About reasonable number of rankings in preference profile when measuring quality; International Symposium on Measurement and Quality Control, (ISMQC 2007)
| Источник: | International Symposium on Measurement and Quality Control, (ISMQC 2007).— 2007.— [P. 331-334] |
|---|---|
| Главный автор: | |
| Автор-организация: | |
| Примечания: | Title screen To plan the quality measuring in the form of consensus relation determination for the given m weak order relations (rankings) it is necessary to know a reasonable number of the rankings. If a ranking is produced by an expert then the number of rankings is equal to number of experts. It is proposed to estimate the expert number using simple probabilistic Bernoulli model, where m experts reveal defects (demerits) of an object. The model assumes that the more the number of an expert group participants, the less the probability of a new defect revealing. Based on this assumption, the probability decrease have been evaluated and graphically presented. The investigations allow to suppose that the number of rankings in preference profile can be from 4 to 10 for typical applications. |
| Язык: | английский |
| Опубликовано: |
2007
|
| Предметы: | |
| Online-ссылка: | https://www.imeko.org/publications/tc14-2007/IMEKO-TC14-2007-61.pdf |
| Формат: | Электронный ресурс Статья |
| Запись в KOHA: | https://koha.lib.tpu.ru/cgi-bin/koha/opac-detail.pl?biblionumber=666469 |
| Примечания: | Title screen To plan the quality measuring in the form of consensus relation determination for the given m weak order relations (rankings) it is necessary to know a reasonable number of the rankings. If a ranking is produced by an expert then the number of rankings is equal to number of experts. It is proposed to estimate the expert number using simple probabilistic Bernoulli model, where m experts reveal defects (demerits) of an object. The model assumes that the more the number of an expert group participants, the less the probability of a new defect revealing. Based on this assumption, the probability decrease have been evaluated and graphically presented. The investigations allow to suppose that the number of rankings in preference profile can be from 4 to 10 for typical applications. |
|---|