English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Named Entity Recognition (NER) systems play a vital role in NLP applicationssuch as machine translation, summarization, and question-answering. Thesesystems identify named entities, which encompass real-world concepts likelocations, persons, and organizations. Despite extensive research on NERsystems for the English language, they have not received adequate attention inthe context of low resource languages. In this work, we focus on NER forlow-resource language and present our case study in the context of the Indianlanguage Marathi. The advancement of NLP research revolves around theutilization of pre-trained transformer models such as BERT for the developmentof NER models. However, we focus on improving the performance of shallow modelsbased on CNN, and LSTM by combining the best of both worlds. In the era oftransformers, these traditional deep learning models are still relevant becauseof their high computational efficiency. We propose a hybrid approach forefficient NER by integrating a BERT-based subword tokenizer into vanillaCNN/LSTM models. We show that this simple approach of replacing a traditionalword-based tokenizer with a BERT-tokenizer brings the accuracy of vanillasingle-layer models closer to that of deep pre-trained models like BERT. Weshow the importance of using sub-word tokenization for NER and present ourstudy toward building efficient NLP systems. The evaluation is performed onL3Cube-MahaNER dataset using tokenizers from MahaBERT, MahaGPT, IndicBERT, andmBERT.

On Significance of Subword tokenization for Low Resource and Efficient Named Entity Recognition: A case study in Marathi