IITM BS BSDA5005: Introduction to Natural Language Processing (i-NLP)

Introduction to Natural Language Processing (i-NLP, BSDA5005) is a 4 credit elective at the degree level of the IITM BS in Data Science. The handbook places it at level 5. The course page lists Deep Learning (BSCS3004) as the prerequisite. It covers how computers handle human language, from basic linguistics and older statistical methods to transformers and LLMs.

The course is taught by Parameswari Krishnamurthy and Rahul Mishra, both Assistant Professors at the Language Technologies Research Centre, IIIT Hyderabad.

Handbook check. Two differences to know:

• The handbook's course table leaves the prerequisite blank. The course page lists Deep Learning. Plan as if Deep Learning is required.
• The course table (updated 18 March 2026) marks i-NLP as not offered in May 2026, September 2026 or January 2027. Check the registration page before you plan around it.

What you learn in i-NLP

Most weeks come with a code demo. The 12 weeks fall into four blocks.

• Weeks 1 and 2: language and text basics. Why human language is hard for computers, the levels of language processing and some linguistics. Then cleaning text: tokenising, normalising, removing stop words, stemming, lemmatising and analysing word forms.
• Weeks 3 to 5: the classical toolkit. Part-of-speech tagging and named entities with HMMs, MEMMs and CRFs, and text classification with Naive Bayes and logistic regression. Then parsing sentence structure and representing meaning. Then word meaning from context with n-grams, Word2Vec and GloVe, linking pronouns to what they refer to, and machine translation.
• Weeks 6 to 8: neural NLP and LLMs. RNNs, LSTMs and GRUs, contextual word embeddings and attention. Then transformers, pre-trained models like BERT and GPT, and fine-tuning. Then LLMs: LoRA and QLoRA, in-context learning, instruction tuning, RLHF and DPO.
• Weeks 9 to 12: generation, uses and responsibility. Decoding methods such as greedy, top-k and top-p sampling. Then applications like question answering, summarisation, fact-checking and retrieval-augmented generation. Then ways to explain model decisions, and a final week on ethics, culture and bias.

The suggested books include Speech and Language Processing by Jurafsky and Martin, and books by Manning and Schuetze, Eisenstein, and Goldberg.

How the course is assessed

The course page points to the standard course structure on the programme's Academics page. It does not list a separate assessment plan.

Where it counts

• BS level stream rule. It is level 5 and tagged BD, which appears to mean the data science stream. See the BS degree level.
• Minors. It is not in any minor list in the handbook.

The handbook says level 5 courses are more complex than level 3 and level 4 courses, with the same credits and fee as level 4. See the full DS electives list for other BD options.

Who finds it hard and how to prepare

The course is very broad. Weeks 1 to 5 are more linguistics and probability than deep learning. If you expect only transformers, this part can feel slow and unfamiliar. Then weeks 6 to 8 move very fast through modern models.

• Revise conditional probability and Naive Bayes. HMMs and CRFs in week 3 build on them.
• Read the early chapters of Jurafsky and Martin on text processing and n-grams before the term.
• Run each code demo yourself and change the input text. Try an Indian language sentence and see where tools break.
• Make a timeline from n-grams to Word2Vec to LSTMs to transformers. Note what problem each step solved.
• Revise attention and transformers from Deep Learning (BSCS3004) before week 6.

What to take before and after

Deep Learning comes first. After i-NLP, Large Language Models (BSDA5004) goes deeper into transformers, training and scaling. Speech Technology (BSEE4001) is another BD elective if you want to work with spoken language.

Common questions

Why can I not find i-NLP on the registration page?

The March 2026 course table marks it as not offered in May 2026, September 2026 and January 2027. Courses also run only if enough students register. Ask support when it will run next.

Should I take i-NLP or LLM first?

i-NLP gives the wider base, from classical methods to transformers. LLM assumes you are ready to go deep on transformers alone. If both are open to you, i-NLP first is the gentler order.

Do I need to know linguistics?

No. Week 1 covers the linguistic basics you need. Expect terms like morphology, syntax and semantics, and give them some extra time.