Disambiguating publication venue titles using association rules

Abstract

Research agencies in several countries evaluate the impact of scientific publications of researcher groups to define their investments, and one of the main used metrics is the quality of the publication venues where their works were published. Several bibliometric indexes have been formulated by measuring the quality of a publication venue. However, given a set of citations extracted, for example, from curricula vitae of a researcher group, to effectively use bibliometric indexes to evaluate their quality it is necessary to identify correctly the publication venue title of each citation. This task is not easy, since there are not unique identifiers for publication venues. Frequently, citations contain abbreviated forms and acronyms, publication venues share similar titles, sometimes they change their titles, divide or merge, creating new ones. Traditional digital libraries deal with this problem by creating Authority Files. In this work, we present a twofold contribution: (i) the creation of a Computer Science publication venue authority file and (ii) the proposal of a method that uses association rules to disambiguate publication venue titles originated from citations. The disambiguator is a supervised learning method that uses the authority file to train a classifier, whose generated model is a set of association rules to identify publication venues. Experiments show that our method obtains better results than three state of art baselines.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 9 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.