and moderately effective queries were processed in less than two seconds.

Post hoc

Cite

Citations (5)

Methods for information server selection

ACM transactions on office information systems (1999)

David Hawking Paul B. Thistlewaite

The problem of using a broker to select a subset of available information servers in order to achieve a good trade-off between document retrieval effectiveness and cost is addressed. Server selection methods which are capable of operating in the absence of global information, and where servers have no knowledge of brokers, are investigated. A novel method using Lightweight Probe queries (LWP method) is compared with several methods based on data from past query processing, while Random and Optimal server rankings serve as controls. Methods are evaluated, using TREC data and relevance judgments, by computing ratios, both empirical and ideal, of recall and early precision for the subset versus the complete set of available servers. Estimates are also made of the best-possible performance of each of the methods. LWP and Topic Similarity methods achieved best results, each being capable of retrieving about 60% of the relevant documents for only one-third of the cost of querying all servers. Subject to the applicable cost model, the LWP method is likely to be preferred because it is suited to dynamic environments. The good results obtained with a simple automatic LWP implementation were replicated using different data and a larger set of query topics.

Relevance

Similarity (geometry)

10.1145/297117.297123

Cite

Citations (125)

Towards efficient "knowledge-based" automated theorem proving for non-standard logics

Springer eBooks (2005)

Michael A. McRobbie Robert K. Meyer Paul B. Thistlewaite

Pruning

Classical logic

Automated proof checking

10.1007/bfb0012833

Cite

Citations (9)

Aglets: A good idea for spidering?

Jason Haines Brendan Humphreys Chris Johnson Nick Craswell Paul B. Thistlewaite Nick Craswell

Many individuals and businesses now rely on the Web for promulgating and finding information, and in particular, rely on centralised search databases. The extent to which these databases reflect the “contents” of the Web in an accurate and timely manner is now under considerable doubt, and in any event, it is apparent that the methods used by the search engines for finding new and modified Web documents are not scaling well. To ameliorate these problems, we have been exploring the use of a “data push” model for notifying Web changes, to replace the current “data pull” model, which uses aglets (aka servlets or peerlets) to distribute the indexing task. Aglets are objects with a thread of control, that can migrate autonomously between processors in a distributed environment. As currently proposed and implemented, aglets have very few of the properties of Persistence. But as they inhabit a similar conceptual space, their properties and applications present some interest to the persistence community. Aglets have unique identities, locally persistent data and methods (in the sense that the aglet can be deactivated onto disk and reactivated), and self-migration in a distributed environment, but they do not facilitate a universal name space. They appear to be modelled on a belief that the processors in the network, while needing to be sufficiently transparent to allow some cooperative processing, will remain sufficiently opaque to thwart a realisation of the Persistence Ideals.

Source

Cite

Citations (5)

A MECHANIZED DECISION PROCEDURE FOR NON-CLASSICAL LOGICS: THE PROGRAM KRIPKE

Michael A. McRobbie Paul B. Thistlewaite Robert K. Meyer

Kripke semantics

Kripke structure

Source

Cite

Citations (0)

Information Retrieval (1999)

David Hawking Paul B. Thistlewaite Donna Harman

Comparability

Sample (material)

Relevance

10.1023/a:1009938405269

Cite

Citations (13)

Relevance weighting using distance between term occurrences

David Hawking Paul B. Thistlewaite

Relevance

Source

Cite

Citations (43)

Overview of TREC-6 Very Large Collection Track.

David Hawking Paul B. Thistlewaite

The emergence of real world applications for text collections orders of magnitude larger than the TREC collection has motivated the introduction of a Very Large Collection track within the TREC framework. The 20 gigabyte data set developed for the track is characterised, track objectives and guidelines are summarised and the measures employed are described. The contribution of the organizations which made data available is gratefully acknowledged and an overview is given of the track participants, the methods used and the results obtained. Alternative options for the future of the track are discussed. 1 Background and Motivation In the overview of the proceedings of TREC-1, Harman [1992] referred to small early test collections such as Cran eld, CACM and NPL and argued the need for a realistically-sized test collection to facilitate the transfer of laboratory-developed retrieval systems into the eld. The 2-gigabyte collection used in TREC-1 was two orders of magnitude larger than previous collections, and legitimately given the label of a very large test collection. Indeed, given the state of contemporary hardware and indexing software, it posed considerable challenges to participants. Two gigabytes remains a realistically-sized test for text retrieval applications typical of universities, research organisations, newspapers, businesses and government departments. However, it is clear that some organisations such as patent o ces and future digital libraries will demand retrieval services over collections at least two orders of magnitude larger, despite trends toward distributed information retrieval. There are already collections of the 100 gigabyte scale in the commercial world and Web search engines such as HotBot claim to index in excess of 50 gigabytes. Accordingly, in line with the initial TREC charter of realism, a need was identi ed for a test collection signi cantly larger than that used in mainstream TREC. It was not intended to replace the main TREC collection but rather to be used in a special-interest Very Large Collection (VLC) track to allow interested developers of commercial and research retrieval systems to investigate the scalability of their methods. It would also help to verify that such systems did not suddenly cease to operate due to machine or operating system limits on virtual addressing, le system The authors wish to acknowledge that this work was carried out within the Cooperative Research Centre for Advanced Computational Systems established under the Australian Government's Cooperative Research Centres Program.

Cite

Citations (34)

Proximity operators : So near and yet so far

David Hawking Paul B. Thistlewaite

Testing of the hypothesis that good precision-recall performance can be based entirely on proximity relationships in a focus of current TREC work at ANU. PADRE's Z-mode method (based on proximity spans) of scoring relevance has been shown to produce reasonable results for hand crafted queries in the Adhoc section. It is capable of producing equally good results in database merging and routing contexts due to its independence from collection statistics. Further work is needed to determine whether Z-mode is capable of achieving top-flight results. A new approach to automatic query generation designed to work with the shorter TREC-4 query produced reasonable results relative to other groups but fell short of expectations based on training performance. Investigation of causes is still under way.

Relevance

Independence

Mode (computer interface)

Cite

Citations (83)