Improving Citation Recommendation Accuracy Using an SVM-Based Model

Abstract

This work presents an SVM-based citation recommendation system, which will path the way for easy identification of relevant research papers. The methodology of the proposed system starts with preprocessing of the input dataset. Here, Using stochastic matching pattern scheme the function words were removed in order to construct a structured library. Canonicalization of documents is achieved through the application of a probabilistic text normalization algorithm. Normalization of data has been done, and now it is ready for further processing.
classified by employing the Support Vector Machine (SVM) algorithm. This algorithm is particularly oriented towards high-dimensional data and effective in making separations of intricate patterns. In the end, the system ranks citations according to recommendations and measures its performance with evaluations like MAP, MRR, precision, recall, and F1-score. Resultant evidence illustrates that this SVM-based system surpasses a number of conventional models—thereby proving its capability to enhance the citation recommendation process.