Sarnasuse mõju viitamise mustritele teaduslikes kogukondades
Date
2011
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Tartu Ülikool
Abstract
Indiviidid kipuvad looma sidemeid pigem nende isikutega, kellega neil on sarnaseid huvisid või muud ühist. Bibliomeetria kontekstis avaldub sama põhimõte viitamise mustrites - artiklites viidatakse muuhulgas ka autorite endi ning nende koostööpartnerite artiklitele. Eelneva valguses on käesoleva töö eesmärgiks projekteerida bibliomeetrilised
indikaatorid, mis võimaldaks meil arvestada sarnasuse mõju teadlaste ning teadusasutuste hindamisel. Konkreetsemalt defineeritakse käesolevas töös sarnasuse alusel kärbitud ja kaalutud versioonid viidete arvu meetrikale. On ilmselge, et sarnasust arvestavad meetrikad annavad kõrgema hinnangu kogukondadele, kus on tava viidata eelkõige teiste
kogukondade autorite artiklitele. Samuti kirjeldatakse ning analüüsitakse antud töös viitamise mustreid, mis tuvastati kirjeldatud meetrikate rakendamisel viidete võrgule kolme
erineva teadlaste kogukonna puhul.
Individuals tend to establish ties in higher rate with individuals that exhibit some kind of affinity than with dissimilar ones. This principle, referred to as homophily, has a important impact in the shape of the social interactions, i.e., topologies of the underlying social networks. In the context of bibliometrics, the effects of homophily can be observed in patterns of citation: references often include a non-negligeable number of self-citations and citations from close collaborators. In light of the above, we aim at designing bibliometric indicators that allow us to modulate the effect of homophily in the ranking induced by metrics. Clearly, homophily-aware metrics would favor communities where citations involve a broader participation. In this thesis, we present homophily-trimmed and homophily-weighted versions of the citation count and report on the patterns of citation uncovered by such metrics over the citation network for three different communities.
Individuals tend to establish ties in higher rate with individuals that exhibit some kind of affinity than with dissimilar ones. This principle, referred to as homophily, has a important impact in the shape of the social interactions, i.e., topologies of the underlying social networks. In the context of bibliometrics, the effects of homophily can be observed in patterns of citation: references often include a non-negligeable number of self-citations and citations from close collaborators. In light of the above, we aim at designing bibliometric indicators that allow us to modulate the effect of homophily in the ranking induced by metrics. Clearly, homophily-aware metrics would favor communities where citations involve a broader participation. In this thesis, we present homophily-trimmed and homophily-weighted versions of the citation count and report on the patterns of citation uncovered by such metrics over the citation network for three different communities.