Motivated by computer privacy issues, we presentthenovelproblemofdocumentrecoveryfromanindex: givenonlyadocument’sbag-of-words(BOW)vectororother typeofindex,reconstructtheoriginalordered document. Weinvestigateavarietyofindex types, includingcount-basedBOWvectors, stopwords-removedcountBOWvectors,indicatorBOWvectors,andbigramcountvectors.

provided invaluable FORTRAN programming expertise and has been the inspiration for this use of data methodology. Both my examiners, Professor Pradeep Yadav and Dr. Andrew Clare, deserve a mention for suggesting amendments that have added clarity to this thesis. On a more personal note I arn extrernely

and for sharing with us his invaluable insights into animal leaxning theory; to Andrew Barto, Richard Granger; Robert Hawkins, Eric Kandel, Joseph Steinmetz, and Richard Sutton for their comments and suggestions on an earlier draft; and to Gordon Bow~ Gregory Clark, Leon Cotg~n; Michael Foy, Stephen Kosslyn

Abstract. Semantic sensor networks use declarative descriptions of sen-sors promote reuse and integration, and to help solve the difficulties of installing, querying and maintaining complex, heterogeneous sensor net-works. This paper reviews the state of the art for the semantic specifi-cation of sensors, one of the fundamental technologies in the semantic sensor network vision. Twelve sensor ontologies are reviewed and anal-ysed for the range and expressive power of their concepts. The reasoning and search technology developed in conjunction with these ontologies is also reviewed, as is technology for annotating OGC standards with links to ontologies. Sensor concepts that cannot be expressed accurately by current sensor ontologies are also discussed. 1