چکیده:
Co-citation forms a relational document network. Co-citation-based measures are found to be effective in retrieving relevant documents. However, they are far from ideal and need further enhancements. Co-opinion concept was proposed and tested in previous research and found to be effective in retrieving relevant documents. The present study endeavors to explore the correlation between opinion (dis)similarity measures and the traditional co-citation-based ones including Citation Proximity Index (CPI), co-citedness and co-citation context similarity. The results show significant, though weak to medium, correlations between the variables. The correlations are direct for co-opinion measure, while being inverse for the opinion distance. Accordingly, the two groups of measures are revealed to represent some similar aspects of the document relation. Moreover, the weakness of the correlations implies that there are different dimensions represented by the two groups.
خلاصه ماشینی:
Co- citation is among the citation-based measures successfully used in retrieving scientific papers (Janssens and Gwinn, 2015; Yoon, Kim, Park, 2016; Eto, 2013; Zhao, 2014; Egghe & Rousseau, 1990; Bichteler & Eaton, 1980; Badran, 1984).
The merit of Co-CCs lies in the fact that they carry not only the co-citers’ objective reports of different features of their co-cited articles, but also their opinions about them (Sendhilkumar, Elakkiya & Mahalakshmi, 2013; Ritchie, Robertson & Teufel, 2008; Agarwal Choubey & Yu, 2010; Doslu & Bingol, 2016).
To answer the question, we conducted a research on the sample of co-cited medical papers studied in our previous work to test the correlation between the co-opinion concept, on the one hand and three traditional co-citation-based measure, including co-citedness, Citation Proximity Index (CPI) and Co-Citation context (Co-CC) textual similarity.
CC opinion mining or sentiment analysis is among those innovative techniques based on content analysis of CCs. Negative and positive citation opinions can be used to evaluate cited papers and delve their actual impact (Abu-Jbara, Ezra, & Radev, 2013; Hernandez-Alvarez & Gomez, 2016; Yu, 2013; Athar & Teufel, 2012; Small, 2011).
there is not yet enough standard infrastructures to ensure the reliability of the results of opinion mining of scientific papers (Hernandez-Alvarez & Gomez, 2016), a growing number of studies have employed the techniques (Teufel, Siddharthan & Tidhar, 2006; Athar, 2011; 2014; Jochim & Schiitze, 2012; Abu-Jbara et al.