Why is PLINTH needed?
PLINTH
was originally designed to support the work of the Building
Directorate of the Scottish Office on the Building Standards Regulations
for Scotland and the associated Technical Standards.
Both hypertext and rule-based systems are well established individually as
tools for work on regulations and standards. Hypertext systems provide
structured access to technical documents, and can maintain and cross-reference
large bodies of text, diagrams and tables. Rule-based expert systems exploit
the inherently rule-oriented nature of regulations. However, both approaches
have limitations:
- cognitive overhead:
- Hypertext has the well known problems of disorientation and digression,
where readers fail to maintain a mental picture of their location and path in
large networks, and get side-tracked or lost.
- expressiveness:
- Expert system rules are limited in expressiveness
compared with texts and diagrams. This is especially true with newer, less
prescriptive regulatory documents where requirements may be expressed using
terms like `adequate means' and `sufficient provision', as well as in precise
weights and measures.
- parsimony:
- Conventional expert systems must reduce all
knowledge to rules, whether or not this is appropriate. For example, although
following the logical dependencies between clauses in regulatory
documents is difficult, the individual clauses are usually expressed
in clear, concise language that is quite easy to understand. This distinction
can be exploited by focusing the system's intelligence on ensuring the
correct and efficient use of documents, and trusting the user's intelligence
to interpret their contents.
- maintenance:
- Creating large rule-bases for regulations, and maintaining
the rules as the regulations are revised, is a time-consuming process
requiring skilled programmers (this is partly a consequence of the
parsimony problem).
- design rationale:
- Neither conventional hypertext nor rule-based systems are well suited to
providing support for handling the design rationale
(or "argumentation") underlying the
technical documents, i.e. the data from research, discussion and
consultation that goes into their production. The ability to structure and
manipulate this information easily is very important for authors.
How does PLINTH solve these problems?