Representation of sentence meaning (A JNLE Special Issue)

Bojar, Ondřej; Bernardi, Raffaella; Webber, Bonnie

doi:10.1017/s1351324919000172

Cited by 3 publications

(1 citation statement)

References 10 publications

(10 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, it is not so clear in which way the embeddings should be evaluated. In an attempt to bring together more traditional representations of sentence meanings and the emerging vector representations, Bojar et al (2019) introduce a number of aspects or desirable properties of sentence embeddings. One of them is denoted as "relatability", which highlights the correspondence between meaningful differences between sentences and geometrical relations between their respective embeddings in the highly dimensional continuous vector space.…”

Section: Introductionmentioning

confidence: 99%

COSTRA 1.0: A Dataset of Complex Sentence Transformations

Barancikova,

Bojar

2019

Preprint

View full text Add to dashboard Cite

We present COSTRA 1.0, a dataset of complex sentence transformations. The dataset is intended for the study of sentence-level embeddings beyond simple word alternations or standard paraphrasing. This first version of the dataset is limited to sentences in Czech but the construction method is universal and we plan to use it also for other languages. The dataset consist of 4,262 unique sentences with average length of 10 words, illustrating 15 types of modifications such as simplification, generalization, or formal and informal language variation. The hope is that with this dataset, we should be able to test semantic properties of sentence embeddings and perhaps even to find some topologically interesting "skeleton" in the sentence embedding space. A preliminary analysis using LASER, multi-purpose multi-lingual sentence embeddings suggests that the LASER space does not exhibit the desired properties.

show abstract

Section: Introductionmentioning

confidence: 99%

COSTRA 1.0: A Dataset of Complex Sentence Transformations

Barancikova,

Bojar

2019

Preprint

View full text Add to dashboard Cite

show abstract

Do Machines Really Understand Meaning? (Again)

Perconti

Plebe

2023

J. AI. Consci.

View full text Add to dashboard Cite

The adventure of artificial intelligence (AI) is based on a revolutionary idea, namely, that machines are able to understand and produce linguistic acts endowed with meaning. Over the past decades, this idea has gained acceptance in the scientific community, but its consequences for human nature were not really appreciated. Recent developments in AI, due especially to Deep Learning (DL), have changed things dramatically by creating computer architectures capable of performing previously impossible tasks in areas such as image recognition and language understanding. Criticisms that were raised decades ago against this possibility have thus been revived. These criticisms are no different in argument from those made in the first place. The reason they are being raised again is because of the social impact that the new machine performances have been able to achieve, and is not driven by truly scientific reasons, which indeed prove to be inadequate when compared to the more recent artificial semantics based on deep neural networks.

show abstract

Therapeutic Exon Skipping Through a CRISPR-Guided Cytidine Deaminase Rescues Dystrophic Cardiomyopathy in Vivo

Wang

Zhang

et al. 2021

Circulation

View full text Add to dashboard Cite

Background: Loss of dystrophin protein causes Duchenne muscular dystrophy (DMD), characterized by progressive degeneration of cardiac and skeletal muscles, and mortality in adolescence or young adult. Although cardiac failure has risen as the leading cause of mortality in patients with DMD, effective therapeutic interventions remain underdeveloped, in part, due to the lack of a suitable preclinical model. Methods: We analyzed a novel murine model of DMD created by introducing a 4-bp deletion into exon 4, one of the exons encoding the actin-binding domain 1 of dystrophin (referred to as Dmd E4* mice). Echocardiography, micro-CT, muscle force measurement, and histological analysis were performed to determine cardiac and skeletal muscle defects in these mice. Using this model, we examined the feasibility of using a cytidine base editor to install exon skipping and rescue dystrophic cardiomyopathy in vivo . AAV9-based CRISPR/Cas9-AID (eTAM) together with AAV9-sgRNA was injected into neonatal Dmd E4* mice, which were analyzed 2- or 12-month post treatment to evaluate the extents of exon skipping, dystrophin restoration, and phenotypic improvements of cardiac and skeletal muscles. Results: Dmd E4* mice recapitulated many aspects of human DMD, including shortened lifespan (by ∼50%), progressive cardiomyopathy, kyphosis, profound loss of muscle strength, and myocyte degeneration. A single-dose administration of AAV9-eTAM instituted over 50% targeted exon skipping in the Dmd transcripts and restored up to 90% dystrophin in the heart. As a result, early ventricular remodeling was prevented and cardiac and skeletal muscle functions were improved, leading to an increased lifespan of the Dmd E4* mice. Despite gradual decline of AAV vector and base editor expression, dystrophin restoration and pathophysiological rescue of muscular dystrophy were long lasted for at least one year. Conclusions: Our study demonstrates the feasibility and efficacy to institute exon skipping via an enhanced TAM (eTAM) for therapeutic application(s).

show abstract

Representation of sentence meaning (A JNLE Special Issue)

Cited by 3 publications

References 10 publications

COSTRA 1.0: A Dataset of Complex Sentence Transformations

COSTRA 1.0: A Dataset of Complex Sentence Transformations

Do Machines Really Understand Meaning? (Again)

Therapeutic Exon Skipping Through a CRISPR-Guided Cytidine Deaminase Rescues Dystrophic Cardiomyopathy in Vivo

Contact Info

Product

Resources

About