To insert individual citation into a bibliography in a word-processor,
select your preferred citation style below and drag-and-drop it into the document.
In Proceedings of the twelfth international conference on Information and knowledge management (2003), pp. 556-559, doi:10.1145/956863.956972 Key: citeulike:595771
Formatted Citation
Show HTML
Likes
(beta)
This copy of the article hasn't been liked by anyone yet.
Given a snapshot of a social network, can we infer which new interactions among its members are likely to occur in the near future? We formalize this question as the link prediction problem, and develop approaches to link prediction based on measures the "proximity" of nodes in a network. Experiments on large co-authorship networks suggest that information about future interactions can be extracted from network topology alone, and that fairly subtle measures for detecting node proximity can outperform more direct measures.
The authors describe various network measures by which the emergence of a publication between two authors can be predicted; there is by the way a full journal version from 2007 that gives more details (Journal of the American Society for Information Science and Technology, 58(7), pp. 1019-1031, 2007). Basically the authors show that by computing one of these measures (like, e.g., the number of common co-authors or various measures based on random walks) and ranking the results by value, they can predict up to 16% of all emerging edges. This procedure can be framed as an 'unsupervised learning' method and they compare the predictive power of all measures.
They also introduce an interesting quality measure which allows easy comparison between different networks, namely the positive predictive value (PPV) in the set of the first n predictions, where n is the (known) number of edges to emerge. As a quality measure, this value has very interesting properties (not actually named in the article): the PPV in the first n ranks is in this case identical to the sensitivity. It also allows to directly compute the specificity and it can be shown that both, sensitivity and specificity increase if PPV in the first n ranks increases. Thus, comparison between different methods on possibly different networks is easily achieved. The article is described as 'state of the art' for unsupervised learning in a 2010 KDD conference proceedings article ("New Perspectives and Methods in Link Prediction") by Lichtenwalder et al. where supervised learning methods are compared.
CiteULike organises scholarly (or academic) papers or literature and provides bibliographic
(which means it makes bibliographies) for universities and higher education establishments.
It helps undergraduates and postgraduates. People studying for PhDs or in postdoctoral (postdoc) positions.
The service is similar in scope to EndNote or RefWorks or any other reference manager
like BibTeX, but it is a social bookmarking service for scientists and humanities researchers.
There is 1 review 
- 2011-01-07 10:51:15
Find related articles from these CiteULike users
