George A. Miller
George A. Miller's papers here are about WordNet, a lexical database of English, and its use in word sense identification, the task of deciding which meaning a word carries in a particular context. The work pairs a hand-built lexicon, in which words are grouped into synonym sets joined by relations such as hyponymy, meronymy, troponymy and entailment, with sense-tagged passages of the Brown Corpus and statistical evidence gathered from larger bodies of text.
The most cited of these is WordNet: A Lexical Database for English (1994). It is a brief progress report on how the database organises lexicalised concepts and how quickly it was growing. It also describes early tests in which local context and WordNet-based similarity measures improved the choice of sense. A Semantic Concordance (1993) proposed connecting every content word in a corpus to the dictionary sense it expresses. The result could serve both as annotated text and as a dictionary illustrated with real usage. Using a Semantic Concordance for Sense Identification (1994) drew on that tagged corpus to ask what baseline accuracy an automatic system must exceed to count as progress. It grounded such a floor in observed sense frequencies rather than in unexplained estimates.
The latest paper, Using Corpus Statistics and WordNet Relations for Sense Identification (1998), turns to the expense of producing sense-tagged training material. It introduces TLC, a statistical classifier that can rely on topical cues, on cues from a narrow window around the target word, or on both. Local context proved the more informative signal. The paper combines corpus statistics with WordNet relations to lessen dependence on hand-tagged examples.
The work here comes from the WordNet project at Princeton University. Bylines under the name also list Harvard University and the Massachusetts Institute of Technology over earlier decades. Co-authors on these papers include Claudia Leacock and Martin Chodorow, with whom Miller developed the corpus-based approach to sense identification.
George A. Miller has 4 papers in our library, in 1993–1998; often with Claudia Leacock, Martin Chodorow.
OpenAlex · h-index 68 · 216 works · 47 683 citations OpenAlex, 2026-10-14
4 works in our library · 2 553 citations in our library · ⭐ Paper Star rank #767
Affiliations
- Massachusetts Institute of Technology · Cambridge, United States — 1952–2013 on 9 papers
- Princeton University · Princeton, United States — 1960–2009 on 43 papers
- Harvard University · Cambridge, United States — 1946–1994 on 49 papers
- Rockefeller University · New York, United States — 1968–1979 on 14 papers
- MIT Lincoln Laboratory · Lexington, United States — 1954–1955 on 3 papers
Works by year
1990s
Papers
- WordNet: A Lexical Database for English (1994 · 2 179 citations)
The paper is a short progress report for the WordNet project at Princeton. WordNet groups words into synonym sets (synsets) that stand for lexicalised concepts, and links them with pointers for synonymy, antonymy, hyponymy, meronymy, troponymy and entailment. As of 9 February 1994 it has 81,658 synsets and 109,570… - Using Corpus Statistics and WordNet Relations for Sense Identification (1998 · 518 citations)
Supervised word sense identification depends on sense-tagged training text, which is expensive to produce. - Using a Semantic Concordance for Sense Identification (1994 · 352 citations)
The paper asks what baseline accuracy an automatic word sense identification system must beat to count as progress. - A Semantic Concordance (1993 · 293 citations)
The authors propose the notion of a "semantic concordance". This is a resource in which a corpus and a dictionary are combined so that each content word occurring in the corpus is connected to the lexicon sense it expresses. Such a resource can be used both as a corpus with syntactic and semantic annotation and as a…
Co-authors here
Cited by
This page in Markdown → https://md.research.elamur.ai/free/en/authors/george-a-miller/