MetaMap is a widely used program that maps concepts from biomedical text to the Unified Medical Language System (UMLS) Metathesaurus. Developed to improve biomedical text retrieval, particularly for MEDLINE/PubMed citations, MetaMap links biomedical literature to knowledge stored in the Metathesaurus, including synonym relationships. It was initially guided by linguistic principles, providing a flexible architecture for exploring mapping strategies. The system processes input text through lexical/syntactic analysis, including tokenization, part-of-speech tagging, and lexical lookup in the SPECIALIST lexicon. It then identifies phrases and performs variant generation, candidate identification, mapping construction, and word sense disambiguation (WSD). The evaluation of candidates and mappings is based on four linguistic measures: centrality, variation, coverage, and cohesiveness. MetaMap is highly configurable, allowing users to choose data, output, and processing options. It supports various output formats, including human-readable, machine output (MMO), XML, and colorized output. It also offers processing options such as term processing, browse mode, negation detection, and enhanced WSD. MetaMap's strengths include its thoroughness, linguistic principles, and ability to handle complex mappings. However, it has limitations, such as being English-centric, relatively slow, and less accurate in ambiguous cases. MetaMap has been used since 1994 and is available via web access, Java implementation (MMTx), and API. It has been applied to various tasks beyond retrieval, including text mining, classification, question answering, and knowledge discovery. Research involving MetaMap has extended beyond the National Library of Medicine (NLM). Recent developments include chemical name recognition and improved WSD. MetaMap's algorithm has been optimized for efficiency, but it still faces challenges in real-time processing due to its computational intensity. Future improvements aim to enhance its accuracy and adaptability to different domains. MetaMap's linguistic foundation is attributed to Thomas C Rindflesch and Allen C Browne. It is supported by the National Institutes of Health and the National Library of Medicine.MetaMap is a widely used program that maps concepts from biomedical text to the Unified Medical Language System (UMLS) Metathesaurus. Developed to improve biomedical text retrieval, particularly for MEDLINE/PubMed citations, MetaMap links biomedical literature to knowledge stored in the Metathesaurus, including synonym relationships. It was initially guided by linguistic principles, providing a flexible architecture for exploring mapping strategies. The system processes input text through lexical/syntactic analysis, including tokenization, part-of-speech tagging, and lexical lookup in the SPECIALIST lexicon. It then identifies phrases and performs variant generation, candidate identification, mapping construction, and word sense disambiguation (WSD). The evaluation of candidates and mappings is based on four linguistic measures: centrality, variation, coverage, and cohesiveness. MetaMap is highly configurable, allowing users to choose data, output, and processing options. It supports various output formats, including human-readable, machine output (MMO), XML, and colorized output. It also offers processing options such as term processing, browse mode, negation detection, and enhanced WSD. MetaMap's strengths include its thoroughness, linguistic principles, and ability to handle complex mappings. However, it has limitations, such as being English-centric, relatively slow, and less accurate in ambiguous cases. MetaMap has been used since 1994 and is available via web access, Java implementation (MMTx), and API. It has been applied to various tasks beyond retrieval, including text mining, classification, question answering, and knowledge discovery. Research involving MetaMap has extended beyond the National Library of Medicine (NLM). Recent developments include chemical name recognition and improved WSD. MetaMap's algorithm has been optimized for efficiency, but it still faces challenges in real-time processing due to its computational intensity. Future improvements aim to enhance its accuracy and adaptability to different domains. MetaMap's linguistic foundation is attributed to Thomas C Rindflesch and Allen C Browne. It is supported by the National Institutes of Health and the National Library of Medicine.