Full Text / Image DBMSs

Full Text / Image DBMSs
Full Text I Image DBMSs
Robert Rowland
155
When dealing with multimedial systems, of which image DBMSs are the clearest
instance, discussion tends to focus on the implications of translating between one medium
and another and on the relationship, for example, between images and the text used to
describe them. Other chapters in this volume examine this relationship, the development
of thesaurus-based systems, and the implications of allowing the end-user some control over
the ways in which text is used to describe and retrieve the information contained in images.
Concentration on the differences between media and their implications may, however,
Iead to neglect of a number of more general issues arising out of the representation of information
within database systems as such. When these are taken into account, multimedial
systems can be approached as a special case among database systems rather than as a distinct
kind of system, and the difliculties involved in controlling the process of translation
between media can be regarded as a particular instance of the more general problem.
In such a context it is helpful to regard the representation of information within any
database system as involving three distinct Ievels:
a) raw information
b) description (of relevant aspects of (a))
c) rules (goveming (b)).
The relationship between these three Ievels form the substance of discussions about
the modelling of historical data1 and need not be expanded on here. Within such a
framework, it is helpful to distinguish between closed and open systems. In the former, all
three of these Ievels are pre-defined, and the user is constrained to work within the set of
rules worked out by the author and embodied in the design of the system. In the latter,
the end-user may be allowed to tailor the description of the raw information to bis own
requirements (while remaining within the pre-defined set of rules), or even to intervene
directly at the Ievel of the rules, changing the criteria according to which relevant aspects
of information are selected for information and embodied in textual keys2.
In a ‚unimedial‘ system containing only textual information, the above schema would
take the following form:
a) raw information full text of the documentary corpus
b) description structured description of (a)
c) rules rules for abstracting information
record structure
thesaurus of descriptive terms.
This is of course familiar ground for all those engaged in historical computing and I
shall not rehearse old arguments, in particular those regarding the respective merits of full
transcription I text retrieval versus more standard (i.e. more structured) database systems
1 Cfr., in this same series, D. Greenstein (Ed.), Modelling Historical Data, Scripta Mercaturae,
St. Katharinen, 1991
2 Cfr. the chapter by Wendy Hall and Frank Colson, pp. 87-96.
156 Robert Rowland
and procedures. It is, however, worth rememberiug that for a system to be open the raw
information must be contained within, or be accessible from within, the database. A
system where each record consists of a structured description of a document is of necessity
relatively closed, and the end-user’s freedom of intervention is limited, for example, to
the choice of the keys used to describe and retrieve a given structured (and usually very
selective) summary description of any particular document.
It is perhaps for this reason that the question of ‚openness‘ bas been raised with particular
reference to multimedial systems where the ‚raw information‘ (a bitmapped image)
is accessible from within the system. In terms of the above schema, a multimedial system
of this kind would be formally equivalent to a hypothetical database system based on a
verbatim transcription of an entire documentary corpus3, and it is therefore understandable
that there should be considerable discussion of the possibilities and implications of
making it more ‚open‘. The same would bave happened in relation to text-based systems
if verbatim transcription of an entire documentary corpus bad ever come to be regarded &s
an appropriate metbod for entering historical information into the Computer and making
it available to otber scholars4• But since bitmapped images are much less costly than
verbatim transcription5 it is natural that the discussion should have arisen here and that
it sbould have focussed on a particular case of structured description – the ‚translation‘
of visual information into verbal descriptions – and on its inherently arbitrary nature.
Cost is one of the reasons why verbatim transcription has not been generally adopted;
another is the fact that information stored in such a form is unwieldy, and that unless
the system includes a structured description of eacb document it runs the risk of being
about as useful as a map drawn to a scale of 1:1. If the question of cost is disregarded
for the time being, everything hinges on the criteria underlying the structured description.
If it were possible to design a system comprising a) verbatim transcription of the entire
3 Perhaps the best-known example of such a system is lhe Earls Colne database created by A1an
Madarlane and bis associates in Cambridge between 1971 and 1983. This database contained the
entire (pre-edited) text of a1l the surviving docwnentation regarding the parish of Earls Colne, in
Essex. The infonnation considered relevant (by Macfarlane) was parsed into a relational database
system (CODD), whence it could be accessed by means of a specially-designed query language
(CHIPS). An end-user with different analytical preoccupations could, in theory, have re-edited
the machine-readable text according to bis own interests and criteria and then proceeded to form
an eotirely different database. For a description oftbe project cfr. Alan Macfarlane, ‚The Origins
and Organization of Research‘, (Appendix H of Sarah Harrison et al., Reconstructing Historical
Communities with a Computer. Final Report to the Social Science Research Council, Department
of Social Anthropology, Cambridge, 1983). I am grateful to Alan Macfarlane for allowing me to
cite this unpubüshed report.
4 I am not of course referring, here, to the computer-based edition of historical texts, where
verbatim transcription is indispensable. Tbe problems arise when considering very extensive
corpora, such as the notarial records of an early modern town. Tbe cost of verbatim transcription
would obviously be out of all proportion to tbe benefits that the scientific community could ever
hope to obtain.
5 In terms of time, material resources and storage requirements.
Full Text / lmuge DBlHSs 157
corpus; b) a structured – and editable – description of each document; and c) a set
of rules specifying the criteria adopted in drawing up (or in modifying) the structured
descriptions, we would have an acceptable ‚open‘ text-based system6• Since verbatim
transcription is extravagant, however, it would make a great deal of sense to replace it
(within the above schema) by scanned images of the relevant documents. Each document
would be represented, within the database, by a structured description. The textual keys
used to retrieve any such descriptive record could equally give access to the raw image of
the document, and provided adequate guidelines bad been laid down, the end-user could
revise, refine and expand the description in accordance with bis analytical preoccupations
and bis own assessment of the document’s meaning and significance.
In most cases, retrieval would be effected by means of a set of textual keys, controlled
by some kind of thesaurus, and these could, as has been mentioned, be redefined by the
end-user. lt would even be possible, in some cases, to remove stich ’subjective‘ criteria from
the database altogether, and to allow interactive information retrieval by means of ‚textual
keys‘ provided by the document itself. Such a solution would require that each document
should already contain some kind of descriptive summary of itself (e.g. the ’sentence‘
in an Inquisition trial record), and that a verbatim transcription of this description be
included as a field in the database record. With suitable information retrieval software
the user could interactively build up and refine a free text query containing a set of terms
which are effectively present and associated with one another in the documents themselves.
Users with different interests and preoccupations could, with such a system, interrogate the
same database in different ways; and the possibility of accessing an irnage of the original
document from within the database can act as an aid to refining the query and as a check
on the appropriateness of the set of terms finally chosen.
An illustration of the possibilities just outlined can be provided by a project on the
records of the Portuguese Inquisition (currently at the planning stage) which I am developing
in association with the Center for Medieval and Renaissance Sturlies of U.C.L.A.
The 40 000 records of the Portuguese Inquisition (1536-1820), of which about 35 000 are
full trial records, constitute a unique source for the social historian. At present they can
only be accessed by means of an obsolete and inadequate catalogue drawn up in the 19th
Century7.
A typical trial record (whose length may vary between ten and several hundred pages)
contains the following main elements: a) transcripts of testimony against the accused,
drawn from depositions taken in other trials or from direct denunciations; b) an order for
the accused to be arrested; c) the testimony of the accused about hirnself and bis family,
and about the accusations made against him; d) a formal Statement of the accusation; e)
8 Such a system would correspond very closely to the Earls Colne project described above. lt
is ironical that this project was brought to an end, Jargely because of the cost of continuing it on
a mainframe, just as the personal computer was about to revolutionize the prospects for historical
computing. lt would not be difficult to resurrect the CODD/CHIPS system and run it on one of
the more powerful microcomputers available today.
7 An adequate catalogue now exists for the Evora tribunal, but even in this case ooly manual
searches are possible.
158 Robert Rou;land
further depositions of witnesses and the accused (possibly und er torture); f} the conclusions
of the court (sentence}, including a final Statement of the proven accusations; g) (usually)
the statement of abjuration made by the accused on publication of the sentence; and h}
follow-up documents (consignment of the accused to the appropriate civil authorities for
execution, banishment or being sent to the galleys; release of the accused after a period
of imprisonment or religious instruction, etc.). Except in the earliest trials, the prisoner
was not informed about the identity of his accusers. But he was given the opportunity of
naming all bis enemies, and thus of disqualifying the testimony of at least some of these
accusers. Since the accusations of enmity were investigated by the court, trials in which
the prisoner tried to name all bis potential accusers contain a great deal of circumstantial
evidence of interest to the social historian.
Verbatim transcription of approximately 2 million pages of difficult manuscript text
would obviously be prohibitively expensive, and quite pointless; but, given the range of
information contained in these trial records, no structured description could possibly correspond
to the full range of interests displayed by those wisbing to consult them. lt is
therefore proposed to construct a database each of whose records would contain an extended
structured description of an individual trial8, including a verbatim transcription
of the sentence. Each such record will be linked to a bitmapped image (stored on CDROM)
of the entire original documentg . The user will be able to inspect on-screen, or
print, the bitmapped reproduction of any document retrieved. The structured description
could be edited at any time by the user so as to include aspects of particular relevance to
him. Records can be retrieved by an interactive procedure making use of any combination
of terms present either in the structured description or in the transcribed text of the
sentence10, and the actual contents of the document (in this case, as summarized in the
sentence) can serve as a check on the textual keys used for information retrieval.
8 The project will build on the experience gained with an earlier project started between 1982
and 1987 at the Gulbenkian Institute of Science, Lisbon. Each trial record was there represented
by an extended structured description of up to 60 fields. This nurober will be considerably reduced
in the proposed project.
g Although no final decision has been taken on the software to be used, the project has been
designed to make use of the capabilities of the database / information retrieval system MUSCAT,
which in its current version allows both interactive information retrieval and cross-referencing,
from within the database, to bitmapped or analog images (on CD-ROM or videodisc).
1° For a description of the information retrieval system see M. F. Porter, ‚lmplementing a
Probabilistic Information Retrieval System •, Information Technology: Research and Development,
1, 1982, 131-156.
Halbgraue Reihe
zur Historischen Fachinformatik
Herausgegeben von
Manfred Thaller
Max-Planck-Institut für Geschichte
Serie A: Historische Quellenkunden
Band 14
Erscheint gleichzeitig als:
MEDIUM AEVUM QUOTIDIANUM
HERAUSGEGEBEN VON GERHARD JARITZ
26
Manfred Thaller (Ed.)
Images and Manuscripts
in Historical Computing
Max-Planck-Institut für Geschichte
In Kommission bei
SCRIPTA MERCATURAE VERLAG
St. Katharinen, 1992
© Max-Planck-Institut für Geschichte, Göttingen 1992
Printed in Cermany
Druck: Konrad Pachnicke, Göttingen
Umschlaggestaltung: Basta Werbeagentur, Göttingen
ISBN: 3-928134-53-1
Table of Contents
lntroduction
Manfred Tb aller . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1
I. Basic Definitions
Image Processing and the (Art) Historical Discipline
.Jörgen van den Berg, Hans Brandhorst and Peter van Huisstede ……………. , .. 5
II. Methodological Opinions
The Processing of Manuscripts
Manfred Tballer …….. . ……….. . . . … …. . . .. . . . ……………… . . .. …… 41
Pietonal Information Systems and the Teaching Imperative
Frank Colson and Wendy Hall . . . . . . . . . . . . . . . . . . . . . .. . .. . . .. … . . . . . .. . . . . . . . . . . . 73
The Open System Approach to Pictorial Information Systems
Wendy Hall and Frank Colson . . . . . . . . . . . . . . . . . . . . . . . . . . . . … . …….. . . . . . . .. . . . 87
111. Projects and Case Studies
Tbe Digital Processing of Images in Archives and Libraries
Pedro Gonzi.lez . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
High Resolution Images
Anthony Hamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . .. . . . . . . . . . . . . . . . 123
A Supra-institutional Infrastructure for Image Processing in the Humanities?
Espen S. Ore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Describing the Indescribable
Gerhard Jaritz and Barbara Schub . . . . . …. . … . . . . . . . . . . . . . . … . . . . . . .. . . . . .. . 143
Full Text / Image DBMSs
Robert Rowland . . . . . .. . …. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
lntrosluctjon
lntroduction
Manfred Thaller
This book is the product of a workshop held at the International University Institute
in Firenze on November 151h, 1991. The intention of that workshop has been to bring
tagether people from as ma.ny different approaches to „ima.ge processing“ as possible.
The reason for this „collecting“ approach to the subject was a feeling, tha.t wbile image
processing in many ways has been the „hattest“ topic in Huma.nities computing 1n recent
years, it may be the least weil defined. It seems also much barder to say in this area., wbat
is specifically important to historia.ns, tha.n to other people. In that situation it was feit,
that a foruin would be helpful, which could sort out what of the various approaches can
be useful in historical resea.rch.
To solve this task, the present volume has been produced: in ma.ny ways, it reflects
the discussions which actually have been going on less, than the two compa.nion volumes
on the workshops at Glasgow a.nd TromS0 do. This is intentional. On the one band,
the pa.rticipa.nts at the workshop in Firenze did strongly feel the need to have projects
represented in the volume, which were not actually present at the workshop. On the other,
the discussions for quite some time were engaged in cla.rifying what the metbodological
issues were. That is: what a.ctua.lly a.re the topics for schola.rly discussion beyond the
description of individual projects, when it comes to the processing of images in historical
resea.rch?
The situation in the a.rea is made difficult, because some of the underlying a.ssumptions
are connected with vigoraus research groups, who use fora of schola.rly debate, which are
only slightly overlapping; so, what is ta.citly a.ssumed to hold true in one group of research
projects may be considered so obviously wrang in a.nother one, that it sca.rcely deserves
explicit refutation.
We hope, that we have been succes:.ful in bringing some of these hidden diJferences
in opinion out into the open. We consider this extremely importa.nt, because only that
cla.rification allows for a fair evaluation of projects which may have sta.rted from different
sets of a.ssumption. So importa.nt, indeed, that we would like to catalogue here some of the
basic differences of opinion which exist between image processing projects. Tbe reader will
rediscover them in many of the contributions; as editor I think however, that suma.rizing
tbem at tbe beginning may make the contributions- which, of course, have been striving
for impartiality – more easily rccognizable as parts of one coherent debate.
Three basic differences in opinion seem to exist today:
(1) Is ima.ge processing a genuine and independent field of Computer ba.sed resea.rcb in
the Humanities, or is it an auxiliary tool“? Many projects a.ssume tacitly – a.nd some do so
quite outspokenly- that imag􀃣 on the computer act as illustrations to more conventional
applications. To retrieval systems, as illustrations in catalogues and the like. Projects of
this type tend to point out, that with currently easily available equipment a.nd currently
clearly understood data processing technologies, the analysis of images, which can quite
easily be ha.ndled as illustrations today, is still costly and of uncertain promise. Wb ich is the
rea.son why they a.ssume, that such analytical approaches, if at all, should be undertaken
2 Introductjon
as side effects of projects only, which focus upon the relatively simple administration of
images. Their opponents think, in a nutshell, that while experiments may be needed, their
overalJ outcome is so promising, that even the more simple techniques of today should be
implemented only, if they can later be made useful for the advanced techniques now only
partially feasible.
(2) Connected to this is another conflict, which might be the most constant one
in Humanities data processing during the last decades, is particularly decisive, however,
when it comes to image processing . Shall we concentrate on Ievels of sopbistication, which
are available for many on today’s equipment or shall we try to make use of the most
sophisticated tools today, trusting that they will become available to an increa.singly !arge
number of projects in the future? This specific battle has been fought since the earliest
years of Humanities computing, and this editor has found bimself on both sides at different
stages. A „right􀅁 answer does not exist: the debate in image processing is probably one
of the best occassions to understand mutually, that both positions are full of merit. It is
pointless to take permanently restrictions into consideration, which obviously will cease to
exist a few years from now. It discredits all of us, if computing in history always promises
results only on next years equipment and does not deliver here and now. Maybe, that is
indeed one of the more important tasks of the Association for History and Computing:
to provide a link between both worlds, Jending vision to those of us burdened down by
the next funding deadline and disciplining the loftier projects by the question of when
sometbing will be affordable for all of us.
(3) The third major underlying difference is inherently connected to the previous ones.
An image as such is beautiful, but not very useful, before it is connected to a description.
Shall such descriptions be arbitrary, formulated in the traditionally clouded langnage of
a historian, perfectly unsuitable for any sophisticated technique of retrieval, maybe not
even unambigously understandable to a fellow historian? Or shall they follow a predefined
catalogue of narrow criteria, using a carefully controlled vocabulary, for both of which it is
somewbat unclear how they will remain relevant for future research questions which have
not been asked so far? – All the contributors to this volume have been much to polite to
pbrase their opinions in this way: scarcely any of them does not have a strong one with
regard to this problem.
More questions than answers. „Image processing“, whether applied to images proper
or to digitalized manuscripts, seems indeed to be an area, where many methodological
questions remain open. Besides that, interestingly, it seems to be one of the most consequential
ones: a project like the digitalization of the Archivo General de Indias will
continue to influence the conditions of historical work for decades in the next century.
There are not only many open questions, it is worthwhile and neccessary to discuss them.
While everybody seems to have encountered image processing in one form or the
other already, precise knowledge about it seems to be relatively scarce. The volume starts,
therefore, with a general introduction into the field by· J. v.d. Berg, H. Brandhorst and
P. v. Huisstede. While most of the following contributions have been written to be as self
supporting as possible, this introduction attempts to give all readers, particularly those
lntroductjon 3
with only a vague notion of the techniques coucerned, a common ground upon which the
more specialized discussions may build.
The contributions that follow have been written to introduce specific areas, where
handling of images is useful and can be integrated into a !arger context. All authors have
been asked in this part to clearly state their own opinion, to produce clearcut statements
about their methodological position in the discussions described above. Originally, four
contributions were planned: the first one, discussing whether the more advanced techniques
of image processing can change the way in which images are analysed and handled by art historians, could unfortunately not be included in this volume due to printing deadlines:
we hope to present it as part of follow up volumes or in one of the next issues of History
and Computing.
The paper of M. Thaller argues that scanning and presenting corpora of manuscripts
on a work station can (a) save the originals, (b) iutroduce new methods for palaeographic
training into university teaching, (c) provide tools for reading damaged manuscripts, the
comparison of band writing and general palaeographic studies. He further proposes to
build upon that a new understanding of editorial work. A fairly long tr.chnical discussion
of the mechanisms needed to link images and transcriptions of manuscripts in a wider
context follows. ·
F. Colson and W. Hall discuss the role of images in teaching systems in university
education. They do so by a detailed description of the mechanism by which images are
integrated into Microcosm I HiDES teaching packages. Their considerations include the
treatment of moving images; furthermore tbey enquire about relationships between image
and text in typical stages in the dialogue between a teaching package and a user.
W. Hall and F. Colson argue in the final contribution to thill part the general case
of open systems, exemplifying their argument with a discussion of the various degrees in
which control about the choices a user has is ascertained in the ways in which navigation is
supported in a hyper-text oriented system containing images. In a outshell the difference
between „open“ and closed systems can be understood as the following: in an „open
system“ the user can dynamically develop further the behaviour of an image-based or
image-related system. On the contrary in static „editions“ the editor has absolute control,
the user none.
Following these general description of approaches, in the third part, several international
projects are presented, which describe in detail the decisions taken in implementing
„real“ image processing based applications, some of them of almost frigthening magnitude.
The contributors of this part were asked to provide a different kind of introduction to the
subject than those to the previous two: all of them should discuss a relatively small topic,
which, however, should be discussed with much greater detail than the relatively broad
overviews of the first two parts.
All the contributions growing out of the workshop came from projects, which had
among their aims the immediate applicability of the tools developed within the next 12-
24 months. As a result they are focusing on corpora not much beyond 20.000 (color) and
100.000 (blw) images, which are supposed to be stored in resolutions manageable within
:::; 5MB I image (color) and :::; 0.5 MB I image (blw). The participants of the workshop
feit strongly, that this view should be augmented by a description of the rationale behind
4 lntroductjon
the creation of a !arge scale projt’Ct for the systematic conversion of a complete archive.
The resulting paper, by P. Gonza!ez, describes the considerations which Iead to the design
of the .’\rchivo General de Indias projt’Ct and the experiences gained du ring the completed
stages. That description is enhanced by a discussion of the stratrgies selected to make the
raw bitmaps accessible via suitable descriptions I transcriptions I keywords. A critical
appraisal, which decisions would be made dilferently after the developments in hardware
tecbnology in recent years, augments the value of the description.
The participants of the workshop feit furthermore strongly, that their view described
above sbould be augmented by a description of the techniques used for the handling of
images in extremely high resolution. A. Hamber’s contribution, dealing with the Vasari
project, gives a very thorough introduction into the technical problems rncountered in
handling images of extremely high quality and also explains the economic rationale behind
an approach to start on purpose with the highest quality of images available today on
prototypical hardware.
As these huge projects both were related to iustitutions which traditionally collect
source material for historical studies, it seemed wise to include also a view on the roJe
images would play in the data archives which traditionally have been of much importance
in the considerations of the AHC. E.S. Ore discusses what implications this type of machine
readable material should bave for tbe infrastructure of institutions specifically dedicated
to Humanities computing.
Image systems which deal with the archiving of pictorial material and manuscript
systems have so far generally fairly „shallow“ descriptions. At least in art history, moreover,
the rely quite frequently on pre-defint’d terminologies. G. Jaritz and 8. Schuh describe
how far and wby historical research needs a different approacb to grasp as much of the
intemal structure and the content of an image as possible.
Last not least R. Rowland, who acted as host of the workshop at Firenze, describes tbe
considerations which currently prepare the creation of another largescale archival database,
to contain !arge arnounts of material from the archives of the inquisition in Portugal. His
contribution tries to explore the way in which the more recent developments of image
processing can be embedded in the general services required for an archival system.
This series of workshop reports shall attempt to providr a broader basis for thorough
discussions of current methodological questions. ‚fheir main virtue shall be, that
it is produced sufficiently quick to become available, before developments in this field of
extremely quick development make them obsolete. We hope we have reached that goal:
the editor has to apologize, however, that due to the necessity to bring this volume out in
time, proofreading has by neccessity be not as intensive as it should have been. To which
􀂰nother shortcoming is added: none of the persons engaged in the final production of this
volume is a native speaker of English; so while we hope to have kept to the standards of
what might be described as „International“ or „Conti111mtal“ English, the native speakers
among the readers can only be asked for their tol(‚rance.
Göttingrn, August 1992