Search results

The paper aims to enhance Arabic machine translation (MT) by proposing novel approaches: (1) a dimensionality reduction technique for word embeddings tailored for Arabic text, optimizing efficiency while retaining semantic information; (2) a comprehensive comparison of meta-embedding techniques to improve translation quality; and (3) a method leveraging self-attention and Gated CNNs to capture token dependencies, including temporal and hierarchical features within sentences, and interactions between different embedding types. These approaches collectively aim to enhance translation quality by combining different embedding schemes and leveraging advanced modeling techniques.

Recent works on MT in general and Arabic MT in particular often pick one type of word embedding model. In this paper, we present a novel approach to enhance Arabic MT by addressing three key aspects. Firstly, we propose a new dimensionality reduction technique for word embeddings, specifically tailored for Arabic text. This technique optimizes the efficiency of embeddings while retaining their semantic information. Secondly, we conduct an extensive comparison of different meta-embedding techniques, exploring the combination of static and contextual embeddings. Through this analysis, we identify the most effective approach to improve translation quality. Lastly, we introduce a novel method that leverages self-attention and Gated convolutional neural networks (CNNs) to capture token dependencies, including temporal and hierarchical features within sentences, as well as interactions between different types of embeddings. Our experimental results demonstrate the effectiveness of our proposed approach in significantly enhancing Arabic MT performance. It outperforms baseline models with a BLEU score increase of 2 points and achieves superior results compared to state-of-the-art approaches, with an average improvement of 4.6 points across all evaluation metrics.

The proposed approaches significantly enhance Arabic MT performance. The dimensionality reduction technique improves the efficiency of word embeddings while preserving semantic information. Comprehensive comparison identifies effective meta-embedding techniques, with the contextualized dynamic meta-embeddings (CDME) model showcasing competitive results. Integration of Gated CNNs with the transformer model surpasses baseline performance, leveraging both architectures' strengths. Overall, these findings demonstrate substantial improvements in translation quality, with a BLEU score increase of 2 points and an average improvement of 4.6 points across all evaluation metrics, outperforming state-of-the-art approaches.

The paper’s originality lies in its departure from simply fine-tuning the transformer model for a specific task. Instead, it introduces modifications to the internal architecture of the transformer, integrating Gated CNNs to enhance translation performance. This departure from traditional fine-tuning approaches demonstrates a novel perspective on model enhancement, offering unique insights into improving translation quality without solely relying on pre-existing architectures. The originality in dimensionality reduction lies in the tailored approach for Arabic text. While dimensionality reduction techniques are not new, the paper introduces a specific method optimized for Arabic word embeddings. By employing independent component analysis (ICA) and a post-processing method, the paper effectively reduces the dimensionality of word embeddings while preserving semantic information which has not been investigated before especially for MT task.

This study aims to examine the characteristics of cross-border central bank digital currencies (CBDCs) while pinpointing research trends and adoption variables at both individual and macroeconomic levels. Additionally, it delves into the impact of terminology within CBDC-related scholarly literature themes.

The authors perform a bibliometric study using the metadata of academic papers about CBDC from ScienceDirect, Scopus and Web of Science (WoS), three reputable research databases. Word maps are produced using VOSviewer, an open-source bibliometric analytics program, to find pertinent and predominate words and phrases based on their frequency, placement, connection and co-occurrence. Additionally, the authors use the R programing language to assess the Jaccard similarity between bibliometric metadata and the financial terms in the Loughran-McDonald Master Dictionary (LMMD).

The study pinpoints the factors that affect CBDC adoption at the micro and macroeconomic levels. Insights into prospective future study themes are provided by the analysis of the metadata corpus, which shows significant and predominate words/phrases and themes in CBDC literature. Notably, the relatively low Jaccard similarity scores in the scholarly literature on CBDC-related topics across all three bibliometric databases suggest a restricted concentration on financial issues. This shows that CBDC research is still in its early stages and that there are still many undiscovered financial aspects.

The identification of literature’s themes using dominant and pertinent words based on bibliometric metadata, considering factors such as frequency and co-occurrence, enriches the evolving field of meta-analysis. Additionally, the use of the Jaccard index to assess the coverage of financial terms within bibliometric metadata represents a unique approach, shedding light on the distinctive aspects of CBDC research.

The aim of this study was to examine whether, when a person reads, there is a boost to all material related to the context in Najdi Arabic (NA) as was found in English in was found in Rodd et al. (2013). The study employs Arabic language features of orthographic style and diglossia to answer the research question.

Forty-two participants were asked to come on two sessions to complete a reading task, a filler task and a word recognition task. The word recognition task included 14 homographs that could mean one thing in Modern Standard Arabic and another in NA.

The findings show that the assumption that all related materials are ready to be used when readers are exposed to the context is not valid for the two Arabic variants.

One limitation of this study was that the participants were all female.

The findings could help writers write better texts to help individuals who struggle with reading comprehension whether it is because of dyslexia, Attention Deficit Disorder (ADD) or Attention Deficit Hyperactivity Disorder (ADHD) as when researchers understand how priming works, they might be able to help readers in their reading fluency and comprehension (Rodd et al., 2016). This could be by producing better texts to comprehend or using semantic priming in classroom setups.

Scrip dividend schemes provide shareholders with the option to choose shares instead of a cash dividend. Scrip dividends became popular in South Africa after the introduction of Secondary Tax on Companies (STC) in 1993. Thus far, no guidance on the recognition, measurement or disclosure of scrip dividends has been issued by the South African Institute of Chartered Accountants (SAICA). This article proposes disclosure regarding scrip dividend schemes that will provide relevant information to the users of financial statements. The proposed disclosure is based on the assumption that entities recognise and measure scrip dividends in accordance with the re‐investment method, as opposed to the capitalisation issue method.

A proper noun entry is an index entry for the name of a particular person, place, or thing. While this type of entry is comparatively easy for humans to recognize, it is difficult to define for the purposes of computer processing. The kind and number of proper nouns which appear in a text vary to some extent with the subject matter. The ratio of proper nouns to all other text words may be higher in such subject areas as history and literature than it is, for example, in the sciences. However, in many instances proper nouns constitute a high proportion of index entries in the sciences.

Screen‐based word processors have been available all through the 1970s, although widespread interest and use dates from about 1975. Conventional offices, generating and manipulating many thousands of words, started to take advantage of the features word processors offer on a much wider scale than they had ever done with their forerunners, the automatic typewriters. Libraries and information departments who were concerned primarily with the already printed and published word were rather slower to realise that they too generated a relatively large number of words in their own right and that these words, in the form of, for example, catalogue cards, accessions lists, abstracts, bibliographies, reports and so on, could perhaps be handled more beneficially by word processors. However, once it began, interest grew rapidly, although actual installation of equipment has been considerably slower—probably a reflection of the limited budgets frequently allocated to such services for the purchase of technological aids.

Reliability of RAM memory systems is impaired by environmental disturbances, causing soft errors, whereby one data bit is transformed to another bit. Single‐error correcting codes with memory scrubbing offer the most effective method to recover from such errors. This paper analyzes the reliability and determines the MTTF for simplex and duplex memory systems with single‐error correction and/or soft‐error scrubbing recovery. It extends previous work on the deterministic scrubbing recovery of simplex memory systems by using a more general model that takes into account cancelling soft errors. In the duplex memory system an additional level of static redundancy is proposed by employing a decoding algorithm at the memory module level. The reliability analysis of the duplex system with soft‐error scrubbing takes into account the decoder output which upon scrubbing transforms words with a number of multiple errors to words with a different number of errors. Computer results show that this combination of data and system redundancy provides more reliability than either data or system redundancy alone.

I had little thought when I chose “words” for this article that I had such a large project before me. I soon found that I had embarked on an ocean and that soon or late I would find myself swamped and sunk if I did not limit myself and keep within territorial waters. This paper will reveal no erudition in etymology or grammar. Neither will it pretend to be an exemplar of the King's English. “The choice word and measured phrase beyond the reach of ordinary men,” as Wordsworth puts it, will only be evident in quotation. The warning of Ecclesiastes, “Let thy words be few,” hangs over me, for, as Pope, a veritable huckster in words, says: “Words are like leaves and where they most abound much fruit of sense beneath is rarely found.”

‘Thesaurus’ comes from the Greek—a storehouse or treasury. The Shorter Oxford English dictionary gives 1736 for the English usage ‘a treasury or storehouse of knowledge, as a dictionary, encyclopaedia or the like’. In 1852, Peter Mark Roget published his Thesaurus of English words and phrases: ‘a collection … arranged, not in alphabetical order as they are in a dictionary, but according to the ideas which they express… The object aimed at is, the idea being given, to find the word, or words, by which that idea may be most fitly and aptly expressed.’ All who write English have heard of this thesaurus.

A simple procedure is used to determine the minimum amount of information required in order to identify a word when contextual information is provided in the form of a sentence with a blank spot into which the word fits. A population of subjects are presented with the sentence and are required to guess at the word which belongs in the blank spot. From the probability with which the expected word is guessed, one can determine the minimum additional information required for the identification of the missing word.