Information Retrieval

CS 6501 Fall 2014

Lecture Contents

Lecture I: Overview -- Course Introduction

We will highlight the basic structure and major topics of this course, and go over some logistic issues and course requirements.

Lecture II: Search Engine Architecture

We will briefly discuss the basic building blocks of a modern search engine system, including web crawler, inverted index, query processing, search result interface.

Lecture III: Retrieval Models

Retrieval model, a.k.a., ranking algorithm, is arguably the most important component of a retrieval system and directly determines search effectiveness. We will discuss classical retrieval models, including Boolean, vector space, probabilistic and language models. We will also introduce the most recent development of learning-based ranking algorithms, i.e., learning-to-rank.

Lecture IV: Retrieval Evaluation

Assessing the quality of deployed system is essential for retrieval system development. Many different measures for evaluating the performance of information retrieval systems have been proposed. We will discuss both the classical evaluation metrics, e.g., Mean Average Precision, and modern advance, e.g., interleaving.

Lecture V: Relevance Feedback

User feedback is important for retrieval systems to evaluate the performance and improve the effectiveness of their service strategies. However, in most practical system, only implicit feedback can be collected from users, e.g., clicks, which are known to be noisy and biased. We will discuss how to properly model implicit user feedback, and enhance retrieval performance via such feedback.

Lecture VI: Link analysis

We will discuss the unique characteristic of web: inter-connection, and introduce Google's winning algorithm PageRank. We will also introduce the application of link analysis techniques in a similar domain: social network analysis.

Lecture VII: Text mining

Text information takes a major portion of online information. Properly modeling text documents is essential for improving search effectiveness and discovering actionable knowledge. We will introduce basic text mining techniques, including text categorization, clustering and topic models.

Lecture VIII: IR applications

We will introduce modern applications in search systems, including recommendation, personalization, and online advertising, if time allows.