Skip to content
Carmelics
TopicsThinkersChangesContributorsLoading account…

    Carmelics

    A reasoning platform. Break down any belief into clear reasons, explore both sides, and weigh the evidence honestly.

    Navigate

    • Topics
    • Search
    • Recent Changes
    • Contribute
    • How It Works
    • Glossary
    • Thinkers
    • Contributors
    • About
    • Statistics
    • Terms
    • Privacy

    Database

    Statements
    —
    Perspectives
    —
    Topics
    —

    Press ? for keyboard shortcuts

    LoyalLoyalJusticeJustice
    Made withinDC&Austin
    Broad-coverage statistical word sense disambiguation tool... — Carmelics
    Statements
    321,452
    Perspectives
    108,905
    Topics
    42
    Home/Philosophy of Language
    HistoryEditSee Inverse

    Broad-coverage statistical word sense disambiguation tools remain elusive

    Philosophy of Language
    ?Rate how convincing each reason is below to see the overall strength.
    1 reason for
    2 reasons against

    Reasons For

    1 perspective
    Reason for
    ?
    • 1.Thousands of words have multiple senses in sources such as WordNet
      ?

      Think about whether this reason is strong or weak

    • 2.Constructing a sense-annotated training corpus with sufficiently many occurrences of all senses is difficult
      ?

      Think about whether this reason is strong or weak

    • 3.Annotations are therefore typically restricted to the senses of a few polysemous words
      ?

      Think about whether this reason is strong or weak

    Reasons Against

    2 perspectives
    Reason against 1 of 2
    ?
    • 1.Large-scale language models like BERT achieve near-human WSD performance on benchmark tasks by leveraging contextual embeddings across broad vocabularies.
      ?

      Think about whether this reason is strong or weak

    • 2.When a system matches human inter-annotator agreement on WSD benchmarks, the claim of elusiveness collapses into a question about evaluation criteria, not capability.
      ?

      Think about whether this reason is strong or weak

    Reason against 2 of 2
    ?
    • 1.Wittgenstein's argument that word meaning is use implies discrete sense inventories like WordNet artificially multiply the 'disambiguation problem' beyond what cognition requires.
      ?

      Think about whether this reason is strong or weak

    • 2.If the problem is partly an artifact of lexicographic sense enumeration rather than natural language structure, the elusiveness of broad-coverage WSD reflects a methodological choice, not an empirical barrier.
      ?

      Think about whether this reason is strong or weak

    Sign in or register to share your perspective on this statement.

    Next step

    Based on where you are in your exploration

    Strongest counterpoint
    Explore the most compelling reason on the other side.

    Topics

    Philosophy of Language

    Related

    Annotations are therefore typically restricted to the senses of a few polysemous...Constructing a sense-annotated training corpus with sufficiently many occurrence...If the problem is partly an artifact of lexicographic sense enumeration rather t...Large-scale language models like BERT achieve near-human WSD performance on benc...
    +3 moreShow less
    Thousands of words have multiple senses in sources such as WordNetWhen a system matches human inter-annotator agreement on WSD benchmarks, the cla...Wittgenstein's argument that word meaning is use implies discrete sense inventor...

    Similar

    Neural net and statistical techniques can help disambiguate word sense...82%Lepore and Stone (2016) deploy disambiguation systematically across a ...74%Word sense information merges with CPI, CD, and SCWD only when all wor...72%Lepore and Stone (2016) argue that disambiguators are more widespread ...71%

    Source

    AI-extracted1/3 agreementValid
    SEP: computational-linguistics
    View source passageHide passage
    Classification of selected words or phrases in sentential or broader contexts: As noted earlier, examples include WSD, named entity recognition, and sentence boundary detection. The only point of distinction from text/document classification is that it is not a chunk of text as a whole, but rather a word or phrase in the context of such a chunk that is to be classified. Therefore features are chosen to reflect both the features of the target word or phrase (such as morphology) and the way it rel
    Extraction notes

    Validity: Extracted via Max plan + API grounding/validity checks

    Details

    Type
    claim
    Perspectives
    3 (1 for, 2 against)
    Edits
    1 edit