Citation Intent Classification Using Word Embedding

Abstract

Citation analysis is an active area of research for various reasons. So far, statistical approaches are mainly used for citation analysis, which does not look into the internal context of the citations. Deep analysis of citation may reveal interesting findings by utilizing deep neural network algorithms. The existing scholarly datasets are best suited for statistical approaches but lack citation context, intent, and section information. Furthermore, the datasets are too small to be used with deep learning approaches. For citation intent analysis, the datasets must have a citation context labeled with different citation intent classes. Most of the datasets either do not have labeled context sentences, or the sample is too small to be generalized. In this study, we critically investigated the available datasets for citation intent and proposed an automated citation intent technique to label the citation context with citation intent. Furthermore, we annotated ten million citation contexts with citation intent from Citation Context Dataset (C2D) dataset with the help of our proposed method. We applied Global Vectors (GloVe), Infersent, and Bidirectional Encoder Representations from Transformers (BERT) word embedding methods and compared their Precision, Recall, and F1 measures. It was found that BERT embedding performs significantly better, having an 89% Precision score. The labeled dataset, which is freely available for research purposes, will enhance the study of citation context analysis. Finally, It can be used as a benchmark dataset for finding the citation motivation and function from in-text citations.

In-text citation’s frequencies-based recommendations of relevant research papers

“…Recently, researchers have analyzed the citing sentences to apprehend the reasons for citation (Muhammad et al, 2021). The same was suggested earlier by Teufel and Kaplan.…”

“…The citation tag is the information in a reference that is used to cite that reference in the body-text of the document, e.g., '' (Muhammad et al, 2021)'', '' (Afzal et al, 2007)''. After manual inspection of the dataset, we found that different authors use different citation tags such as 20, [Afzal, 2009], and (Muhammad et al, 2021). We have considered all these patterns in our implementation by employing regular expressions.…”

Exploiting Contextual Word Embedding of Authorship and Title of Articles for Discovering Citation Intent Classification

“…Roman et al [4] used contextual embedding for capturing the context of citation context. ey used an automated method for annotating the unannotated dataset for citation intent and achieved good precision, recall, and F1 score.…”

“…Table 4 discusses various types of word embeddings along with their strengths and weaknesses. We have selected BERT word embedding as BERT is good in capturing the contextual information from a text and has been used by Roman et al [4] for a similar task. BERTuses the transformers model [35,36] for encoding the vector representation, using encoding-decoding architecture.…”

“…Various attempts have been made in order to understand the reason and intent of a citation. e most recent techniques have used deep networks for reading the citation context of a citation [4][5][6][7].…”