Speaker Series

5/23

Chris Olston, Yahoo! Research  Website

6/7-6/8

Ed Hovy, ISI-USC  Website

6/14

Josef Ruppenhoffer, Pitt  Website

Dr. Ruppenhofer will give a talk and a related tutorial, both of which are open to the public.

Title: Manual and Automatic Subjectivity and Sentiment Analysis

Subjectivity analysis focuses on the expression of emotions, evaluations, and sentiments in language. This tutorial will cover:

• problem definitions (e.g., what is subjectivity?) and manual annotations;
• methods for identifying opinion-bearing words and phrases (lexicon development)
• methods for identifying polarity/orientation (positive, negative, or neutral) of expressions in context;
• applications of subjectivity analysis, with an emphasis on project review mining.

The tutorial session will be a working session involving manual annotations. Note that we will only touch on subjectivity and sentiment classification at the document level, and will focus on fine-grained analysis at the sentence, phrase, word, and word-sense levels.

6/25

Deva Ramanan, TTI  Website

If you would like to schedule a meeting with Deva on Tuesday, 6/26, please send email to: heeren@cs.uiuc.edu.

7/5

Dan Roth, MIAS-UIUC  Website

Dr. Roth will give a talk entitled: Global Learning with Constraints

Abstract: The maturity of machine learning techniques allows us today to learn many low level predicates and generate an appropriate vocabulary over which reasoning methods can be used to make significant progress in higher level domain decisions.

I will describe research on a framework that combines learning and inference and exhibit its use in the natural language processing domain. Key in this framework is the ability to incorporate declarative and expressive global information into the learning and decision stage. I will discuss the use of this framework as (1) a way to allow the output of local classifiers for different problem components to be assembled into a whole that reflects global preferences and constraints; (2) a way to improve probabilistic models by enforcing additional expressive constraints and (3) a way to significantly improve semi-supervised training of structured models.

Examples will be drawn from 'wh' attribution in natural language processing (determining who did what to whom when and where) and from information extraction problems.

Bio: Dan Roth is a Professor in the Department of Computer Science at the University of Illinois at Urbana-Champaign and the Beckman Institute of Advanced Science and Technology (UIUC) and a Willett Faculty Scholar of the College of Engineering. He is also the Director of MIAS, a DHS Institute of Discrete Science Center for Multimodal Information Access & Synthesis.

Roth has published broadly in machine learning, natural language processing, knowledge representation and reasoning and has developed advanced machine learning based tools for natural language applications that are being used widely by the research community. Among his paper awards are the best paper award in IJCAI-99 and the 2001 AAAI Innovative Applications of AI Award. Roth was the program chair of CoNLL'02 and of ACL'03 and is an associate editor for JAIR and the Machine Learning Journal. Roth got his Ph.D. in Computer Science from Harvard University in 1995.

7/6

Tina Eliassi-Rad, LLNL Website

Dr. Eliassi-Rad will give a talk entitled: Leveraging Network Structure to Infer Missing Values in Relational Data

Abstract: Inference techniques for relational data improve classification performance by exploiting dependencies between attributes of related instances. In particular, a great deal of recent attention has been paid to collective inference procedures, which make simultaneous inferences over attributes of related instances. Collective inference has been shown to be particularly effective for overcoming substantial amounts of missing attribute information. We propose a novel approach for inference in relational data, which leverages information about the relational network structure. We show that when structural characteristics are informative, our approach leads to consistent, and sometimes dramatic, improvement in classification performance regardless of the amount of attribute information available. We demonstrate the utility of our method on several real-world classification tasks. Interestingly, for many of these tasks, collective inference does not perform well, apparently due to low amounts of relational autocorrelation. Understanding data characteristics that influence collective inference is a largely unexplored area for further study. This work is joint work with Brian Gallagher (Lawrence Livermore National Laboratory) and Lise Getoor (University of Maryland).

Bio: Tina Eliassi-Rad is a computer scientist and a technical lead at the Center for Applied Scientific Computing at Lawrence Livermore National Laboratory. She earned her Ph.D. at UW-Madison in 2001, her M.S. at UIUC in 1995, and her B.S. at UW-Madison in 1993. All of her degrees are in computer science. Her research interests include artificial intelligence, machine learning, knowledge discovery and data mining. Her work has been applied to the World-Wide Web, scientific simulation data, and complex networks. For more details, visit http://www.cs.wisc.edu/~eliassi/.

7/9

Patrick Pantel, ISI-USC  Website

Dr. Pantel will give a 3 hour talk and tutorial which is open to the public:

Title: Lexical semantics and large-scale similarity modeling

Abstract: In this tutorial, we will explore recent explorations in computational lexical semantics, using corpus-based and web-based techniques, and unsupervised and semi-supervised learning methods. With a focus on similarity modeling, we will learn the art of mapping problem statements to feature representations, information theoretic feature weighting, comparison measures, and clustering algorithms. We will apply this framework to automatically learn the concepts in a textual corpus, the senses of words, the topics in a collection of documents, paraphrases, and even detecting aliases and groups of related individuals in a homeland security setting. We will also explore Google's famed MapReduce infrastructure for seamless very large-scale data processing, introducing open-source efforts under way for making this technology to the public.

Bio: Patrick Pantel is currently a Research Assistant Professor and Research Scientist in the Natural Language Group at the USC Information Sciences Institute where he does research in large-scale natural language processing, ontology learning, text mining, knowledge acquisition, and predictive systems. In 2003, he received a Ph.D. in Computing Science from the University of Alberta in Edmonton, Canada.

7/10

Anhai Doan, University of Wisconsin  Website

Dr. Doan's talk is open to the public.

Title: The Cimple Project on Community Information Management

Abstract: In this talk I will give an overview of Cimple, a joint project between the University of Wisconsin-Madison and Yahoo! Research. Cimple develops a generic solution that crawls, extracts, and integrates data, to build structured "portals" for online communities. I will first describe the envisioned working of Cimple and our prototype, DBlife, which is a structured portal being developed for the database research community. Next, I describe the technical challenges underlying Cimple and our solution approaches. Finally, I discuss the connections between Cimple and research in data integration, information extraction, human computation, and Web data management. More information about Cimple can be found at http://www.cs.wisc.edu/~anhai/projects/cimple

Bio: AnHai Doan is an assistant professor in Computer Science at the University of Wisconsin-Madison, since July 2006. His interests cover databases, AI, and Web. His current research focuses on data integration, Web community management, mass collaboration, text management, information extraction, and schema and ontology matching. Selected recent honors include the ACM Doctoral Dissertation Award in 2003, CAREER Award in 2004, and Alfred P. Sloan Research Fellowship in 2007. Selected recent professional activities include co-chairing WebDB at SIGMOD-05 and the AI Nectar track at AAAI-06.