Probabilistic Retrieval

This lecture introduces Probabilistic Information Retrieval, focusing on modeling relevance as a probability using the Query Likelihood Model and Language Modeling. It covers concepts like smoothing, learning and using the model, query expansion, and automatic thesaurus generation. The instructor discusses the Rocchio Algorithm, User Relevance Feedback, and SMART algorithm for practical relevance feedback. The lecture also explores weighting schemes, query expansion methods, and the challenges of query drift. Various examples and algorithms are presented to illustrate the concepts discussed.