560 likes | 815 Views
Generative Topic Models for Community Analysis . Ramesh Nallapati. Objectives. Provide an overview of topic models and their learning techniques Mixture models, PLSA, LDA EM, variational EM, Gibbs sampling Convince you that topic models are an attractive framework for community analysis
E N D
Generative Topic Models for Community Analysis Ramesh Nallapati
Objectives • Provide an overview of topic models and their learning techniques • Mixture models, PLSA, LDA • EM, variational EM, Gibbs sampling • Convince you that topic models are an attractive framework for community analysis • 5 definitive papers 10-802: Guest Lecture
Outline • Part I: Introduction to Topic Models • Naive Bayes model • Mixture Models • Expectation Maximization • PLSA • LDA • Variational EM • Gibbs Sampling • Part II: Topic Models for Community Analysis • Citation modeling with PLSA • Citation Modeling with LDA • Author Topic Model • Author Topic Recipient Model • Modeling influence of Citations • Mixed membership Stochastic Block Model 10-802: Guest Lecture
Introduction to Topic Models • Multinomial Naïve Bayes • For each document d = 1,, M • Generate Cd ~ Mult( ¢ | ) • For each position n = 1,, Nd • Generate wn ~ Mult(¢|,Cd) C ….. WN W1 W2 W3 M b 10-802: Guest Lecture
Introduction to Topic Models • Naïve Bayes Model: Compact representation C C ….. WN W1 W2 W3 W M N b M b 10-802: Guest Lecture
Introduction to Topic Models • Multinomial naïve Bayes: Learning • Maximize the log-likelihood of observed variables w.r.t. the parameters: • Convex function: global optimum • Solution: 10-802: Guest Lecture
Introduction to Topic Models • Mixture model: unsupervised naïve Bayes model • Joint probability of words and classes: • But classes are not visible: C Z W N M b 10-802: Guest Lecture
Introduction to Topic Models • Mixture model: learning • Not a convex function • No global optimum solution • Solution: Expectation Maximization • Iterative algorithm • Finds local optimum • Guaranteed to maximize a lower-bound on the log-likelihood of the observed data 10-802: Guest Lecture
Introduction to Topic Models log(0.5x1+0.5x2) • Quick summary of EM: • Log is a concave function • Lower-bound is convex! • Optimize this lower-bound w.r.t. each variable instead 0.5log(x1)+0.5log(x2) X2 X1 0.5x1+0.5x2 H() 10-802: Guest Lecture
Introduction to Topic Models • Mixture model: EM solution E-step: M-step: 10-802: Guest Lecture
Introduction to Topic Models 10-802: Guest Lecture
Introduction to Topic Models • Probabilistic Latent Semantic Analysis Model d d • Select document d ~ Mult() • For each position n = 1,, Nd • generate zn ~ Mult( ¢ | d) • generate wn ~ Mult( ¢ | zn) Topic distribution z w N M 10-802: Guest Lecture
Introduction to Topic Models • Probabilistic Latent Semantic Analysis Model • Learning using EM • Not a complete generative model • Has a distribution over the training set of documents: no new document can be generated! • Nevertheless, more realistic than mixture model • Documents can discuss multiple topics! 10-802: Guest Lecture
Introduction to Topic Models • PLSA topics (TDT-1 corpus) 10-802: Guest Lecture
Introduction to Topic Models 10-802: Guest Lecture
Introduction to Topic Models • Latent Dirichlet Allocation • For each document d = 1,,M • Generate d ~ Dir(¢ | ) • For each position n = 1,, Nd • generate zn ~ Mult( ¢ | d) • generate wn ~ Mult( ¢ | zn) a z w N M 10-802: Guest Lecture
Introduction to Topic Models • Latent Dirichlet Allocation • Overcomes the issues with PLSA • Can generate any random document • Parameter learning: • Variational EM • Numerical approximation using lower-bounds • Results in biased solutions • Convergence has numerical guarantees • Gibbs Sampling • Stochastic simulation • unbiased solutions • Stochastic convergence 10-802: Guest Lecture
Introduction to Topic Models • Variational EM for LDA • Approximate the posterior by a simpler distribution • A convex function in each parameter! 10-802: Guest Lecture
Introduction to Topic Models • Gibbs sampling • Applicable when joint distribution is hard to evaluate but conditional distribution is known • Sequence of samples comprises a Markov Chain • Stationary distribution of the chain is the joint distribution 10-802: Guest Lecture
Introduction to Topic Models • LDA topics 10-802: Guest Lecture
Introduction to Topic Models • LDA’s view of a document 10-802: Guest Lecture
Introduction to Topic Models • Perplexity comparison of various models Unigram Mixture model PLSA Lower is better LDA 10-802: Guest Lecture
Introduction to Topic Models • Summary • Generative models for exchangeable data • Unsupervised models • Automatically discover topics • Well developed approximate techniques available for inference and learning 10-802: Guest Lecture
Outline • Part I: Introduction to Topic Models • Naive Bayes model • Mixture Models • Expectation Maximization • PLSA • LDA • Variational EM • Gibbs Sampling • Part II: Topic Models for Community Analysis • Citation modeling with PLSA • Citation Modeling with LDA • Author Topic Model • Author Topic Recipient Model • Modeling influence of Citations • Mixed membership Stochastic Block Model 10-802: Guest Lecture
Hyperlink modeling using PLSA 10-802: Guest Lecture
Hyperlink modeling using PLSA[Cohn and Hoffman, NIPS, 2001] • Select document d ~ Mult() • For each position n = 1,, Nd • generate zn ~ Mult( ¢ | d) • generate wn ~ Mult( ¢ | zn) • For each citation j = 1,, Ld • generate zj ~ Mult( ¢ | d) • generate cj ~ Mult( ¢ | zj) d d z z w c N L M g 10-802: Guest Lecture
Hyperlink modeling using PLSA[Cohn and Hoffman, NIPS, 2001] PLSA likelihood: d d z z New likelihood: w c N L M g Learning using EM 10-802: Guest Lecture
Hyperlink modeling using PLSA[Cohn and Hoffman, NIPS, 2001] Heuristic: (1-) 0 ·· 1 determines the relative importance of content and hyperlinks 10-802: Guest Lecture
Hyperlink modeling using PLSA[Cohn and Hoffman, NIPS, 2001] • Experiments: Text Classification • Datasets: • Web KB • 6000 CS dept web pages with hyperlinks • 6 Classes: faculty, course, student, staff, etc. • Cora • 2000 Machine learning abstracts with citations • 7 classes: sub-areas of machine learning • Methodology: • Learn the model on complete data and obtain d for each document • Test documents classified into the label of the nearest neighbor in training set • Distance measured as cosine similarity in the space • Measure the performance as a function of 10-802: Guest Lecture
Hyperlink modeling using PLSA[Cohn and Hoffman, NIPS, 2001] • Classification performance content Hyperlink Hyperlink content 10-802: Guest Lecture
Hyperlink modeling using LDA 10-802: Guest Lecture
Hyperlink modeling using LDA[Erosheva, Fienberg, Lafferty, PNAS, 2004] a • For each document d = 1,,M • Generate d ~ Dir(¢ | ) • For each position n = 1,, Nd • generate zn ~ Mult( ¢ | d) • generate wn ~ Mult( ¢ | zn) • For each citation j = 1,, Ld • generate zj ~ Mult( . | d) • generate cj ~ Mult( . | zj) z z w c N L M g Learning using variational EM 10-802: Guest Lecture
Hyperlink modeling using LDA[Erosheva, Fienberg, Lafferty, PNAS, 2004] 10-802: Guest Lecture
Author-Topic Model for Scientific Literature 10-802: Guest Lecture
Author-Topic Model for Scientific Literature[Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004] a P • For each author a = 1,,A • Generate a ~ Dir(¢ | ) • For each topic k = 1,,K • Generate fk ~ Dir( ¢ | ) • For each document d = 1,,M • For each position n = 1,, Nd • Generate author x ~ Unif(¢ | ad) • generate zn ~ Mult( ¢ | a) • generate wn ~ Mult( ¢ | fzn) a x z A w N M f b K 10-802: Guest Lecture
Author-Topic Model for Scientific Literature[Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004] a Learning: Gibbs sampling P x z A w N M f b K 10-802: Guest Lecture
Author-Topic Model for Scientific Literature[Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004] • Perplexity results 10-802: Guest Lecture
Author-Topic Model for Scientific Literature[Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004] • Topic-Author visualization 10-802: Guest Lecture
Author-Topic Model for Scientific Literature[Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004] • Application 1: Author similarity 10-802: Guest Lecture
Author-Topic Model for Scientific Literature[Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004] • Application 2: Author entropy 10-802: Guest Lecture
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05] 10-802: Guest Lecture
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05] Gibbs sampling 10-802: Guest Lecture
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05] • Datasets • Enron email data • 23,488 messages between 147 users • McCallum’s personal email • 23,488(?) messages with 128 authors 10-802: Guest Lecture
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05] • Topic Visualization: Enron set 10-802: Guest Lecture
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05] • Topic Visualization: McCallum’s data 10-802: Guest Lecture
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05] 10-802: Guest Lecture
Modeling Citation Influences 10-802: Guest Lecture
Modeling Citation Influences[Dietz, Bickel, Scheffer, ICML 2007] • Copycat model 10-802: Guest Lecture
Modeling Citation Influences[Dietz, Bickel, Scheffer, ICML 2007] • Citation influence model 10-802: Guest Lecture
Modeling Citation Influences[Dietz, Bickel, Scheffer, ICML 2007] • Citation influence graph for LDA paper 10-802: Guest Lecture