On 23rd April 2025, the IEEE RAS BRAC University Student Branch Chapter, hosted a engaging workshop titled “Talking to Machines: The Power of Natural Language Processing.” Held on campus, the event brought together technology enthusiasts to explore the fascinating world of Natural Language Processing (NLP) — a field at the intersection of computer science and linguistics. The event drew an enthusiastic crowd of approximately 140 participants and marked a significant step toward cultivating interest in theoretical and applied research in NLP.

The session commenced at 3:30 PM with opening remarks from the chapter advisor, Kholilur Rahman, who emphasized the pivotal role of IEEE and the IEEE Robotics and Automation Society (RAS) in fostering research and innovation. He highlighted the importance of advancing theoretical research alongside practical applications and introduced the purpose of IEEE RAS in expanding the frontier of robotics-related scholarship, including areas such as Natural Language Processing.

The first speaker of the day was Aryan Hossain, Lecturer at the Department of Computer Science and Engineering, BRAC University. He opened his session with a thought-provoking question: “Why teach machines to understand us?” This led into a discussion on what NLP is — an interdisciplinary field combining computer science and linguistics. Aryan Hossain explained how NLP handles various forms of human communication, including text, speech, and even sign language, providing compelling real-world examples.

He outlined three main challenges in NLP: ambiguity, sarcasm or irony, and linguistic variability, all of which make language a complex input for machines. His focus remained on text-based NLP, and he presented a clear NLP pipeline: data collection, text processing, text representation, model setup and evaluation, and finally, deployment.

In the “Text Processing” section, he explained the importance of data cleaning through techniques like lowercasing, special character removal, tokenization, and stop word elimination. He then transitioned into traditional methods of text representation such as Bag of Words and TF-IDF, emphasizing their role in quantifying word importance.

Moving on to modern techniques, Aryan Hossain introduced Word2Vec, which generates vector representations based on context. He broke down Skip-gram and CBOW models for intuitive understanding. He also discussed BERT, a contextual embedding model that dynamically assigns word vectors based on sentence context.

The session continued with an overview of key NLP tasks: text classification, machine translation, summarization, question answering, and domain-specific applications. Using clear diagrams, he walked the audience through sentiment analysis pipelines and machine translation systems, explaining the significance of sequence-to-sequence models and evaluation metrics like BLEU scores.

He wrapped up his session with an engaging two-minute interactive segment where students asked questions and shared reflections on the material.

At around 4:50 PM, Niloy Farhan sir, Adjunct Lecturer in the Department of CSE, BRAC University, took over to guide participants through the historical and modern evolution of NLP. He began with a timeline dating back to the 1950s, referencing the Georgetown-IBM experiment of 1954 and the development of ELIZA in 1966 — one of the first NLP-based programs. He identified pivotal years like 2017 and 2023 that marked significant breakthroughs in the NLP landscape.

Farhan sir elaborated on the shift from rule-based systems to statistical models, and finally to neural-network-based approaches including the rise of transformer architectures. He emphasized the LLM (Large Language Model) revolution with examples such as BERT, BART, and GPT, diving into their internal mechanics — from pre-training and instruction tuning to reinforcement learning from human feedback.

The audience was then introduced to autoregressive models, Retrieval-Augmented Generation (RAG), and real-world examples of retrieval-based methods. Farhan sir also discussed about Prompt Engineering and the concept of autonomous agents capable of planning, deciding, and executing complex tasks using NLP. He further connected NLP to the robotics domain through examples of language understanding and task execution.

Toward the end of his session, he addressed ethical considerations in NLP and challenges that arise with large-scale deployment. He concluded by offering actionable advice on how beginners can get started with NLP, including LLM project ideas tailored for newcomers.

As the session drew to a close, the workshop succeeded in bridging fundamental theory with modern applications of NLP. Participants were given a comprehensive overview of the field, sparking both curiosity and inspiration. The event concluded with an interactive photo session and tokens of appreciation for the speakers. Light refreshments were provided, and the organizers ensured all participants were acknowledged for their involvement and enthusiasm.

“Talking to Machines: The Power of Natural Language Processing” proved to be a dynamic and enlightening workshop that not only deepened participants’ understanding of NLP but also aligned with the broader mission of IEEE to foster innovation and interdisciplinary collaboration in engineering and technology.