In memoriam : Karen

Karen Spärck Jones, one of the pioneers of information retrieval (IR), died on 4 April 2007 at the age of 71. Her career history can be stated simply. She was born in Huddersfield in 1935 and brought up there before going to Cambridge University in 1953 to read history. With the exception of a brief period as a teacher after graduating, she spent her entire life at the University, initially as a PhD student and then research worker in the famous Cambridge Language Research Unit under the supervision of Margaret Masterman. She was awarded her doctorate in 1964, her thesis, ‘Synonymy and Semantic Classification’, having the distinction of being republished in full by Edinburgh University Press over 20 years later, by which time the importance of her doctoral research had become evident to the wider community [1]. Following her time with the Unit, she was successively a Research Fellow of Newnham College, Royal Society Research Fellow, Senior Research Associate, GEC Fellow and then Assistant Director of Research. She only achieved a permanent position in the University in 1994 when she was made Reader in Computers and Information, and was finally awarded her chair in 1999, just three years before her retirement. She was a Fellow of the American Association of Artificial Intelligence (AAAI) and of the European Coordinating Committee for Artificial Intelligence, and the President of the Association for Computational Linguistics (ACL) in 1994. Her major awards included the Gerard Salton Award of the Association for Computing Machinery (ACM) SIGIR in 1988, the ACL Lifetime Achievement Award in 2004, and the British Computer Society Lovelace Medal and the ACM/AAAI Allen Newell Award in 2007. She was also the recipient of a substantial festschrift, containing contributions from around the world from those who had worked with her throughout her career [2]. Karen’s PhD had involved the construction of a thesaurus using text distribution data and statistical classification methods, and this led directly to her pioneering studies of keyword classifications. The work involved grouping words on the basis of co-occurrence data, with the aim of enhancing, principally, the recall of IR systems by allowing matches on clusters of words, rather than just the individual words in a query or document [3]. Although this work was ultimately unsuccessful, it involved the development of a systematic approach to the testing and evaluation of IR systems, with rigorous experimental design being used to control the many variables that can affect retrieval effectiveness. This detailed approach, which is clearly demonstrated in the two major British Library projects that she undertook in the 1970s [4, 5], characterized all of Karen’s work and set a standard for researchers in the field that they needed to aspire to if their results were