# Department of Statistics Event Calendar

Back to Listing

## Statistics Seminar - Dr. Annie Qu (UIUC)

Event Type
Other
Department of Statistics
Location
106B1 Engineering Hall
Date
Mar 3, 2016   3:30 pm
Speaker
Annie Qu, Department of Statistics, University of Illinois at Urbana-Champaign
Views
80

Classification with unstructured predictors and an application to sentiment analysis

Abstract:
Unstructured data  refers to information that lacks certain  structures and cannot be organized in a predefined fashion. Unstructured data often involves words,  texts,  graphs, objects or  multimedia types of files that are difficult to process and analyze with traditional computational tools
and statistical methods. This  work explores  ordinal classification for  unstructured predictors with ordered class categories, where imprecise information concerning strengths of association between predictors is available for predicting class labels. However,  imprecise information  here is expressed in terms of a directed graph, with each node  representing a predictor and a directed edge  containing pairwise strengths of association between two nodes. One of the targeted applications  for unstructured data arises from sentiment analysis,  which identifies and extracts    the relevant content or opinion of a document  concerning a specific event of interest. We integrate the imprecise predictor relations into linear relational constraints over classification function coefficients,  where large margin ordinal classifiers are introduced, subject to many  quadratically linear constraints. The proposed classifiers are then applied in sentiment analysis using binary word predictors. Computationally, we implement ordinal support vector machines and $\psi$-learning through a scalable quadratic programming package based on sparse word representations. Theoretically, we show that utilizing  relationships  among  unstructured predictors improves prediction accuracy of classification significantly. We illustrate an application for sentiment analysis using consumer text reviews and movie review data. Supplementary materials for this article are available online. This is joint work with Junhui Wang, Xiaotong Shen and Yiwen Sun.