arXiv Research
Abstract:India's linguistic landscape is one of the most diverse in the world, comprising over 120 majorlanguagesand approximately 1,600 additional…▽ MoreIndia's linguistic landscape is one of the most diverse in the world, comprising over 120 majorlanguagesand approximately 1,600 additionallanguages, with 22 officially recognized as scheduledlanguagesin the Indian Constitution. Despite recent pro
Open Full News
arXiv Research
Abstract:Hope speechlanguagethat fosters encouragement and optimism plays a vital role in promoting positive discourse online. However, its detection remains challenging, especially in multilingual and low-resource settings. This paper presents a multilingual framework for hope speech detection using an active learning approach and transformer-based…▽ MoreHope speechlanguagethat fosters encouragem
Open Full News
arXiv Research
Abstract:…as a critical task for promoting positive discourse and well-being. In this paper, we present a machine learning approach to multiclass hope speech detection across multiplelanguages, including English,…▽ MoreThe detection of hopeful speech in social media has emerged as a critical task for promoting positive discourse and well-being. In this paper, we present a machine learning approach
Open Full News
arXiv Research
Abstract:The use of derogatory terms inlanguagesthat employ code mixing, such as Roman…▽ MoreThe use of derogatory terms inlanguagesthat employ code mixing, such as RomanUrdu, presents challenges for NaturalLanguageProcessing systems due to unstated grammar, inconsistent spelling, and a scarcity of labeled data. In this work, we propose a QLoRA based fine tuning framework to improve offensivelangu
Open Full News
arXiv Research
Abstract:Benchmarks for largelanguage…▽ MoreBenchmarks for largelanguagemodels(LLMs) often rely on rubric-scented prompts that request visible reasoning and strict formatting, whereas real deployments demand terse, contract-bound answers. We investigate whether such "evaluation scent" inflates measured performance without commensurate capability gains. Using a single open-weightsmodel(GPT-OSS-20B)
Open Full News
arXiv Research
Abstract:Developing a high-performing largelanguage…▽ MoreDeveloping a high-performing largelanguagemodels(LLMs) for low-resourcelanguagessuch asUrdu, present several challenges. These challenges include the scarcity of high-quality datasets, multilingual inconsistencies, and safety concerns. Existing multilingual LLMs often address these issues by translating large volumes of available data. Howe
Open Full News
arXiv Research
Abstract:…among the most active users and often reveal aspects of their personal and professional lives through online posts. Platforms such as Twitter provide an opportunity to analyzelanguageand behavior for understanding demographic and social patterns. Since followers frequently share linguistic traits and interests with the celebrities they follow, textual data…▽ MoreSocial media has become a
Open Full News
arXiv Research
Abstract:VisionLanguage…▽ MoreVisionLanguageModels(VLMs) are pivotal for advancing perception in intelligent agents. Yet, evaluation of VLMs remains limited to predominantly English-centric benchmarks in which the image-text pairs comprise short texts. To evaluate VLM fine-grained abilities, in fourlanguagesunder long-text settings, we introduce a novel multilingual benchmark VLURes featuring eigh
Open Full News
arXiv Research
Abstract:Idiomatic translation remains a significant challenge in machine translation, especially for low resourcelanguagessuch as…▽ MoreIdiomatic translation remains a significant challenge in machine translation, especially for low resourcelanguagessuch asUrdu, and has received limited prior attention. To advance research in this area, we introduce the first evaluation datasets forUrduto English
Open Full News
arXiv Research
Abstract:…affective computing technology that enables emotionally intelligent artificial intelligence. While SER is challenging in general, it is particularly difficult for low-resourcelanguagessuch as…▽ MoreSpeech Emotion Recognition (SER) is a key affective computing technology that enables emotionally intelligent artificial intelligence. While SER is challenging in general, it is particularly d
Open Full News
arXiv Research
Abstract:Despite considerable progress in handwritten text recognition, paragraph-level handwritten text recognition, especially in low-resourcelanguages, such as Hindi,…▽ MoreDespite considerable progress in handwritten text recognition, paragraph-level handwritten text recognition, especially in low-resourcelanguages, such as Hindi,Urduand similar scripts, remains a challenging problem. Theselan
Open Full News
arXiv Research
Abstract:Automatic speech recognition for low-resourcelanguagesremains fundamentally constrained by the scarcity of labeled data and computational resources required by state-of-the-art…▽ MoreAutomatic speech recognition for low-resourcelanguagesremains fundamentally constrained by the scarcity of labeled data and computational resources required by state-of-the-artmodels. We present a systematic
Open Full News