1 Introduction

Electronic media have many advantages over paper. These advantages include

flexibility with regard to editing, displaying, and annotating existing documents,
easy and quick transport of documents by means of computer networks, possibly over long distances,
support of several (virtual) copies of a document in addition to the central original document,
support of fast and convenient search mechanisms like full text searches and hypertext links.

Though these advantages of electronic media are undisputed, paper is still the most important medium for information exchange. Partly this sticking to paper documents is done from force of habit. However, paper is not favoured for subjective reasons only: in addition, the creation and reading of paper documents has been developed for thousands of years. Thus, methods have been developed that support the presentation and perception of information within these kind of documents. Most of us are used to these methods/mechanisms providing for cognitive support in situations like e. g. ``I can't remember the exact location of the information, but it is in the last quarter of the book and on the same page there is a picture of a WWW browser.'' Therefore, it seems sensible to incorporate such a cognitive support of readers within electronic access systems to documents, especially if paper is the original source of the document. In this case, the document was specifically designed for this medium and each transformation to pure electronic representation would necessarily lead to a loss of information.

Due to the huge heritage as well as the on-going creation of paper documents this incorporation of cognitive support in electronic document access systems by means of facsimile delivery seems to be necessary. Besides, it may also provide for a positive side effect: converting paper documents into electronic documents involves or - at least with regard to large quantities of documents - should involve optical character recognition (OCR). Though OCR has improved a lot over the past years, it is still far from being perfect [3]. The display of facsimiles hides these recognition errors from the users. Of course, at the same time methods have to be used that take these errors into consideration internally e. g. in case of full text searches (see section 3.2).

Hypertext representation is advisable in order to make use of the extended functionality an electronic document access system may provide. However, with regard to an electronic library that consists of a large quantity of documents, a manual conversion of linear documents into hypertext documents is not feasible. In order to manage the automatic conversion of paper documents into hypertext documents, we have designed a system called HYPERFACS that is currently being implemented [8]. This system includes an already existing elaborated browsing tool with the following characteristics, i. a.:

integration of distinct hypertext documents,
integration of distinct hypertext document types (ASCII and facsimile),
public and private hypertext links and annotations,
manual modification of the link and annotation base,
full text searches with keyword in context (KWIC) display,
graphical browser showing the hierarchical relationship between different hypertext nodes.
implementation for SUN SPARCSTATIONS using the XVIEW toolkit.

Though this system is already available across a network making use of the X window system, the additional usage of a different access system, WWW and MOSAIC, is sensible in the following cases:

the full functionality of the original HYPERFACS system is not needed,
the library system should be implemented locally without having access to the needed hardware and/or software (database management system TRANSBASE\) environment,
the library should be integrated seamlessly into the WWW [2].

In these cases, the results of the hypertext conversion of the HYPERFACS system have to be transformed into data that can be handled by means of the WWW or HTTP, respectively.

In the following we will describe some of the functionality of this WWW library. This includes already implemented, tested and projected solutions together with some of their advantages or disadvantages, respectively. Section 2 deals with the basic process of converting paper documents into WWW pages. Section 3 focuses on the access mechanisms to the prepared WWW pages. These access mechanisms include direct access by means of URIs (cf. [1]), direct search mechanisms like full text searches and associative search mechanisms like links. Section 4 gives some ideas on possible enhancements of such a system as well as the environments in which such a system could be used. All of the screendumps in this paper have been taken from one hypertext base that was built from transforming the software manual ``Benutzerhandbuch iXVIEW/SQL'' (Copyright 1991 iXOS Software GmbH).

Next: Document preparation Up: Putting Paper Documents in Previous: Abstract