Unsupervised Keyword Extraction Models Using Word-Set Averaging

With the rapid growth of online documents, digital publications, and Q\&A platforms, keywords play a critical role in document classification, information retrieval, and text processing. This paper introduces a new unsupervised keyword-extraction framework that enhances the performance of existing models by incorporating lexical information through word-averaging algorithms. The proposed framework requires no prior training and can be applied to keyword extraction from both individual documents and larger corpora. Experimental evaluations demonstrate that the proposed methods achieve acceptable level of performance and can improve the baseline results.