Graph-Based Arabic Key-phrases Extraction
Dana Halabi, Arafat Awajan · 2019
This paper proposes Arabic key-phrases extraction using graph representation. The proposed approach based on representing the text of an individual document as a graph, where the nodes within the graph hold the words' stem and the edges represent the co-occurrence relation between stems in specific window size. After building the graph, graph-based centrality measures were used in ranking the nodes according to their importance. Then the ranking results are sorted decently to determine the top n nodes. The stems that are represented by the top n nodes will be considered as the key-stems of the individual document. The performance of our work is measured using the three accuracy measures: Precision, Recall, and F-Measure. The obtained result reached 54%, 82% and 64% for Precision, Recall, and F-measure respectively.