Both positive and negative citations indicate the interest of researchers to the problem, so all of them are usefull. The share of negativa citations can not be estimated in any way except expert opinion.... Thus it is impossible to use it as scientometric quantitative indicator.

I think negative citations may be an indication of quality research, especially in the case of scientists who are pushing at the frontiers of contemporary consensus. Largely positive citations may be indicative of a researcher who is more interested in playing the academic game than in expanding our knowledge in a particular field.

For me, both positive and negative citations, have the same value. We can never know the influence that a citation, positive or negative, can reach on scientific community. Also, it is difficult to apply criteria to determine if the influence of a citation will be positive or negative, in the fture.

The question is not only about negative and positive citations, but about adding semantics to our citation analysis. Citation counts do not allow to undertake this kind of research. Eugene Garfield (see for instance "can citation indexing be automated?"), among others, have shown that researchers' citation practices do not only relate to approval or disapproval.

There is a very interesting project called Citation Typing Ontology - CITO that tries to develop a markup language to account for these semantics. This paper describes the project:

Cito, the citation typing ontology: http://www.jbiomedsem.com/content/1/S1/S6.

One of the problems is, as Antonio says, the difficulty to agree on whether a citation is positive, negative, additive, etc. But this is definitely an unexplored terrain.

This is an interesting question with many different possible answers. Say for example an article is published to initial good reviews so it garners some positive cites. Then somebody comes out with an article that publishes a refutation. There may then be a number of negative citations. When studying influence as the SIR group does, all of these citations denote some influence on the field. However the nature of that influence is another question. Quite possible after the series of negative articles, there may never be another citation of this article except in a review piece that discusses the different research streams in the field.

As has been previously mentioned, there is the question of distinguishing positive and negative from each other. There is also the citation for citations sake where a paper is cited simply because it refers to a certain subject or where it is cited as an example of a certain type of research. There are even mis-citations where papers are cited as saying something that they in fact did not say.

It seems as though a lot of research needs to be done in this area to determine how to distinguish between types of citations and what if any impact they have on the field.

Since citation counts are supposed to measure impact rather than quality, the positive/negative distinction doesn't seem to make much difference. If citation links are used to indicate thematic similarity, the positive/negative distinction doesn't seem to make mcuh difference, either.

Thus, the most intriguing question seems to be: For which research questions does the positive/negative distinction make a difference?

Anything that is referred in a research output is having the same value. There should not be any negative or positive categorization in citations. When we value add to the existing knowledge we are developing our output by agreeing or differing based on the merits and demerits of any prior publication. So every citation serves the same purpose.

Some rare articles have tried to tackle this issue using some natural langage processing techniques, see for instance : Small H., 2011. Interpreting maps of science using citation context sentiments: a preliminary investigation. Scientometrics, vol. 87, n° 2, pp. 373-88.

If the knowledge system is considered as a process of communications exchange, then any citation relationship indicates communication (Pulgarin, Moskaleva, Kumar). If the network of connections between papers and citations is seen as communicating positive or negative feedback (or open/closed "gates") then this is a feature of the system that should be measured (Chavarro, Cuellar, Glaser). Imagine a network in which we see one article gets a larger-than-expected number of 'negative' citations--what can we hypothesize as going on here? fraud? crisis (in the Kuhnian sense)? error? A citation network that shows green lines as positive and red lines as negative would reveal some amazing underlying dynamics about exchange, path dependence, and knowledge flows. With the advent of full-text analysis, we should be able to parse these differences and develop testable hypotheses about what communications dynamics are happening in the citation network.

Very interesting topic. I think the definition of Impact, as citations either positive or negative, is a function of the difficulty of automatically associating citations with their original works. Its only relatively recently that this has been possible with any degree of reliability. The next development in my view is ORCID [ http:www.orcid.org ] (or another system like it) to correctly identify authors to improve the precision and reliability of citations.

A reliable citation system would allow us to raises the question of the intention of citations, to add the next level or granularity. I could see authors being asked to apply a simplified mark-up to their manuscripts to express intention in a form that can be read by a computer. In the end it must be the authors intention rather than one applied by a researcher that counts, if only to allow such a system to be scaleable.

Surely it depends on the contexts in which a citation is used. The same citation could be reflected as negative in one context and positive in another. It depends on how the researcher uses the citation and what he wants to achieve with using it.

Even though it is sometimes indeed impossible to tell whether a particular citation is positive or negative, there is at least one kind of citations which are unequivocally negative, namely when the result from the cited paper is convincingly proven to be wrong/false.

Dear All,

The value or the “direction” of citations is much more important as one can perceive it reading a study.

I feel that the negative adjective is not a suitable term because the notion of disagreement or uncertainty may have various stages. I remember a biologically exaggerated and inaccurate paper which collected over 100 citations but each of them wanted to disprove its statements.

There are citations of politeness which have nothing to do with the real message or merit of a study. I have heard about institutions where it is officially compulsory to cite the publications of all other fellows of this organization. Even some big potatoes allow to submit a publication at certain “conferences” and “journals” if the future participants or authors cite the publications of these great men.

Thus, it would be useful to have a mechanism to somehow signal the directions of the citations.

For an example of the use of semantics in citation analysis I recommend reading Lowell L. Hargens (2004) paper 'What is Mertonian Sociology of Science'? (scientometrics, 60(1)), in which the author analyses how Merton has been cited by different communities. A complete school of thought can be seen that cited Merton in order to establish their divergence from the author.