Article citation study: Context enhanced citation sentiment detection
This work addresses citation sentiment detection for scientometric analysis, presenting an incremental improvement with new datasets and feature engineering.
The paper tackled citation sentiment analysis by creating eight manually annotated datasets, including three with full citation context, and proposed an ensembled feature method combining word embeddings, POS tags, and dependency relations for deep learning classification. Results showed deep learning performed better with larger samples, support vector machines with smaller ones, and context-based samples were more effective than context-less ones.
Citation sentimet analysis is one of the little studied tasks for scientometric analysis. For citation analysis, we developed eight datasets comprising citation sentences, which are manually annotated by us into three sentiment polarities viz. positive, negative, and neutral. Among eight datasets, three were developed by considering the whole context of citations. Furthermore, we proposed an ensembled feature engineering method comprising word embeddings obtained for texts, parts-of-speech tags, and dependency relationships together. Ensembled features were considered as input to deep learning based approaches for citation sentiment classification, which is in turn compared with Bag-of-Words approach. Experimental results demonstrate that deep learning is useful for higher number of samples, whereas support vector machine is the winner for smaller number of samples. Moreover, context-based samples are proved to be more effective than context-less samples for citation sentiment analysis.