Metraipolis › archive › dc_ddbf6f0c8cf7

Ambiguity-Robust Dictionaries using Contextualized Embeddings (2026) · DOI 10.31235/osf.io/m8sjn_v1

Source
Metraipolis archive
Original
https://doi.org/10.31235/osf.io/m8sjn_v1
License (as archived)
CC-BY-4.0
Archived
2026-10-08T02:12:26.469Z
SHA-256 of the archived text
428ff6ebddfe27c010de052d77b24782a4dd4af11be0e49ef8ea3bf333e8fc73
Stable URL of this page
https://metraipolis.com/documento/dc_ddbf6f0c8cf7

How to cite

Ambiguity-Robust Dictionaries using Contextualized Embeddings (2026) · DOI 10.31235/osf.io/m8sjn_v1 — Metraipolis archive, https://doi.org/10.31235/osf.io/m8sjn_v1 (CC-BY-4.0); archived copy: https://metraipolis.com/documento/dc_ddbf6f0c8cf7 (Metraipolis archive dc_ddbf6f0c8cf7, sha256 428ff6ebddfe27c010de052d77b24782a4dd4af11be0e49ef8ea3bf333e8fc73)
BibTeX
@misc{dcddbf6f0c8cf7,
  title = {Ambiguity-Robust Dictionaries using Contextualized Embeddings (2026) · DOI 10.31235/osf.io/m8sjn_v1},
  howpublished = {Metraipolis archive},
  url = {https://doi.org/10.31235/osf.io/m8sjn_v1},
  note = {License CC-BY-4.0; archived copy https://metraipolis.com/documento/dc_ddbf6f0c8cf7; sha256 428ff6ebddfe27c010de052d77b24782a4dd4af11be0e49ef8ea3bf333e8fc73},
  year = {2026}
}

Archived text

Reproduced under CC-BY-4.0. Attribution: Metraipolis archive, https://doi.org/10.31235/osf.io/m8sjn_v1. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Articolo recente con licenza aperta: scheda dai metadati Crossref (posted-content).
Titolo: Ambiguity-Robust Dictionaries using Contextualized Embeddings
Autori: Jamil Civitarese, Sylvan Zheng
Pubblicato: 2026-08-21 · editore: Center for Open Science
DOI: https://doi.org/10.31235/osf.io/m8sjn_v1
Licenza dell'articolo: CC-BY-4.0

Abstract (testo degli autori, CC-BY-4.0):
Keyword dictionaries remain widely used in text analysis for their simplicity, transparency, and replicability. However, sensitivity to keyword selection and inability to account for context introduce measurement error that attenuates estimates and obscures real relationships. We introduce ambiguity-robust dictionaries, a text-as-data method that leverages contextual word embeddings to produce more precise measures while preserving the key advantages of dictionary approaches. Our method generates contextualized embeddings for all keyword instances, then applies constrained fuzzy clustering to determine the degree of each instance's membership in a target cluster defined by a small set of researcher-identified anchor words. This approach allows the construction of ambiguity-robust document level word counts, increasing precision and reducing sensitivity to keyword selection. We demonstrate the method using three applications: UN environmental discourse, ethnic bias in Kenyan judicial politics, and moral language in US Congressional speech. In each application, our method yields substantively more robust and precise estimates than conventional approaches.

Fonte: https://api.crossref.org/works/10.31235%2Fosf.io%2Fm8sjn_v1 — metadati Crossref, consultati il 2026-10-07. Interrogazione: «multilingual embeddings retrieval».