This paper introduces TextRank, a graph-based ranking model for text processing, and demonstrates its effectiveness in natural language applications. The model is applied to two tasks: unsupervised keyword and sentence extraction. The results show that TextRank performs competitively with state-of-the-art systems. TextRank is based on graph-based ranking algorithms, such as PageRank and HITS, which determine the importance of a vertex in a graph by recursively computing global information. In the context of text processing, TextRank constructs a graph from natural language texts, where vertices represent text units (e.g., words, sentences) and edges represent relationships between them. The importance of each text unit is determined by the number and strength of connections it has with other units. For keyword extraction, TextRank uses a co-occurrence relation between words, where words that appear together within a window are connected. The algorithm then ranks the words based on their importance, which is determined by the number and strength of their connections. The results show that TextRank outperforms other methods in terms of precision and F-measure. For sentence extraction, TextRank constructs a graph where each sentence is a vertex, and edges represent the similarity between sentences. The similarity is measured based on the overlap of words between sentences. The algorithm then ranks the sentences based on their importance, which is determined by the number and strength of their connections. The results show that TextRank produces summaries that are competitive with other systems. TextRank is fully unsupervised and does not require training data or linguistic knowledge. It is therefore highly portable to other domains, languages, and text types. The model is effective because it considers the global structure of the text, rather than just local information, allowing it to identify important text units based on their connections to other units. This makes TextRank a powerful tool for text processing tasks such as keyword and sentence extraction.This paper introduces TextRank, a graph-based ranking model for text processing, and demonstrates its effectiveness in natural language applications. The model is applied to two tasks: unsupervised keyword and sentence extraction. The results show that TextRank performs competitively with state-of-the-art systems. TextRank is based on graph-based ranking algorithms, such as PageRank and HITS, which determine the importance of a vertex in a graph by recursively computing global information. In the context of text processing, TextRank constructs a graph from natural language texts, where vertices represent text units (e.g., words, sentences) and edges represent relationships between them. The importance of each text unit is determined by the number and strength of connections it has with other units. For keyword extraction, TextRank uses a co-occurrence relation between words, where words that appear together within a window are connected. The algorithm then ranks the words based on their importance, which is determined by the number and strength of their connections. The results show that TextRank outperforms other methods in terms of precision and F-measure. For sentence extraction, TextRank constructs a graph where each sentence is a vertex, and edges represent the similarity between sentences. The similarity is measured based on the overlap of words between sentences. The algorithm then ranks the sentences based on their importance, which is determined by the number and strength of their connections. The results show that TextRank produces summaries that are competitive with other systems. TextRank is fully unsupervised and does not require training data or linguistic knowledge. It is therefore highly portable to other domains, languages, and text types. The model is effective because it considers the global structure of the text, rather than just local information, allowing it to identify important text units based on their connections to other units. This makes TextRank a powerful tool for text processing tasks such as keyword and sentence extraction.