Е. Ф. Бычкова The authors examine the possibility of computerized abstracting of publications based on generative abstracting models. They review the approaches toward automatic control including that of neural networks, characterize popular software environments, and discuss their advantages and disadvantages for computerized abstracting. The problems under discussion are relevant as the technology of computerized abstracting increases the accessibility of publications, in particular, those out of the open access while decreases bibliographic processing costs. The authors insist that augmented bibliographic record supplied with annotation or abstract is very important for providing information on new environmental technologies. At the same time, the appropriate abstract consumes intellectual efforts and time of competent professionals. The database “Ecology: Science and Technologies” was chosen as the study object; the database comprises publications on implementation of new environmental-friendly and resource-saving technologies. The authors conclude that computerized abstracting as opposed to similar manual process requires no specialists highly qualified in the discipline of the document being processed while the quality of the abstracts is rather high even when the standard datasets are used.
Научная статья 
датасет