Vietnamese Antonyms Detection Based on Specialized Word Embeddings using Semantic Knowledge and Distributional Information

In this paper, we present one of the solutions to the Taxonomy Enrichment shared task co-located with the Dialogue conference. The proposed method blends distributional information from fastText and BERT word embeddings to predict the most likely parent hypernym node for a new term in a taxonomy. More specifically, we are using both the information on hypernym frequency among the most similar entries in the taxonomy and the similarity of hypernyms themselves. DeepPavlov-based fastText and RuBERT finetuned on news texts and Russian Wikipedia achieve a MAP of 0.3939 and MRR of 0.4353.

Download Full-text

Multimodal Distributional Semantics

Journal of Artificial Intelligence Research ◽

10.1613/jair.4135 ◽

2014 ◽

Vol 49 ◽

pp. 1-47 ◽

Cited By ~ 142

Author(s):

E. Bruni ◽

N. K. Tran ◽

M. Baroni

Keyword(s):

Computational Linguistics ◽

Word Meaning ◽

Semantic Relatedness ◽

Semantic Knowledge ◽

Visual Words ◽

Distributional Information ◽

Flexible Architecture ◽

The Rich ◽

The Many ◽

Distributional Semantic Models

Distributional semantic models derive computational representations of word meaning from the patterns of co-occurrence of words in text. Such models have been a success story of computational linguistics, being able to provide reliable estimates of semantic relatedness for the many semantic tasks requiring them. However, distributional models extract meaning information exclusively from text, which is an extremely impoverished basis compared to the rich perceptual sources that ground human semantic knowledge. We address the lack of perceptual grounding of distributional models by exploiting computer vision techniques that automatically identify discrete visual words in images, so that the distributional representation of a word can be extended to also encompass its co-occurrence with the visual words of images it is associated with. We propose a flexible architecture to integrate text- and image-based distributional information, and we show in a set of empirical tests that our integrated model is superior to the purely text-based approach, and it provides somewhat complementary semantic information with respect to the latter.

Download Full-text

Learning Word Embeddings from Portuguese Lexical-Semantic Knowledge Bases

Lecture Notes in Computer Science - Computational Processing of the Portuguese Language ◽

10.1007/978-3-319-99722-3_27 ◽

2018 ◽

pp. 265-271 ◽

Cited By ~ 2

Author(s):

Hugo Gonçalo Oliveira

Keyword(s):

Knowledge Bases ◽

Semantic Knowledge ◽

Word Embeddings ◽

Lexical Semantic ◽

Lexical Semantic Knowledge

Download Full-text

Phonomotor Versus Semantic Feature Analysis Treatment for Anomia in 58 Persons With Aphasia: A Randomized Controlled Trial

Journal of Speech Language and Hearing Research ◽

10.1044/2019_jslhr-l-18-0257 ◽

2019 ◽

Vol 62 (12) ◽

pp. 4464-4482 ◽

Cited By ~ 9

Author(s):

Diane L. Kendall ◽

Megan Oelke Moldestad ◽

Wesley Allen ◽

Janaki Torrence ◽

Stephen E. Nadeau

Keyword(s):

Randomized Controlled Trial ◽

Response Latency ◽

Controlled Trial ◽

Semantic Knowledge ◽

Semantic Feature ◽

Feature Analysis ◽

Group Differences ◽

Randomized Controlled ◽

Confrontation Naming ◽

Semantic Feature Analysis

Purpose The ultimate goal of anomia treatment should be to achieve gains in exemplars trained in the therapy session, as well as generalization to untrained exemplars and contexts. The purpose of this study was to test the efficacy of phonomotor treatment, a treatment focusing on enhancement of phonological sequence knowledge, against semantic feature analysis (SFA), a lexical-semantic therapy that focuses on enhancement of semantic knowledge and is well known and commonly used to treat anomia in aphasia. Method In a between-groups randomized controlled trial, 58 persons with aphasia characterized by anomia and phonological dysfunction were randomized to receive 56–60 hr of intensively delivered treatment over 6 weeks with testing pretreatment, posttreatment, and 3 months posttreatment termination. Results There was no significant between-groups difference on the primary outcome measure (untrained nouns phonologically and semantically unrelated to each treatment) at 3 months posttreatment. Significant within-group immediately posttreatment acquisition effects for confrontation naming and response latency were observed for both groups. Treatment-specific generalization effects for confrontation naming were observed for both groups immediately and 3 months posttreatment; a significant decrease in response latency was observed at both time points for the SFA group only. Finally, significant within-group differences on the Comprehensive Aphasia Test–Disability Questionnaire ( Swinburn, Porter, & Howard, 2004 ) were observed both immediately and 3 months posttreatment for the SFA group, and significant within-group differences on the Functional Outcome Questionnaire ( Glueckauf et al., 2003 ) were found for both treatment groups 3 months posttreatment. Discussion Our results are consistent with those of prior studies that have shown that SFA treatment and phonomotor treatment generalize to untrained words that share features (semantic or phonological sequence, respectively) with the training set. However, they show that there is no significant generalization to untrained words that do not share semantic features or phonological sequence features.

Download Full-text