A hybrid BERT-CNN-BiLSTM model classifies Bangla news headlines by topic and sentiment using the 9,014-headline BAN-ABSA dataset, marking a new NLP approach for low-resource Bengali news analysis.

#BanglaNLP #HeadlineClassification #SentimentAnalysis #LowResourceAI
https://arxiv.org/abs/2511.18618

Detecting Indian languages is harder than it looks—new paper drops baseline transformer models that tackle code‑mixing and low‑resource quirks. Curious how they fine‑tuned the AI? Dive in for the details! #LanguageID #CodeMixing #LowResourceAI

🔗 aidailypost.com/news/indian-...