TextRank Algorithm is an unsupervised graph-based algorithm by Mihalcea with two primary applications, namely in text summarization and keyword extraction of a text document. This study will focus on the enhancement of the TextRank algorithm on the keyword extraction side. This paper introduces an enhanced version of the algorithm wherein the whole text document is preprocessed with a method known as Coreference Resolution, wherein this method normalizes every referenced entity in a text into a single entity. The application of this method to a document text with a longer sequence, outperforms the Precision, Recall, F1-Measure, and the Mean Average Precision metrics of the original algorithm.