U.S. DEPARTMENT OF COMMERCE
National Institute of Standards and Technology
National Engineering Laboratory
Center for Manufacturing Engineering
Factory Automation Systems Division
PDES Testbed Group
Gaithersburg, MD. 20899
September 1988
Sanford Ressler
National Institute of Standards and Technology
September 1988
Faced with implementing a standard an individual must treat the standards document as the specification against which to compare their interpretation of that document. Typically a terse concise document which seeks to eliminate ambiguity of interpretation, the standard is often little help for the implementor.
No approval or endorsement of any products by the NIST is intended or implied.
A hyperstandard is proposed which uses several existing computing technologies (hypertext[3,16], CD-ROM[14], document processing) combined in a novel way to address a number of problems in the development and implementation of standards. A hyperstandard system is the embodiment of a particular standard, related documents, images, sounds and programs in the context of a particular physically implemented system. The term "Hyper" is derived from hypertext systems. A term coined by Ted Nelson[17]. A hypertext document consists of a collection of "interconnected writings." The standard's implementor or user could browse through the document and upon encountering some point of interest could follow that point and travel along its connections to find other (cross referenced) points in that collection of writings. Rather than limiting these "documents" to pure text we may, given present technology, also incorporate graphics, audio and computer programs into the content of the document creating what is known as a hypermedia document. (The words hypertext and hypermedia are used interchangeably in this paper.) The concept of a hyperstandard system is explored primarily in the context of a particular ongoing standards effort known as PDES[10] (Product Data Exchange Specification).
First let's examine the standards development process taking place for PDES. The standard is being developed by the IGES/PDES Voluntary Organization which is open to any interested individuals. The overall organization has been broken into a number of committees. Each committee works on separate portions of the standard and produces working documents about their domains. The problems increase when each of the committees' work must be integrated with the others into a uniform consistent whole. Overlap in the committees must be reduced with the goal being a standard that is as unambiguous as possible. The following diagram illustrates a simplified view of the development process.
One question which needs to be asked is: Where are there opportunities for the application of systems which can aid in this development process?
Definition is perhaps the lengthiest phase where experts in various domains draw upon their knowledge and experience to identify and create relevant definitions. A mechanism to allow different committees access to documents produced in the definition process and to conveniently look at unresolved issues (see figure 5) would help in communicating the rationale behind the decision making process. Quite often in the course of the development of PDES a great deal of research and thoughtful reasoning which has gone into a decision simply disappears. This information, while not part of the standards document, would be useful, if available as supplementary material.
The reviewing process has been a particularly troublesome area in PDES development. Many working "informal" documents are produced representing individual opinions and positions. Reliable access to the latest versions of drafts are often a problem in voluntary standards development where funding is very limited and resources are stretched to the limit. Determining what documents exist, their status and who "owns" them (see figure 11) would be extremely useful.
Integration is an area which is also quite problematic. The information from the various committees must be tied together into a single coherent document. Ambiguities must be identified and eliminated. When problems occur there must be a systematized mechanism for feedback to the definition phase. Software which aids in identifying duplicate definitions and which produces various cross-references and dictionaries could help. Figure 8 illustrates a display where a user is able to look at the results of the two information modeling methodologies (EXPRESS[11] and IDEF1x[9] as text) being used in PDES to define entities. This information comes from different activities and sources, yet can be placed together and presented to a user in a coherent manner.
Solutions to the many complex development problems are certainly not presented here, however one can start to think in terms which may lead to a framework for the development of standards that takes advantage of existing hypertext technology. Currently the entire development process is done "by hand." One clear requirement is the automated "computer sensible" form of all documents, including illustrations, source code and prototype implementations when available. This requirement clearly is expensive and difficult to attain with the variety of systems being used.
Some standards documents are very obtuse and difficult to comprehend. Everyone can agree that PDES is very large and complex. Hypertext can be used to aid the process of learning about the standard. The process of educating potential implementors and the user community is a formidable task. There is no magic, of course. A poorly written standard will remain poorly written when put into a hypertext system. However one can try to reorganize the documents into more accessible and usable information. Placement of the information into an on-line form also allows for information queries and for an intermixing of graphics and text in a number of combinations. In addition, given this method of distribution, one can allow users to add commentary (see figure 9), first stored on the user's local disk and later incorporated into subsequent releases of the CD-ROM, creating a permanent record.
Documentation about the kinds of implementations and methodologies for creating implementations would be extremely useful in the context of the specification itself.
A hyperstandard system can also serve the role of software librarian by embedding public domain tools and making them accessible to implementors. In the case of PDES a number of software tools to parse the various types of representations (EXPRESS and IDEF1x) do exist and are a good starting point for implementations. The difficult area of validation and testing can also be aided by packaging some of the testing tools with the specification. Pre-packaged "officially approved and tested" data structures and functions in a variety of languages may also be included as an aid to producing implementations. Rapid, convenient access to the specification will also aid in achieving high quality implementations.
PDES is chosen because it offers a chance to examine an evolving standard which is in great demand by industry. PDES is an extremely large, complex and ambitious standards project which could greatly benefit from this technology. Industry participants are being presented with a daunting array of implementation possibilities and problems. A PDES hyperstandard would gather together not only the official standards documents but related research papers, and public domain software tools which will aid implementors in their tasks.
It must be stated very clearly that the way one might go about creating this collection of "documents" is not at all clear or well defined. No methodology exists to guide one in the creation of such a collection of multi-media information. No standards for hypertext exist, as yet. The best one can do is look at the existing systems in the literature and learn from others. The work done at Brown[19], CMU[1], Xerox[5,7,8], and Tektronix[4] is vital and must be studied carefully.
The following figures represent a sample of how a user browses through a hyperstandard version of PDES:
As an example of how one might do this let's again look at PDES. Let's also assume that our target system is going to be a Macintosh with HyperCard. The specification itself exists as TeX (TeX is a text formatting language developed by Donald Knuth) documents, which are ASCII files that may be parsed. One way of getting documents into HyperCard would be to preprocess the documents on a system which has good parsing tools, such as a UNIX based machine with YACC[12] and Lex[15]. Parsing would generate a set of reformatted documents. A simple HyperCard script could be written to format the new documents into a variety of cards.
HyperCard's programming language is called HyperTalk and while it has some interesting features it lacks good tools for parsing. The next problem is formatting the document into meaningful and useful cards. A reasonable approach is to create named template cards by hand on the Macintosh and refer to those names on the UNIX machine. Figure 19 illustrates the overall process.
This technique works for the EXPRESS specification document which is a highly structured document. Informal documents and illustrations are still a problem. The IDEF1x diagrams of PDES can be represented textually by a language called SML. An SML parser could generate a series of drawing commands interpretable by HyperTalk which would automatically create the drawing. However one can only see small portions of the IDEF1x diagrams on a single card. A partial solution to this problem is to display only context diagrams. These diagrams show relationships for only one entity at a time.
Other diagrams and illustrations would be scanned and represented in a computer sensible form. Given a raster type file of an illustration one could get it into HyperCard by creating some simple translation programs. Another approach is to provide a print button which would cause the diagrams to be printed out rather than displayed. The CD-ROM in this case would function as an image library.
Another problem in creating a satisfying hyperstandard system is generating the links. For example, a link between an IDEF diagram of an entity and its definition should be created automatically. This could be accomplished by the preprocessing technique discussed previously and by using consistent naming schemes. An IDEF1x rectangle has text associated with it and that rectangle could be made active by creating a button over it. The link to the EXPRESS definition would be by a reference to the correct HyperCard stack and card.
This entire scheme, while not particularly elegant, would work for most of the PDES standard. In addition, the information obtained directly from the standard and preprocessors could be used to generate useful statistical information, cross references, and indices. Clearly most of the work involved takes place in the preprocessing phase. This has the advantage of providing a platform, at a later time, for moving towards an application independent implementation. Defining the functionality of hypertext/media systems is a ripe domain for standards. SGML[18] (Standard Graphical Markup Language) is an ISO standard which can be used to define the structure of documents. It has great potential for use in mapping documents into a hypertext structure. This would aid in porting a hyperstandard system to a variety of hypertext systems.
A specific prototypical instance of a standard in development (PDES) has been implemented and demonstrated to illustrate the viability of a hyperstandard system.