方法对比

	组成分析	实体链接	命名实体识别 (NER)
领域	文本挖掘	文本挖掘	文本挖掘
方法族	Process / pipeline	Process / pipeline	Process / pipeline
起源年份	2003	2008	—
提出者	Michael Collins (statistical models, 2003)	Milne & Witten	—
类型	NLP syntactic-analysis task	NLP knowledge-base grounding task	NLP sequence-labelling task
开创性文献	Collins, M. (2003). Head-Driven Statistical Models for Natural Language Parsing. Computational Linguistics, 29(4), 589-637.	Milne, D. & Witten, I.H. (2008). Learning to Link with Wikipedia. CIKM (Proceedings of the 17th ACM Conference on Information and Knowledge Management).	Nadeau, D. & Sekine, S. (2007). A survey of named entity recognition. Lingvisticae Investigationes.
别名	phrase-structure parsing, constituent parsing, Kurucu Öbek Ayrıştırma (Constituency Parsing)	named entity disambiguation, entity disambiguation, entity resolution to knowledge base, Varlık Bağlama (Entity Linking)	NER, entity tagging, Adlandırılmış Varlık Tanıma (NER)
相关	3	3	3
摘要	Constituency parsing is a natural-language-processing task that represents a sentence as a tree of recursively nested phrase-structure constituents — for example S → NP + VP. Building on the head-driven statistical parsing models introduced by Collins (2003) and the later neural parsers of Kitaev and colleagues (2019), it exposes the hierarchical syntactic skeleton of a sentence for grammatical pattern extraction and grammar research.	Entity linking is a natural-language-processing task that matches ambiguous entity mentions in text — people, places, organisations — to the correct record in a knowledge base such as Wikidata, DBpedia, or a domain dictionary. Surveyed and shaped by Milne and Witten (2008) and later neural approaches reviewed by Sevgili and colleagues (2022), it grounds free text into structured, unambiguous references used in knowledge-graph building and multi-source text analysis.	Named entity recognition (NER) is a natural-language-processing task that automatically detects and labels entities in text — such as people, organisations, locations, and dates. Surveyed by Nadeau and Sekine (2007) and later advanced with neural architectures by Lample et al. (2016), it turns free-running text into tagged spans that downstream tools can use.