Multiple Scripts Research Articles

Use of prompt-based learning for code-mixed and code-switched text classification

Code-mixing and code-switching (CMCS) are prevalent phenomena observed in social media conversations and various other modes of communication. When developing applications such as sentiment analysers and hate-speech detectors that operate on this social media data, CMCS text poses challenges. Recent studies have demonstrated that prompt-based learning of pre-trained language models outperforms full fine-tuning across various tasks. Despite the growing interest in classifying CMCS text, the effectiveness of prompt-based learning for the task remains unexplored. This paper presents an extensive exploration of prompt-based learning for CMCS text classification and the first comprehensive analysis of the impact of the script on classifying CMCS text. Our study reveals that the performance in classifying CMCS text is significantly influenced by the inclusion of multiple scripts and the intensity of code-mixing. In response, we introduce a novel method, Dynamic+AdapterPrompt, which employs distinct models for each script, integrated with adapters. While DynamicPrompt captures the script-specific representation of the text, AdapterPrompt emphasizes capturing the task-oriented functionality. Our experiments on Sinhala-English, Kannada-English, and Hindi-English datasets for sentiment classification, hate-speech detection, and humour detection tasks show that our method outperforms strong fine-tuning baselines and basic prompting strategies.

Optical Character Recognition of Balochi Script

Optical Character Recognition is considered one of the fastest methods of data entry. OCR converts the text image representation of x and y coordinates representing pixel information to be converted into text data in a particular language. OCR as a field of pattern recognition and document image understanding, OCR requires a challenging job once a different language text is available on the image. Difference in language script will pose different challenges for OCR which requires entirely different approaches and algorithms. Latin scripts require a different approach whereas the Balochi adopted language scripts require a different approach. In this regard, various solutions have been proposed for different languages. Segmentation is considered one of the important tasks in the process of OCR. A good segmentation will definitely increase the accuracy of an OCR. Segmentation includes the segmentation of text lines from text images which are further divided into words. These segmented words are further divided into characters which are to be recognized. A single segmentation algorithm to segment various scripts of the languages is proposed in this study which checks the script and then segments the text image for the further processing in OCR. The proposed generalized algorithm will check the style, direction and other properties of the script and then adopts the segmentation process to segment text lines, words and characters of the language. The proposed algorithm segments more than ten languages of three scripts and segments for their OCRs. These images can be further processed for feature extraction and classification further. The process of OCR for selected languages will be made easier to recognize. Multiple scripts, languages and images were experimented, and the proposed algorithm successfully segmented 42,833 images of text line, words and character image. The algorithm provides 97% accuracy while segmenting these images and can be extended to further languages as well as scripts .

Multiple scripts, multiple institutions: Introducing complexity into the understanding of women’s empowerment

Partial enactment of women’s rights is at the crux of this analysis, which identifies factors associated with the adoption of some global women’s rights scripts but not others. Women who partially enact global principles are an important group, and focusing on them provides clues into when, where, and how institutionalized scripts are in competition. To explore this issue, Demographic and Health Survey data from 25 low- and middle-income countries across two time periods are used, with a focus on two dimensions of women’s empowerment: a woman’s household decision-making power and her attitudes toward intimate partner violence. Multinomial regressions reveal that exposure to global culture is associated with dual enactment of the two dimensions. Among partial adopters, enactment privileging physical integrity is mediated through local community institutions, including religions, whereas partial-enactment privileging decision making is associated with women’s household bargaining power.

Ferry: Toward Better Understanding of Input/Output Space for Data Wrangling Scripts.

Understanding the input and output of data wrangling scripts is crucial for various tasks like debugging code and onboarding new data. However, existing research on script understanding primarily focuses on revealing the process of data transformations, lacking the ability to analyze the potential scope, i.e., the space of script inputs and outputs. Meanwhile, constructing input/output space during script analysis is challenging, as the wrangling scripts could be semantically complex and diverse, and the association between different data objects is intricate. To facilitate data workers in understanding the input and output space of wrangling scripts, we summarize ten types of constraints to express table space and build a mapping between data transformations and these constraints to guide the construction of the input/output for individual transformations. Then, we propose a constraint generation model for integrating table constraints across multiple transformations. Based on the model, we develop Ferry, an interactive system that extracts and visualizes the data constraints describing the input and output space of data wrangling scripts, thereby enabling users to grasp the high-level semantics of complex scripts and locate the origins of faulty data transformations. Besides, Ferry provides example input and output data to assist users in interpreting the extracted constraints and checking and resolving the conflicts between these constraints and any uploaded dataset. Ferry's effectiveness and usability are evaluated through two usage scenarios and two case studies, including understanding, debugging, and checking both single and multiple scripts, with and without executable data. Furthermore, an illustrative application is presented to demonstrate Ferry's flexibility.

Chaos or system?

Abstract This paper seeks to reassess the balance between chaos and systematicity within the Japanese writing system (JWS), which is noted for its complexity. As potential factors for chaos, Section 2 focuses on two important conventions. The first is the simultaneous use of multiple scripts as components of a largely systematic whole, even though it also affords considerable levels of graphematic variation. The second convention of dual-readings turns on kanji graphematically mapping to both Native-Japanese (NJ) and Sino-Japanese (SJ) morphemes, which yields the JWS’s intriguing form of morphography. However, as a factor that is, to a considerable degree, a major source of systematicity, Section 3 outlines the graphematic representation of SJ compound words. More specifically, starting from the morphological structures of two-kanji compound words (2KCWs), Section 3 introduces the dominant morphological patterns of three-kanji (3KCWs) and four-kanji compound words (4KCWs), which underscore the significance of 2KCWs within the Japanese mental lexicon, and concludes by noting Hatano, Kuhara, and Akiyama’s (1981) study about inferring the meanings of SJ compound words.

An end-to-end model for multi-view scene text recognition

Due to the increasing applications of surveillance and monitoring such as person re-identification, vehicle re-identification and sports events tracking, the necessity of text detection and end-to-end recognition is also growing. Although the past deep learning-based models have addressed several challenges such as arbitrary-shaped text, multiple scripts, and variations in the geometric structure of characters, the scope of the models is limited to a single view. This paper presents an end-to-end model for text recognition through refining the multi-views of the same scene, which is called E2EMVSTR (End-to-End Model for Multi-View Scene Text Recognition). Considering the common characteristics shared in multi-view texts, we propose a cycle consistency pairwise similarity-based deep learning model to find texts more efficiently in three input views. Further, the extracted texts are supplied to a Siamese network and semi-supervised attention embedding combinational network for obtaining recognition results. The proposed model combines natural language processing and genetic algorithm models to restore missing character information and correct wrong recognition results. In experiments on our multi-view dataset and several benchmark datasets, the proposed method is proven effective compared to the state-of-the-art methods. The dataset and codes will be made available to the public upon acceptance.

Generalized Segmentation Algorithm for Dissimilar Script Languages

Optical Character Recognition is considered one of the fastest methods of data entry. OCR converts the text image representation of x and y coordinates representing pixel information to be converted into text data in a particular language. OCR as a field of pattern recognition and document image understanding, OCR requires a challenging job once a different language text is available on the image. Difference in language script will pose different challenges for OCR which requires entirely different approaches and algorithms. Latin scripts require a different approach whereas the Arabic adopted language scripts require a different approach. In this regard, various solutions have been proposed for different languages. Segmentation is considered one of the important tasks in the process of OCR. A good segmentation will definitely increase the accuracy of an OCR. Segmentation includes the segmentation of text lines from text images which are further divided into words. These segmented words are further divided into characters which are to be recognized. A single segmentation algorithm to segment various scripts of the languages is proposed in this study which checks the script and then segments the text image for the further processing in OCR. The proposed generalized algorithm will check the style, direction and other properties of the script and then adopts the segmentation process to segment text lines, words and characters of the language. The proposed algorithm segments more than ten languages of three scripts and segments for their OCRs. These images can be further processed for feature extraction and classification further. The process of OCR for selected languages will be made easier to recognize. Multiple scripts, languages and images were experimented, and the proposed algorithm successfully segmented 32,833 images of text line, words and character image. The algorithm provides 97% accuracy while segmenting these images and can be extended to further languages as well as scripts.

Script-switching in Japanese pop culture: a social semiotic multimodal approach

This article explores the nature of multimodal meaning-making in writing by analyzing the script-switching used in Japanese pop culture. Japanese intermixes multiple script systems in one text. Although official and public documents follow standard orthographic conventions, creative works in pop culture texts freely and regularly break these norms by switching script types to create new meanings. However, while many previous works recognize the semiotic effects of script-switching, to this day, there is no systematic account for how and why such meaning-making operates. Notably, script-switching triggers meanings independent of linguistic meaning. Yet, scripts are a visual manifestation of language, so one cannot say they are independent of language. Accordingly, this work attempts to resolve this dilemma by exploring a social semiotic multimodal approach (see Van Leeuwen ‘Typographic meaning’, 2005, and ‘Towards a semiotics of typography’, 2006, and Stöckl ‘Typography: Body and dress of a text’, 2005), which claims the semiotic independence of typography and its interdependency with language. By recognizing the basic unit of the writing system as graphemes, the author argues that the Japanese script systems are categorized into four types of graphemes depending on the linguistic unit they relate to and she demonstrates how the visual unit of the grapheme and the linguistic unit interact in meaning-making. In this analysis, script-switching occurs when grapheme types are intentionally changed to foreground a visual meaning associated with each script system. This article shows that such a meaning reflects the historical development and discursive social practices of writing with multiple script systems throughout the history of the Japanese writing system.

Multiscriptality within the European Union: the case of a Greek and a Bulgarian urban landscape

ABSTRACT In an increasingly globalised and multilingual world, the use of different scripts in the same semiotic landscape is an increasingly frequent and widespread phenomenon. For this reason, it is vital to conduct research focusing on multiscriptality in order to better understand the linguistic and semiotic functions of the use of multiple scripts that have become more or less officialised by the political-administrative power. This study examines the use of Latin characters in Greece and Bulgaria, the only two European Union member societies that formally adopt different scripts. Using both a qualitative and a quantitative approach, we analyse the coexistence of the national and Latin alphabets in the graphic texts of commercial signs in two urban areas: the Ladadika district in Thessaloniki, Greece and the Kapana district in Plovdiv, Bulgaria. The results show that the advancement of the Latin alphabet is not homogeneous; sometimes unconventionally represents the national language; is not completely preferred to the recessive national script, but creates with it new orthographic-communicative forms; and can take place regardless of the overwhelming prestige of Global English. This paper demonstrates the importance of studying multiscriptality and the Romanisation process in the context of contemporary multilingualism.

Development and Evaluation of a Web-Based and Interactive Flood Management Tool for Awash and Omo-Gibe Basins, Ethiopia

Flood risk management is used to monitor floodwater and mitigate flooding that impacts people, properties and infrastructures, and the environment. This study developed an interactive web-based “flood tool” for Awash and Omo-Gibe basins in Ethiopia to improve the flood monitoring services and facilities. The data used were real-time and seasonal rainfall-runoff forecasts, flood inundations, and other forecast products for the 2021 flood season (June to September) in a case study. Methods used were multiple scripts written in the Hypertext Markup Language (HTML) and the Visual Studio Code as a coding environment. The coefficient-of-determination (R2) and efficiency (NSE) were used to evaluate the forecast products. The R2 values for selected river stations were the Awash-Hombole (0.79), Mojo (0.64), Awash-7 (0.66), Awash-Adaitu (0.62), Gibe-Tolai (0.78), and Gibe-Abelti (0.70) rivers. The R2 values for Koka and Gibe-3 reservoirs inflows (water levels) forecasts were 0.97 (0.96) and 0.93 (0.99), and the NSE values were 0.89 (0.88) and 0.92 (0.95) for each reservoir, respectively. Besides, the flood inundation extents (km2) from satellite observation and model were compared for the main flood-prone areas and in agreement with very good performance. The flood tool can therefore present early warning forecast products and convey advice to decision-makers to take action for the people at risk.

Handwriting quality analysis using online-offline models

This work is part of an innovative e-learning project allowing the development of an advanced digital educational tool that provides feedback during the process of learning handwriting for young school children (three to eight years old). In this paper, we describe a new method for children handwriting quality analysis. It automatically detects mistakes, gives real-time on-line feedback for children’s writing, and helps teachers comprehend and evaluate children’s writing skills. The proposed method adjudges five main criteria: shape, direction, stroke order, position respect to the reference lines, and kinematics of the trace. It analyzes the handwriting quality and automatically gives feedback based on the combination of three extracted models: Beta-Elliptic Model (BEM) using similarity detection (SD) and dissimilarity distance (DD) measure, Fourier Descriptor Model (FDM), and perceptive Convolutional Neural Network (CNN) with Support Vector Machine (SVM) comparison engine. The originality of our work lies partly in the system architecture which apprehends complementary dynamic, geometric, and visual representation of the examined handwritten scripts and in the efficient selected features adapted to various handwriting styles and multiple script languages such as Arabic, Latin, digits, and symbol drawing. The application offers two interactive interfaces respectively dedicated to learners, educators, experts or teachers and allows them to adapt it easily to the specificity of their disciples. The evaluation of our framework is enhanced by a database collected in Tunisia primary school with 400 children. Experimental results show the efficiency and robustness of our suggested framework that helps teachers and children by offering positive feedback throughout the handwriting learning process using tactile digital devices.

Predicting movie revenue before committing significant investments

ABSTRACT A movie originates at the development stage when investors evaluate multiple scripts, attach key cast and crew, and shape the budget. Investors prefer to consider various scenarios before committing a large sum of money when selecting the key cast and crew. An accurate revenue prediction at this early stage reduces the risk of investment. We use a little information available at the early stage to predict movie revenue to enhance potential investors’ decision-making before greenlighting the project. In this study, we develop a movie revenue prediction model using multiple predictors, which are essentially based on key cast and crew, the storyline (movie genre), and the distinctive characteristics of the movie, which are available at the development stage. Our study applies multiple machine learning algorithms to predict movie revenue and demonstrates significant improvement over traditional statistical regression methods. The most accurate machine learning algorithm reduces the forecast error of total and opening-week box office revenue by 10.02% and 9.20%, respectively, over the benchmark multivariate linear regression method. Our results exhibit that including a movie’s distinctive characteristics reduces the average forecast error of total and opening-week box office revenue across all algorithms by 9.22% and 8.43%, respectively.

How Do Individuals Understand Multiple Conceptual Modeling Scripts?

Because most real-world domains intended to be supported by an information system are complex, practitioners often use multiple different types of conceptual modeling scripts to understand them. We performed two experiments to examine how two theoretical factors of multiple scripts— combined ontological completeness and ontological overlap—influence how users develop an understanding of a real-world domain from multiple scripts. Results of the first experiment show that to some degree, ontological overlap improves participants’ understanding of a domain, more so than combined ontological completeness. In the second experiment, we tracked the eye movement data of participants to understand how ontological overlap between scripts impacts users’ information search and cognitive integration processes. We found that some occurrence of semantically similar constructs between scripts helps individuals to identify and relate constructs presented in different scripts. Users, therefore, can identify and focus on script areas that are relevant to their problem tasks. However, a high level of ontological overlap decreases the attention paid by participants to relevant task-specific areas because they spend more time searching for relevant information. Together, our findings both refine and extend existing conceptual modeling theory. We clarify the dialectics between the full and parsimonious real-world representations offered through multiple scripts and the individual’s understanding of the domain that is represented by those scripts.

Is Hanja represented in the Korean mental lexicon?: Encoding cross-script semantic cohorts in the representation of Sino-Korean

Korean can be transcribed in two different scripts, one alphabetic (Hangul) and one logographic (Hanja). How does the mental lexicon represent the contributions of multiple scripts? Hangul's highly transparent one-to-one relationship between spellings and sounds creates homophones in spoken Korean that are also homographs in Hangul, which can only be disambiguated through Hanja. We thus tested whether native speakers encoded the semantic contributions of the different Hanja characters sharing the same homographic form in Hangul in their mental representation of Sino-Korean. Is processing modulated by the number of available meanings, that is, the size of the semantic cohort? In two cross-modal lexical decision tasks with semantic priming, participants were presented with auditory primes that were either syllables (Experiment 1) or full Sino-Korean words (Experiment 2), followed by visual Sino-Korean full word targets. In Experiment 1, reaction times were not significantly modulated by the size of the semantic cohort. However, in Experiment 2, we observed significantly faster reaction times for targets preceded by primes with larger semantic cohorts. We discuss these findings in relation to the structure of the mental lexicon for bi-scriptal languages and the representation of semantic cohorts across different scripts.

Understanding movie poster: transfer-deep learning approach for graphic-rich text recognition

Graphic-rich texts are common in posters. In a movie poster, information, such as movie title, tag lines, and names of the actors, director, and production house, is available. Graphic-rich texts in movie titles represent not only sentiments but also their genre. Understanding the poster requires graphic-rich text recognition. Prior to that, one requires text localization, so background and foreground graffiti can be well segmented. In this paper, we propose a transfer learning-based approach for graphic-rich text localization, which was tuned by introducing reverse augmentation and rotated/inclined rectangle drawing technique. A convolution neural network-based model is then applied to identify their corresponding scripts. In our experiments, on a newly developed dataset (available upon request) that is composed of movie posters with multiple scripts of 1154 images, we achieved an average accuracy of 99.30%. Our results outperformed previously developed tools that are relying on handcrafted features.

Texture feature column scheme for single‐ and multi‐script writer identification

Identification of writers from images of handwriting is an interesting research problem in the handwriting recognition community. Application of image analysis and machine learning techniques to this problem allows development of computerised solutions which can facilitate forensic experts in reducing the search space against a questioned document. This article investigates the effectiveness of textural measures in characterising the writer of a handwritten document. A novel descriptor by crossing the local binary patterns (LBP) with different configurations that allows capturing the local textural information in handwriting using a column histogram is introduced. The representation is enriched with the oriented Basic Image Features (oBIFs) column histogram. Support vector machine (SVM) is employed as the classifier, and the experimental study is carried out on five different datasets in single as well as multi-script evaluation scenarios. Multi-script evaluations allow evaluating the hypothesis that writers share common characteristics across multiple scripts and the reported results validate the effectiveness of textural measures in capturing this script-independent, writer-specific information.

Automatic Reverse Engineering of Script Engine Binaries for Building Script API Tracers

Script languages are designed to be easy-to-use and require low learning costs. These features provide attackers options to choose a script language for developing their malicious scripts. This diversity of choice in the attacker side unexpectedly imposes a significant cost on the preparation for analysis tools in the defense side. That is, we have to prepare for multiple script languages to analyze malicious scripts written in them. We call this unbalanced cost for script languages asymmetry problem . To solve this problem, we propose a method for automatically detecting the hook and tap points in a script engine binary that is essential for building a script Application Programming Interface (API) tracer. Our method allows us to reduce the cost of reverse engineering of a script engine binary, which is the largest portion of the development of a script API tracer, and build a script API tracer for a script language with minimum manual intervention. This advantage results in solving the asymmetry problem. The experimental results showed that our method generated the script API tracers for the three script languages popular among attackers (Visual Basic for Applications (VBA), Microsoft Visual Basic Scripting Edition (VBScript), and PowerShell). The results also demonstrated that these script API tracers successfully analyzed real-world malicious scripts.

Benchmarked multi-script Thai scene text dataset and its multi-class detection solution

Detecting text portion from scene images can be found to be one of the prevalent research topics. Text detection is considered challenging and non-interoperable since there could be multiple scripts in a scene image. Each of these scripts can have different properties, therefore, it is crucial to research the scene text detection based on the geographical location owing to different scripts. As no work on large-scale multi-script Thai scene text detection is found in the literature, the work conducted in this study focuses on multi-script text that includes Thai, English (Roman), Chinese or Chinese-like script, and Arabic. These scripts can generally be seen around Thailand. Thai script contains more consonants, vowels, and has numerals when compared to the Roman/ English script. Furthermore, the placement of letters, intonation marks, as well as vowels, are different from English or Chinese-like script. Hence, it could be considered challenging to detect and recognise the Thai text. This study proposed a multi-script dataset which includes the aforementioned scripts and numerals, along with a benchmarking employing Single Shot Multi-Box Detector (SSD) and Faster Regions with Convolutional Neural Networks (F-RCNN). The proposed dataset contains scene images which were recorded in Thailand. The dataset consists of 600 images, together with their manual detection annotation. This study also proposed a detection technique hypothesising a multiscript scene text detection problem as a multi-class detection problem which found to work more effective than legacy approaches. The experimental results from employing the proposed technique with the dataset achieved encouraging precision and recall rates when compared with such methods. The proposed dataset is available upon email request to the corresponding authors.

Multi-script morphological transducers and transcribers for seven Turkic languages

This paper describes ongoing work to augment morphological transducers for seven Turkic languages with support for multiple scripts each, as well as preliminary work adding IPA transcription systems. Evaluation demonstrates that our approach yields coverage equivalent to or not much lower than that of the base transducers.

A diachronic analysis of Sindhi multiscriptality

Abstract Recent debates in modern India on which script to use for the Sindhi language in the country present the situation as a binary choice between Perso-Arabic and Devanagari. However, such debates almost always fail to take into account the fact that the Sindhi language has, for most of its written history, been written in multiple scripts by different user groups and for different purposes. This study investigates the rich history of multiscriptality in the Sindhi language by analysing data on the use of various scripts for the language from the tenth century to the present day. I show that, historically, the Sindhi community chose and used scripts based on utilitarian and pragmatic principles, rather than on ideology or prejudice. I also demonstrate that script choice in the Sindhi community was determined by religio-occupational needs, one’s gender affiliation as well as the purpose and function of writing. In doing so, I argue that the case of Sindhi multiscriptality makes significant contributions to our understanding of the sociolinguistics of writing, of script choice, and of the paradigm of biscriptality (Bunčić, Daniel, Sandra L. Lippert & Achim Rabus (eds.). 2016. Biscriptality: A sociolinguistic typology. Heidelberg: Universitätsverlag Winter).