Notice bibliographique
Résumé
Words in natural language often take on new meanings, so that language users can express an infinite set of intended meanings with a finite vocabulary. Such processes of word meaning extension (WME) are highly productive, diverse, but also non-arbitrary. However, it remains unclear what common cognitive mechanisms and knowledge are driving various types of word meaning extension. Modeling WME is also relevant for natural language processing (NLP), since novel lexical expressions are constantly emerging through time, and NLP systems should effectively interpret and generate such novel word usages in a human-like way. In this dissertation, I develop a computational framework that not only accounts for word meaning extension in historical language development, but also supports NLP systems to flexibly interpret and generate novel word usages. My dissertation is organized into two main parts. In the first part, I study word meaning extension from a computational cognitive science perspective. I first propose a probabilistic generative model of word meaning extension that relies on multimodal semantic knowledge, and show that the model, when incorporated with the cognitive processes of chaining, can accurately predict historical emergence of novel verb-noun syntactic compositions. Next, I present a novel and general account of semantic chaining through the lens of cognitive efficiency. This account explains different chaining mechanisms as a tradeoff between representation and complexity. I show that the efficiency-based framework can be formulated as an infinite mixture model from Bayesian non-parametric statistics, which adaptively constructs word meaning through time under limited computational resources. In the second part of my dissertation, I study how modeling human-like word meaning extension can enhance natural language processing. Specifically, I propose the problem of word sense extension, where a neural language model trained on limited linguistic data is asked to generate novel usages based on previously unseen word senses. I develop a generative framework that combines deep few-shot learning with semantic chaining to capture incremental word sense extension, and I show that this framework can be leveraged to fine-tune and improve word sense disambiguation on rare word senses. Furthermore, I investigate how modeling the systematicity in word meaning extension in combination with language models helps the construction of non-literal word usages like metaphors. I show that learning the analogical similarity between word meanings effectively improves language model systematicity in making both incremental and irregular types of word meaning extension. I therefore suggest that learning systematic word meaning extension benefits language models on multiple tasks pertaining to figurative language understanding. In summary, my dissertation contributes a principled paradigm for modeling the generative processes of word meaning extension, and it opens up future opportunities for human-like automated processing of creative language use.
Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.
Comment cette classification a été obtenuedéplier
Prédiction distillée sur la base complète
Imitation des enseignantsNi prévalence calibrée, ni vérité terrain. Validation humaine à venir. Apprise à partir de 10 348 étiquettes directes de Codex et de 10 348 étiquettes directes de Gemma. Le mode candidate est l'union des têtes enseignantes seuillées; le consensus est leur intersection. Ces sorties portent le statut machine_predicted_unvalidated et ne sont ni des étiquettes humaines ni des étiquettes directes de modèles de pointe.
Scores Codex et Gemma par catégorie
| Catégorie | Codex | Gemma |
|---|---|---|
| Métarecherche | 0,000 | 0,000 |
| Méta-épidémiologie (sens strict) | 0,000 | 0,000 |
| Méta-épidémiologie (sens large) | 0,000 | 0,000 |
| Bibliométrie | 0,000 | 0,001 |
| Études des sciences et des technologies | 0,001 | 0,000 |
| Communication savante | 0,000 | 0,000 |
| Science ouverte | 0,000 | 0,000 |
| Intégrité de la recherche | 0,000 | 0,000 |
| Charge utile insuffisante (le modèle a refusé de juger) | 0,000 | 0,000 |
Scores machine (provisoires)
Les deux têtes enseignantes du modèle étudiant, lues sur ce travail. Un score ordonne la base pour la relecture; il n'affirme jamais une catégorie, et le statut de validation accompagne chaque rangée tel quel.
Scores de référence d'un modèle non mature (critères de maturité non atteints, 7 itérations). Un score ordonne; il n'affirme jamais une catégorie.
score_only:v0-immature-baseline · tel quel depuis la passe de notation : score_only signifie que le nombre peut ordonner les travaux, et qu'aucune étiquette de catégorie n'en découleClassification
machine, non validéePrédiction automatique; un appel candidat d’une seule tête enseignante, pas un consensus.
Le détail, modèle par modèle et score par score, se trouve en fin de page sous « Comment cette classification a été obtenue ».