joseph reisinger

Joseph Reisinger joeraii@cs.utexas.edu
@josephreisinger vita

I'm a PhD candidate in Computer Sciences at The University of Texas at Austin. My advisor is Ray Mooney. I'm currently supported by the Google PhD Fellowship in Natural Language Processing. I spent a wonderful 2 years as co-mastermind of FAI.

I'm currently in the Bay area. Let's collaborate on something!

Research Interests

I'm interested primarily in models of latent combinatorial structure in unlabeled data; particularly inferring semantic information automatically. My previous home was in theoretical evolutionary biology where I studied evolvability under randomized search schemes (here is a recent review).

My thesis is on Cross-cutting models of distributional lexical semantics. Current projects are: subspace clustering, parselets, and cross-cultural concept organization.

Diversions

I'm fascinated by inefficient markets; 日本語も昔から趣味だけど、中国語だとまだ「我是美国人！」ぐらい。

Software

vowpal_porpoise -- Lightweight python wrapper for vowpal_wabbit.
github

dist_lda -- Lightweight python implementation of a distributed, collapsed gibbs sampler for LDA. Uses redis to coordinate multiple nodes
github

Latent Variable Modeling Toolkit -- Collection of Gibbs samplers for Multinomial-Dirichlet latent variable models including DPMM, LDA, hLDA, tiered clustering, and cross-cutting categorization.
github

Spherical Admixture Model -- MATLAB code for a spherically embedded (L2-normalized) topic model; developed by Austin Waters.
matlab pdf

Lexical Semantics

Cross-Cutting Models of Lexical Semantics, with Ray Mooney EMNLP-2011.
pdf bibtex slides

Fine-Grained Class Label Markup of Search Queries, with Marius Pasca. ACL-2011.
pdf bibtex

A Mixture Model with Sharing for Lexical Semantics, with Ray Mooney. EMNLP-2010.
pdf bibtex slides

Multi-Prototype Vector-Space Models of Word Meaning, with Ray Mooney. NAACL-2010.
pdf bibtex slides

Information Extraction

Latent Variable Models of Concept-Attribute Attachment, with Marius Pasca. ACL-2009.
pdf bibtex slides poster

Bootstrapped Extraction of Class Attributes, with Marius Pasca. WWW-2009.
pdf bibtex

Weakly-Supervised Acquisition of Labeled Class Instances using Graph Random Walks.
Partha Pratim Talukdar, Joseph Reisinger, Marius Pasca, Deepak Ravichandran, Rahul Bhagat and Fernando Pereira. EMNLP-2008.
pdf bibtex slides