Start Date: 2008-05-01

End Date: 2011-04-30


The LiquidPub project proposes a paradigm shift in the way scientific knowledge is created, disseminated, evaluated and maintained. This shift is enabled by the notion of Liquid Publications, which are evolutionary, collaborative, and composable scientific contributions. Many Liquid Publication concepts are based on a parallel between scientific knowledge artifacts and software artifacts, and hence on lessons learned in (agile, collaborative, open source) software development, as well as on lessons learned from Web 2.0 in terms of collaborative evaluation of knowledge artifacts.

The adoption of the new paradigm for scientific publications will succeed only if we are able to guarantee the fairness of the system. This implies ensuring that the quality attributed to research work is adjusted to the reality (avoiding biased opinions) and that each participant/researcher (from authors to editors) receives the right credit according to its participation in the work.

IIIA’s focus is on computing the reputation of both research and researchers. The basic idea is that researchers may leave opinions about existing research. These could either be direct opinions (such as a review) or indirect ones (such as a citation in the case of research, which may be assumed to represent a positive opinion, or existing metrics such as the h-index in the case of authors).

For computing the reputation of research, we propose a propagation mechanism for opinions in structural graphs and an aggregation mechanism that handles the aggregation of all opinions based on the reliability of each (which is in fact the reliability of the opinion holder, or the reviewer).

For computing the reputation of a researcher, the computation is essentially depends on the role in question. For instance, authors reputation is different than that of reviewers. Authors’ reputation is based on that of their research, taking into consideration how their individual research work is linked together in the structural graph. On the other hand, reviewers’ reputation is based on their experience in the field, the history of their previous reviews (e.g. how close were they to the group opinion, how convincing were they, etc.), social network relations and potential bias, amongst other things.

In all cases, attack detection and avoidance is a crucial issue and a main point of investigation. During the design of the model it is necessary to take into account the specific kind of attacks it can suffer in the environment where it is intended to be used. Collusion, bad mouthing or discrimination for non scientific reasons are examples of attacks to be considered. It is important to provide mechanisms to detect the attacks and be sure the system is able to avoid (or at least minimize) their effects. For the detection, we plan to identify the different patterns that typify each attack so that the model can detect that there is something anomalous going on.

More information