Classifying documents with link-based bibliometric measures

Abstract

Automatic document classification can be used to organize documents in a digital library, construct on-line directories, improve the precision of web searching, or help the interactions between user and search engines. In this paper we explore how linkage information inherent to different document collections can be used to enhance the effectiveness of classification algorithms. We have experimented with three link-based bibliometric measures, co-citation, bibliographic coupling and Amsler, on three different document collections: a digital library of computer science papers, a web directory and an on-line encyclopedia. Results show that both hyperlink and citation information can be used to learn reliable and effective classifiers based on a kNN classifier. In one of the test collections used, we obtained improvements of up to 69.8% of macro-averaged F 1 over the traditional text-based kNN classifier, considered as the baseline measure in our experiments. We also present alternative ways of combining bibliometric based classifiers with text based classifiers. Finally, we conducted studies to analyze the situation in which the bibliometric-based classifiers failed and show that in such cases it is hard to reach consensus regarding the correct classes, even for human judges.

Information and data management at PUC-rio and UFMG

“…In [34,35], we focused on web directories, producing high levels of classification effectiveness (around 90%) which is good for a noisy environment such as the Web. Similar results could be obtained in other scenarios where hyperlinked information also existed such as those provided by citations among scientific documents and references among encyclopedia articles [53,54].…”

Hybrid self-optimized clustering model based on citation links and textual features to detect research topics

“…The advantages of the proposed model are reflected in three main aspects: (a) it can make citation-based and text-based methods complementary to improve the efficiency and overcome some limitations of these two methods [ 10 ]; (b) considering both the common out- and in-links between papers can express the bibliometric relations between papers more accurately [ 24 ]; and (c) adding the topological feature into the statistical feature of terms can improve the performance of existing textual statistical methods [ 27 ]. Moreover, the Louvain method is an efficient and self-optimal clustering method based on modularity; thus, it does not require the number of clusters to be set beforehand.…”

“…The difference between this study and theirs is that the link strength is calculated based on the Amsler network rather than on the bibliographic coupling network. Although Glänzel and Czerwon [ 26 ] showed that the bibliographic coupling network has several advantages compared to the co-citation network, some studies [ 23 , 24 ] also verified that different distributions of citation links have different influences on the citation-based clustering. Therefore, to weaken the influence of the different distributions of citation links, both the common out- and in-links between pairs of papers are involved in the computation of citation-based similarity in this study.…”

“…However, there are three issues to be noted. First, Calado et al [ 23 ] and Couto et al [ 24 ] found that different distributions of citation links in direct citation, co-citation, and bibliographic coupling networks can influence the clustering performance. Second, previous studies usually only took textual statistical features based on TF-IDF as the text-based similarity, whereas Chen and Xiao [ 15 ] pointed out that only taking the term frequency as the importance of terms is not sufficient.…”

A New Metric for Measuring Relatedness of ScientificPapers Based on Non-Textual Features

“…McNee et al [15] use the citation graph attributes like co-citation for finding similar papers in their proposed collaborative filtering algorithm to recommend citations for an input text and Couto et al [16] use these bibliometrics measures to classify documents.…”