Project 2: Classification		

The group studied the area of classification for a few years, but
no document was agreed upon.

--------------------------------------------------------------------

Oxford 1975: 
============

Special session on environmental constants and classification summarised 
in the minutes, pp 10-12.

SESSION ON ENVIRONMENTAL CONSTANTS AND CLASSIFICATION,
11,15 AM, JANUARY 15

Professor Fosdick introduced discussion of classification schemes. 
There were two basic approaches: some form of fixed, alpha-numeric 
system or the use of keywords.	His own preference was towards the 
latter approach because of its greater flexibility. He believed 
that after an initial period of fluctuation, an agreed, stable set of 
keywords would emerge.	He did not feel able to support the 
approach of the SHARE classification index nor the revision recently 
proposed by Bolstad of Stanford University. Dr Pool gave the 
background to the Stanford reclassification proposal. Pointing out 
the existing software investment using the SHARE index, Dr Ford 
wanted to see a refinement of the index, for use in the short to
medium term. Material could be mechanically processed to an agreed 
set of keywords if and when they emerged. He drew the meeting's 
attention to the suggestions on page 2 of his working paper called 
'Classification of Numerical Algorithms Libraries'.

Dr Reid described how the ordering and keyword systems were combined 
in the Harwell library. Professor Fosdick discussed ways in which 
both approaches could be used to provide several pointers to an 
algorithm. Dr Battiste felt that the classification problem was 
one of access to information, and there were several ways of providing 
that access. The Chairman noted the twin problems of information 
retrieval and information storage within the classification discussion.

Professor Fosdick agreed to assemble comments on classification and to 
collate them for circulation.

Oak Brook 1976:
=============== 

Professor Fosdick reported that J. Lemone (Purdue) and 
J. Bolstad (Stanford) had each been working on classification 
schemes for numerical algorithms. These schemes represented an 
improvement on existing schemes but were not ideal. There was a 
problem of incompatibility with currently used schemes. Professor 
Fosdick asked whether the Working Group would wish to comment on
the schemes. It was agreed to circulate details of Bolstad's scheme 
to the Group and then to reconsider the topic at the next Group 
meeting.

Amsterdam 1977:
===============

Dr F. Krogh had requested the group's opinion of his proposed 
classification scheme, but it was felt that the group could not respond 
sufficiently quickly. Dr Brown reported that he had written to Krogh 
on the matter.

Madison 1982:
=============

Rice pointed out that Hanson is working on the ACM classification
scheme. Schnabel drew attention to the work of Rus Rew at NCAR. Hull
pointed out that the guidelines suggest that the group is obliged to
look at this topic. It was agreed to follow progress but take no
action for the present. [A classification scheme from NBS is almost
finished.]

Soederkoeping 1983:
===================

GAMS classification (see IFIP/WG 2.5 (Soederkoeping-10) 1010).

Einarsson introduced the discussion and the following resolution 
was approved by the group:

IFIP WG 2.5 on Numerical Software has studied the proposed GAMS 
Classification Scheme for Mathematical and Statistical Software, 
and recognises the value of the work done by the group at the 
National Bureau of Standards (Boisvert, Howe, Kahaner). There is 
a strong need for a refined classification scheme within the 
numerical software community, and we find the NBS activity 
pertinent and worthwhile. As the document is not yet complete, 
the working group feels unable to evaluate the work at the 
present time. Schemes of classification are so important in the 
field that their consideration for IFIP approval is a continuing 
interest and concern to WG 2.5.