ÖйúAƬ

Times ÖйúAƬ reports that Eric Anicich recommends classifying a paper according to the type of citation it is, and in particular, to try to eliminate negative citations from citation counts for evaluating the quality of research (¡°Six citation categories could help neutralise the negatives¡±, News, 11 December). The idea is not a new one. In the 1970s, there were several attempts to develop and apply such classifications, and such research continues to this day. The results of these studies, mainly focused in the science and medicine fields, showed that only a tiny proportion, typically about 2 per cent, of citations are negative. Mr Anicich recommends that the authors themselves assign the reasons for the citation at the time they write their article. An alternative approach is to relieve them of this duty, but instead for others to use automated text and data mining techniques, already applied to sentiment analysis, to assess the reasons for the citations.

Charles Oppenheim
Former professor of information science Loughborough University