Type of Proposal: Paper Title: Reading the Papyrologist: Building Systems to Aid the Humanities Expert Keywords: Papyrology, Knowledge Elicitation, Cognitive Systems Author: Melissa M. Terras Affiliation: Department of Engineering Science / Centre for the Study of Ancient Documents, University of Oxford. Email: mmt@robots.ox.ac.uk Contact Address: Christ Church, Oxford, OX1 1DP Fax Number: c/o Professor Mike Brady, 01865 273908 Phone Number: 01865 282181, mobile 07977 445062 The act of Papyrology, simply defined as obtaining “a body of knowledge …from the study of papyri” and now taken to cover “ as a matter of convenience…all materials carrying writing… done by a pen” (Turner 1968, p vi) is an important aspect to our understanding of ancient societies. The reading and analysis of ancient texts can provide a vast array of historical and linguistic information, providing an important, often first hand, companion to archaeological and scientific evidence of sites and peoples. Although there has been much written regarding the history of papyrology (Pattie and Turner 1974), and the contribution the transcription of such texts has made to both literary and non-literary classical studies (Turner 1968), the process entailed in transcribing a text has never been made explicit. Papyrology is in essence a “self consuming labor which leaves little or no trace of itself,” (Youtie 1963, p 11) and the expertise of papyrologists, as with the expertise of any professional, is a valuable but surprisingly elusive resource. An EPSRC jointly funded project at the Department of Engineering Science, and the Centre for the Study of Ancient Documents (CSAD), University of Oxford, was initiated three years ago to analyse ancient texts and develop new image processing techniques to retrieve information from small incisions in damaged surfaces, the techniques under development being applicable to a wide variety of engineering problems. The project is concentrating on the analysis of the Vindolanda stylus tablets: a collection of some 200 texts discovered near a Roman Fort built in the late 80s AD near Hadrian’s Wall at modern day Chesterholm (Birley 1977; Bowman 1983). Some significant progress has been made using wavelet filtering to remove woodgrain in images of the stylus tablets, and developing and appropriating Shadow Stereo techniques to identify candidate writing strokes (Bowman, Brady et al. 1997; Molton 1999); these developments were presented at ALLC 2000. However, to be able to help the papyrologists, it is important to develop tools which can be utilised easily and confidently by them, and also provide techniques which can mobilise the disparate linguistic and visual knowledge accessed whilst transcribing such a text. In order to do this, it is first important to understand the process that they go through in analysing a text. However, experts are notoriously bad at describing what they are expert at (McGraw and Harbison-Briggs 1989). Experts utilise and develop many skills which become automated and so they are increasingly unable to explain their behaviour, resulting in the troublesome “knowledge engineering paradox”: the more competent domain experts become, the less able they are to describe the knowledge they use to solve problems (Waterman 1986). A program of knowledge elicitation was undertaken to identify the areas in which a computer program could aid the papyrologists, and also how the development of such a system would add to our understanding of the papyrology process. This has resulted in the development of a knowledge based program, firstly to try and make explicit the reading techniques that papyrologists use whilst deciphering such texts (which have so far remained implicit), and secondly, to provide a tool to aid the papyrologists in recording the recursive hypotheses they develop in the transcription of the texts. This program incorporates lexical and visual knowledge that the papyrologists rely on to help them read such ancient texts, and will compliment the development of the image processing techniques, eventually contributing to the development of a Cognitive Visual System which will aim to replicate the human ability to combine image processing, reasoning, memory, and knowledge. (It should be stressed that this is not an attempt to build a prescriptive “expert system” that will automatically generate a reading of the texts, but a tool through which the papyrologists can record the various hypotheses and reasoning generated during the transcription of such texts.) Developing such a program requires much interaction with the papyrologists to understand the process of reading such ancient texts, analysis of existing linguistic sources from the same period, and development of appropriate computer skills to undertake the building of the program. This paper will present an overview of the project, and a demonstration of the computer program developed so far, followed by a discussion regarding what the development of such a program has revealed about the nature of the papyrology process and how experts approach and read ancient texts. Birley, R. (1977). Vindolanda, A Roman Frontier Post on Hadrian's Wall. London, Thames and Hudson. Bowman, A. K. (1983). The Roman Writing Tablets From Vindolanda. London, British Museum Publications LTD. Bowman, A. K., J. M. Brady, et al. (1997). “Imaging Incised Documents.” Literary and Linguistic Computing 12(3): 169 - 176. McGraw, K. L. and K. Harbison-Briggs (1989). Knowledge Acquisition: Principles and Guidelines. London, Prentice-Hall International Editions. Molton, N. and J. M. Brady (1999). “The Choice of Light Position for Shadow Stereo with Inscribed Tablets.” Forthcoming. Pattie, T. S. and E. G. Turner (1974). The Written Word on Papyrus. London, British Museum Publications. . Turner, E. G. (1968). Greek Papyri, An Introduction. Oxford, Clarendon Press. Waterman, D. A. (1986). A Guide to Expert Systems. Reading, MA, Addison- Wesley. Youtie, H. C. (1963). “The Papyrologist: Artificer of Fact.” GRBS 4 (1963): 19-32.