Yunong Wu

Bottom up: Exploring word emotions for Chinese sentence chief sentiment classification

In this paper we demonstrate the effectiveness of employing basic sentiment components for analyzing the chief sentiment of Chinese sentence among nine categories of sentiments (including “No emotion”). Compared to traditional lexicon based methods, our research explores emotion intensities of words and phrases in an eight dimensional sentiment space as features. An emotion matrix kernel is designed to evaluate inner product of these sentiment features for SVM classification with O(n) time complexity. Experimental result shows our method significantly improves performance of sentiment classification.

Exploring latent semantic information for textual emotion recognition in blog articles

Understanding people U+02BC s emotions through natural language is a challenging task for intelligent systems based on Internet of Things U+0028 IoT U+0029. The major difficulty is caused by the lack of basic knowledge in emotion expressions with respect to a variety of real world contexts. In this paper, we propose a Bayesian inference method to explore the latent semantic dimensions as contextual information in natural language and to learn the knowledge of emotion expressions based on these semantic dimensions. Our method synchronously infers the latent semantic dimensions as topics in words and predicts the emotion labels in both word-level and document-level texts. The Bayesian inference results enable us to visualize the connection between words and emotions with respect to different semantic dimensions. And by further incorporating a corpus-level hierarchy in the document emotion distribution assumption, we could balance the document emotion recognition results and achieve even better word and document emotion predictions. Our experiment of the wordlevel and the document-level emotion predictions, based on a well-developed Chinese emotion corpus Ren-CECps, renders both higher accuracy and better robustness in the word-level and the document-level emotion predictions compared to the state-of-theart emotion prediction algorithms.

Modification Relations Based Emotional Keywords Annotation Using Conditional Random Fields

In this study, we demonstrate the effectiveness of modification features for emotional keywords annotation on Chinese sentences extracted from a Chinese emotion corpus. Eight basic word emotion categories have been selected and Conditional Random Fields is employed in eight experiments with different sets of features. Our result shows that modification features make the contribution to emotional keywords annotation.

Emoticon-Based Emotion Analysis for Weibo Articles in Sentence Level

In this study, we propose a multi-label emotion study for Weibo articles in sentence level based on word and emoticon feature. We crawl articles from Weibo randomly, and extract sample sentences based on word emotion lexicon which has been constructed from a Chinese emotion corpus (Ren-CECps). Two machine learning methods including Support Vector Machine and Logistic Regression are employed to conduct the emotion classification experiments with the word feature and the combination feature of words and emoticons respectively. The significantly improved results given by classification experiments with the emoticon feature prove the effectiveness of taking emoticons in emotion analysis.

Learning Salient Samples and Distributed Representations for Topic-Based Chinese Message Polarity Classification

We describe our participation in the TopicBased Chinese Message Polarity Classification Task, based on the restricted and unrestricted resources respectively. In the restricted resource based classification, we focus on the selection of parameters in a multi-class classification model with highly-biased training data. In the unrestricted resource based classification, we explore the distributed representation of Chinese words through unsupervised feature learning and the annotation of salient samples through active learning, with a raw corpus of over 90 million messages extracted from Chinese Weibo Platform. For two classification subtasks, our submitted results ranked the 4th and the 2nd respectively.