A Systematic Approach to Key Phrase Extraction for Turkish: Adapting Embedding-Based Models

Key phrase extraction (KPE) is the identification of informative phrases to provide a summary of a given document. Although unsupervised methods are highly valued for their domain independence and generalization capabilities, research in this area has predominantly focused on English texts. Conversely, KPE for agglutinative languages like Turkish remains understudied, primarily due to the challenges posed by complex morphology and a scarcity of high-quality, organized benchmarks. This study explores a systematic approach to adapting existing embedding-based KPE models, which have proven successful in English, for the Turkish language. We establish a systematic heuristic-based extraction methodology to construct candidate pools, specifically analyzing the impact of chunking strategies on model performance and addressing the critical challenge of semantic redundancy. Our study provides a comprehensive analysis of three embedding-based models (AttentionRank, SAMRank, and JointGL) utilizing two newly compiled Turkish datasets from the Information and Health Sciences domains. We further evaluate the adaptability of these architectures under varying constraints, assessing their potential as robust benchmarks for future research. Experimental findings confirm that leveraging Turkish pre-trained language models (PLMs) provides a promising baseline, with AttentionRank demonstrating superior stability among embedding-based approaches. Furthermore, we demonstrate that hybrid methodologies, which fuse embedding-based and statistical approaches, offer high performance, establishing a strong baseline for the field.