Browsing by Author "Anjum, Anika"

Now showing 1 - 6 of 6

A study on Nestle start healthy stay healthy Bangladesh: Is it successful enough at educating the parents on children's health?
(BRAC University, 2022-06) Anjum, Anika; Choudhury, Ahmed Abir
This report thoroughly explores all the departments that work in a coordinative manner to obtain the organizational goals successfully. I made an effort to summarize my experience, knowledge, and contribution to NBL. With a sophisticated research and development division, as well as a thorough logistics, manufacturing, and procurement control team, Nestlé is renowned as a corporation that focuses on nutrition, well-being, and wellness and offers the highest-quality goods. I have tried to capture how the nutrition department in NBL operates, what are the common methods of marketing and promoting the brands under this specific department and how exact the nontraditional method of marketing works here. I have also emphasized on how the Facebook page Nestlé Start Healthy Stay Healthy operates and who this page targets to promote the nutritional information of children’s diet. Moreover, I have conducted thorough research with survey questionnaire on the effectiveness of the aforementioned media, how the parents perceive this method of subtle marketing. This paper includes an overview and the analysis of how much social media pages can educate a certain target group of parents in Bangladesh. At the very end, I have provided some recommendations as to how the page can improve, how new strategies can be implemented in order to successfully provide authentic information to the parents in a way they enthusiastically accept and also knows about the brand.
Automated Phrasal Verb and Key-Phrase Checking with LSTM-Based Attention Mechanism
(MDPI Publications, 2023-09-13) Tusher, Abdur Nur; Anjum, Anika; Anjum, Anika; Das, Shudipta; Sammy, Mst. Sakira Rezowana; Hossain, Dr. Gahangir
Text prediction and classification are crucial tasks in modern Natural Language Processing (NLP) techniques. Long short-term memory (LSTM), a type of Recurrent Neural Network (RNN), is well-known for its outstanding performance in text classification. Phrasal verbs, also known as Bagdhara in Bangla, play a vital role in making language more expressive and poetic in any language, including Bangla. These two or three-word phrases help us convey our emotions and thoughts more effectively. However, determining whether a phrase is a phrasal verb and appropriate for a given context can be challenging for writers, poets, and the general public. To address this issue, an automatic system capable of identifying and using phrasal verbs is necessary. In this study, we propose a system that can instantly and accurately predict phrasal verbs using the LSTM algorithm, a part of the RNN, and an attention mechanism. Our system achieved an overall phrasal verb prediction accuracy of 78.63%.
The endogeneity of domestic violence: understanding women empowerment through autonomy
(© 2016 Elsevier Ltd., 2016-10) Fakir, Adnan M.S.; Anjum, Anika; Bushra, Fabiha; Nawar, Nabilah
Women’s autonomy is known to incite intimate partner violence (IPV) in developing countries. We argue for the endogeneity of women’s autonomy with IPV, which is often ignored in the existing literature, for understanding the causal association. Using the Bangladesh Demography and Health Survey (2007), we isolate the effect of autonomy on IPV taking into account the possible endogeneity using instrument variables and special regressor estimations, proposing the special regressor as a more reliable approach to estimating binary choice models with discrete endogenous regressors. Our study finds increased women’s autonomy to lead to higher incidents of IPV for a South Asian patriarchal society such as Bangladesh. Thereby, policies catered towards women empowerment by increasing women’s autonomy should concurrently focus on other determinants of IPV. Undue male controlling behaviour, witnessing inter-parental abuse as a child and early marriage are also found to aggravate IPV. Hence there should be simultaneous programmes that aim to relax male controlling behaviour over women, provide counselling for those who have witnessed inter-parental abuse as a child, and laws prohibiting child marriage, still prevalent in South Asian societies, should be reinforced. These findings have strong policy implications suggesting the dual nature of improving women’s autonomy in empowering women while aggravating IPV.
Zero- and Few-Shot Prompting with LLMs:
(Daffodil International University, 2024-05-30) Hasan, Md. Arid; Das, Shudipta; Anjum, Afiyat; Alam, Firoj; Anjum, Anika; Sarke, Avijit; Noori, Sheak Rashed Haider
The rapid expansion of the digital world has propelled sentiment analysis into a critical tool across diverse sectors such as marketing, politics, customer service, and healthcare. While there have been significant advancements in sentiment analysis for widely spoken languages, low-resource languages, such as Bangla, remain largely under-researched due to resource constraints. Furthermore, the recent unprecedented performance of Large Language Models (LLMs) in various applications highlights the need to evaluate them in the context of low-resource languages. In this study, we present a sizeable manually annotated dataset encompassing 33,606 Bangla news tweets and Facebook comments. We also investigate zero- and few-shot in-context learning with several language models, including Flan-T5, GPT-4, and Bloomz, offering a comparative analysis against fine-tuned models. Our findings suggest that monolingual transformer-based models consistently outperform other models, even in zero and few-shot scenarios. To foster continued exploration, we intend to make this dataset and our research tools publicly available to the broader research community.
Zero- and Few-Shot Prompting with LLMs: A Comparative Study with Fine-tuned Models for Bangla Sentiment Analysis
(Scopus, 2024) Hasan, Md. Arid; Das, Shudipta; Anjum, Afiyat; Alam, Firoj; Anjum, Anika; Sarker, Avijit; Noori, Sheak Rashed Haider
The rapid expansion of the digital world has propelled sentiment analysis into a critical tool across diverse sectors such as marketing, politics, customer service, and healthcare. While there have been significant advancements in sentiment analysis for widely spoken languages, low-resource languages, such as Bangla, remain largely under-researched due to resource constraints. Furthermore, the recent unprecedented performance of Large Language Models (LLMs) in various applications highlights the need to evaluate them in the context of low-resource languages. In this study, we present a sizeable manually annotated dataset encompassing 33,606 Bangla news tweets and Facebook comments. We also investigate zero- and few-shot in-context learning with several language models, including Flan-T5, GPT-4, and Bloomz, offering a comparative analysis against fine-tuned models. Our findings suggest that monolingual transformer-based models consistently outperform other models, even in zero and few-shot scenarios. To foster continued exploration, we intend to make this dataset and our research tools publicly available to the broader research community.
Zero- and Few-Shot Prompting with LLMs: A Comparative Study with Fine-tuned Models for Bangla Sentiment Analysis
(Elsevier, 2024-05-25) Hasan, Md. Arid; Das, Shudipta; Anjum, Afiyat; Alam, Firoj; Anjum, Anika; Sarker, Avijit; Sheak Rashed Haider; Noori, Sheak Rashed Haider
The rapid expansion of the digital world has propelled sentiment analysis into a critical tool across diverse sectors such as marketing, politics, customer service, and healthcare. While there have been significant advancements in sentiment analysis for widely spoken languages, low-resource languages, such as Bangla, remain largely under-researched due to resource constraints. Furthermore, the recent unprecedented performance of Large Language Models (LLMs) in various applications highlights the need to evaluate them in the context of low-resource languages. In this study, we present a sizeable manually annotated dataset encompassing 33,606 Bangla news tweets and Facebook comments. We also investigate zero- and few-shot in-context learning with several language models, including Flan-T5, GPT-4, and Bloomz, offering a comparative analysis against fine-tuned models. Our findings suggest that monolingual transformer-based models consistently outperform other models, even in zero and few-shot scenarios. To foster continued exploration, we intend to make this dataset and our research tools publicly available to the broader research community.