Long non-coding RNAs (lncRNAs) are increasingly recognized as crucial regulatory molecules in numerous biological processes, yet their identification and functional annotation remain challenging. This paper presents a novel machine-learning pipeline, leveraging natural language processing (NLP) concepts, for distinguishing lncRNA sequences from protein-coding transcripts. We use a k-mer bag-of-words approach to transform RNA sequences into numerical vectors and train a Support Vector Machine (SVM) classifier to predict lncRNA status. Our proposed framework demonstrates high accuracy in classifying validated lncRNA and protein-coding transcripts, offering a promising approach for large-scale lncRNA discovery in diverse organisms. We discuss methodological innovations, statistical insights, and potential biological applications, with an emphasis on how data science drives innovation in genomics research.