Software Classification



next up previous
Next: Search Interfaces Up: Indexing and Searching Previous: Data Models

Software Classification

 

A number of studies have shown that proper classification of software objects contributes to effective location of the objects by potential reusers [22][31][15]. Classification is carried out by assigning codes and/or keywords from a classification scheme or thesaurus. Classifications and thesauri developed for indexing documents, such as the INSPEC classification [1] and INSPEC thesaurus [2], are inadequate for indexing software objects. Firstly, these tools cover a broad range of topics and cover software-related topics only superficially. Thus, they do not allow the user to discriminate finely enough among the available software objects. Secondly, effective classification of software objects requires that the function of the object be indexed [10]. Because documents are not used as software is, terms related to function have not generally been included in thesauri developed for document indexing.

A classification scheme that has been developed specifically for use in indexing mathematical software is the GAMS hierarchy [9]. The GAMS scheme has been widely adopted by network-accessible repositories such as Netlib and by commercial mathematical software libraries such as NAG and IMSL. A successor to GAMS is currently under development. The new scheme will refine areas needing better discrimination and will add new categories to encompass recent developments in numerical algorithms. The new scheme will also be reorganized so as to be less cumbersome.

An HPCC software thesaurus is under development as part of the NHSE. This thesaurus uses GAMS categories for mathematical software but defines new terms for other areas of high performance computing. Some of these terms are drawn from an existing HPCC glossary [25] and from a book that gives an overview of parallel computing [21]. Terms from the INSPEC thesaurus [2] are being used for application areas. The thesaurus has a faceted structure with facets for application area, problem area, function, algorithm, and target architecture. The thesaurus will be made available on-line in hypertext form to assist with searching the NHSE, similar to [30]. Hypertext links from terms to scope and definition notes will also be provided.



next up previous
Next: Search Interfaces Up: Indexing and Searching Previous: Data Models



Jack Dongarra
Thu Feb 23 09:42:15 EST 1995