| Autor: |
Kirby, G H1 (AUTHOR), Rayner, J D (AUTHOR) |
| Předmět: |
|
| Zdroj: |
Journal of Chemical Information & Computer Sciences. Feb1991, Vol. 31 Issue 1, p153-159. 7p. |
| Abstrakt: |
The problem of attempting to correct by computer software erroneous systematic organic chemical names is addressed. Software has been implemented and tested within the Hull University Nomenclature Translator so that the latter can handle submitted names found to by syntactically or semantically incorrect. A preprocessing stage before parsing removes many minor errors in punctuation. More substantial errors of syntax are detected during parsing and then processed. Errors in punctuation may be corrected by modifying the grammar rules to accept particular cases of erroneous syntax, which is then corrected by semantic code associated with the rules. However, this proved to be practicable for only a few constructions. The more general approach adopted involves a combination of spell-checking suspect fragments to find similarly spelled or sounding fragments, known to the translator, and referencing the grammar tables to determine whether any of these are acceptable in the context of the parsing to that point. Nonserious semantic faults are easily corrected, but faults with the magnitudes of locants and with their associated multiplying terms need interaction with the user. A successful correction rate of 70% was achieved on a small sample of externally generated incorrect names. The conclusion is that it is possible to semiautomatically correct, that is with some user interaction, many of the errors commonly made in systematic nomenclature. Errors that are not satisfactorily handled by the techniques adopted are discussed and suggestions are made for improvement. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Library, Information Science & Technology Abstracts |
| Externí odkaz: |
|
