Quality Estimation (QE) plays an essential role in applications of Machine Translation (MT). Traditionally, a QE system accepts the original source text and translation from a black-box MT system as input. Recently, a few studies indicate that as a b
y-product of translation, QE benefits from the model and training data's information of the MT system where the translations come from, and it is called the glass-box QE''. In this paper, we extend the definition of glass-box QE'' generally to uncertainty quantification with both black-box'' and glass-box'' approaches and design several features deduced from them to blaze a new trial in improving QE's performance. We propose a framework to fuse the feature engineering of uncertainty quantification into a pre-trained cross-lingual language model to predict the translation quality. Experiment results show that our method achieves state-of-the-art performances on the datasets of WMT 2020 QE shared task.
This paper, intended for the ISA-17 Quantification Annotation track, provides background information for the shared quantification annotation task at the ISA-17 workshop, a.k.a. the Quantification Challenge. In particular, the role of the abstract an
d concrete syntax of the QuantML markup language are explained, and the semantic interpretation of QuantML annotations in relation to the ISO principles of semantic annotation. Additionally, the choice is motivated of the test suite of the Quantification Challenge, along with the suggested markables for the sentences of the suite.
Formaldehyde is classified by the World Health Organization (International Agency
for Research on Cancer) as a carcinogen in Group 1. The upper limit allowed for use in
cosmetic products as a preservative has to not exceed 0.2%. This study aims to
detect and
determine formaldehyde level in some cosmetic products available in the local market
using acetyl acetone method. Formaldehyde was detected in 77% of the studied cosmetic
products (49 samples) despite the fact that most of them were not labeled formaldehyde or
formaldehyde donor preservatives (about 70%). The amount of formaldehyde in keratin
samples were higher than the allowed limit (0.34% -11.73%) except one sample 0.18%. pH
decrease and temperature increase of keratin samples led to an increase of formaldehyde
level as a result of its release from preservatives.
The topic of this research aims at minimizing carpet image color number
(acquired by a scanner) from 16 million colors to 5 colors, in order to restore
the shapes of the original image automatically without distortions.
For this purpose an algorit
hm was developed for the application called CDS
(Carpet Design System)- which is an application developed locally for use in
sewing carpets in General Institution of Texture Factory in Syria-, but the
results are still not acceptable, because the resulting images have a lot of
distortions, and they need some processing using an image processing
application (ex. Photoshop). This task needs more than tow weeks.
In this research work, we study the CDS algorithm, and other color
quantification algorithms, developed in research laboratories specialized in
image processing. We apply this algorithms on carpet images, after doing the
necessary modifications for tuning and adapting to our special problem. Finally
we compare the results, and suggest the best solution for the problem.
