This workshop aimed to equip participants with foundational and advanced knowledge in Artificial Intelligence (AI), Natural Language Processing (NLP), and Data Analytics, with a special focus on Information Retrieval and Extraction.
The key objectives of the course are to introduce core concepts in data, AI, and NLP; explore real-world applications of information retrieval and extraction; present essential methods and techniques for building intelligent applications; and discuss advanced AI and NLP approaches for practical deployment.
Information is knowledge, power, and value—and today, vast amounts are freely available on the Web. Social media, blogs, and online communities have fueled exponential growth in unstructured data across domains like commerce, education, and health. Extracting meaningful knowledge from this data is challenging since it’s designed for humans, not machines. While traditional methods rely on explicit keywords, much of the information—such as opinions and intentions—is implicit and hidden in latent semantics, making simple syntactic approaches ineffective. This seminar explores advanced AI techniques and reasoning frameworks that go beyond word-level analysis to transform unstructured text into structured, machine-readable data, enabling more powerful information retrieval and extraction across diverse fields.

With growing industry interest in chatbots, conversational AI has emerged as a rapidly advancing research area within Natural Language Processing (NLP), Machine Learning, and Deep Learning. The primary goal of conversational AI is to generate human-like dialogue, a challenging task due to the complexity of human conversation, including aspects like co-reference and context. Conversations generally fall into two categories: task-oriented and chit-chat (non-task oriented). Each type is influenced by different pragmatic factors such as topic, speaker personality, argumentation logic, intent, and viewpoint. Effectively modeling these factors is crucial for accurate conversational understanding and generation.
This advanced course explores state-of-the-art deep learning technologies and word representations—such as BERT, GPT-2, Seq2Seq, and Transformer architectures—for conversation analysis and generation. Beyond generation, the course delves into classification tasks including emotion recognition in conversation, emotion-driven dialogue generation, conversational question answering, and intent classification. It is designed to equip students, faculty, researchers, and practitioners with a solid understanding of both foundational and cutting-edge techniques in conversational AI.

The 17th International Conference on Natural Language Processing (ICON-2020) was held at IIT Patna from December 18–21, 2020. Serving as a key forum for researchers in Natural Language Processing (NLP) and Computational Linguistics (CL) from India and abroad, the conference featured the main sessions on December 19–20, preceded by pre-conference tutorials on December 18 and followed by post-conference workshops on December 21.
Click Here

The CEP on Deep Learning for Natural Language Processing provided an in-depth introduction to core NLP concepts, recent research trends, and hands-on training in machine learning and deep learning techniques. Topics included N-gram models, word sense disambiguation, parsing, POS tagging, sentiment analysis, machine translation, QA, NLU, and NLG. Emphasizing end-to-end neural models, the course explored how deep learning is reshaping NLP by reducing the need for manual feature engineering. Through lectures, tutorials, and practical sessions, participants gained the skills to design and implement their own NLP solutions.

Unsupervised data mining involves extracting meaningful patterns and structures from unlabeled data without prior knowledge of classes or categories. While batch mining algorithms operate on fixed datasets, stream mining algorithms are designed to handle continuous, high-velocity data streams, requiring incremental, memory-efficient, and adaptive methods. This shift enables real-time analysis and pattern detection in dynamic environments such as sensor networks, social media feeds, and financial transactions.

This topic focuses on quantifying and modeling the cognitive effort involved during the translation process. It examines factors such as attention, memory load, and decision-making complexity that affect translator performance. By analyzing behavioral and neurocognitive data, researchers develop computational models to better understand and predict the mental workload and challenges faced during translation production, aiming to improve translation tools and training.

Big Social Data Analysis involves processing and extracting insights from massive datasets generated by social media platforms, online communities, and digital interactions. It combines techniques from data mining, machine learning, and natural language processing to understand trends, sentiments, behaviors, and social dynamics at scale. This analysis helps in areas like marketing, public opinion monitoring, and crisis management by revealing patterns hidden within complex and high-velocity social data streams.

Neural Machine Translation (NMT) is an advanced approach to automated language translation that uses deep neural networks to model the entire translation process end-to-end. Unlike traditional phrase-based methods, NMT captures context and semantics more effectively by encoding the source sentence into a continuous vector representation and then decoding it into the target language. Techniques like sequence-to-sequence models, attention mechanisms, and Transformer architectures have significantly improved translation accuracy and fluency.

This topic explores the intersection of cognitive science and computational linguistics, focusing on how human cognitive processes—such as perception, memory, learning, and language understanding—inform the development of computational models for natural language processing. By studying how humans process and produce language, computational linguists can design more effective algorithms that mimic human-like language comprehension and generation.

Multi-Objective Optimization involves solving problems that have two or more conflicting objectives simultaneously. Instead of a single optimal solution, the goal is to find a set of trade-off solutions, known as the Pareto front, where no objective can be improved without worsening another. This approach is widely used in fields like engineering, economics, and machine learning to balance competing criteria effectively.

Natural Language Processing (NLP) is a field at the intersection of computer science, linguistics, and artificial intelligence that focuses on enabling machines to understand, interpret, and generate human language. NLP encompasses a variety of tasks such as text analysis, language modeling, part-of-speech tagging, parsing, sentiment analysis, machine translation, and question answering, aiming to bridge the gap between human communication and computer understanding.