Course

COM-308: Internet analytics

Lectures in this course (12)

Social and Information Networks: Structure

Explores the structure of social and information networks, focusing on giant components, clustering, tie formation, and network connectivity.

Social and Information Networks 2: Evolution

Delves into social and information networks dynamics, including herding behavior, information cascades, preferential attachment, and the Friendship Paradox.

Social and Information Networks: Processes

Covers computing measures over large networks, random walks, conductance bound, and epidemic models.

Social and Information Networks: Ranking

Explores the significance of ranking in networks, emphasizing algorithms like PageRank and HITS for web page ranking.

Dimensionality Reduction

Explores Singular Value Decomposition and Principal Component Analysis for dimensionality reduction, with applications in visualization and efficiency.

Recommender Systems: Overview and Methods

Explores the evolution of recommenders, collaborative filtering, Netflix Prize, model training, and optimization techniques.

Recommender Systems: Text Classification & Naïve Bayes

Explores text classification using Naïve Bayes in content-based recommenders.

Clusters and Communities

Explores clustering, community detection, K-means, GMM, modularity, and the Louvain method.

Text Models: Word Embeddings and Topic Models

Explores word embeddings, topic models, Word2vec, Bayesian Networks, and inference methods like Gibbs sampling.

Probabilistic Topic Models: Latent Dirichlet Allocation

Explores Latent Dirichlet Allocation, a probabilistic topic model for document clustering and analysis using distributions over words and topics.

Data Streams: Algorithms and Applications

Covers data streams, sub-linear memory computation, document similarity, and randomized dimension reduction techniques for handling 'Big Data' challenges efficiently.

Ad Auctions

Explores the online advertisement market, ad types, Google AdWords, auction formats, and bidding strategies.