International Journal of Computational Linguistics & Chinese Language
Vol. 23, No. 1, June 2018
Sentiment Analysis on Social Network: Using Emoticon Characteristics for Twitter Polarity Classification
Chia-Ping Chen, Tzu-Hsuan Tseng and Tzu-Hsuan Yang
In this paper, we describe a sentiment analysis system implemented for the semantic-evaluation task of message polarity classification for English on Twitter. Our system contains modules of data pre-processing, word embedding, and sentiment classification. In order to decrease the data complexity and increase the coverage of the word vector model for better learning, we perform a series of data pre-processing tasks, including emoticon normalization, specific suffix splitting, and hashtag segmentation. In word embedding, we utilize the pre-trained word vector provided by GloVe. We believe that emojis in tweets are important characteristics for Twitter sentiment classification, but most pre-trained sets of word vectors contain few or no emoji representations. Thus, we propose embedding emojis into the vector space by neural network models. We train the emoji vector with relevant words that contain descriptions and contexts of emojis. The models of long short-term memory (LSTM) and convolutional neural network (CNN) are used as our sentiment classifiers. The proposed emoji embedding is evaluated on the SemEval 2017 tasks. Using emoji embedding, we achieved recall rates of 0.652 with the LSTM classifier and 0.640 with the CNN classifier.
Keywords: Sentiment Analysis, Polarity Classification, Machine Learning, Neural Network, Word Embedding
A Lexical Coherence Representation Computational Framework using LSTM Forget Gate For Autism Recognition
Yu-Shuo Liu, Chin-Po Chen, Susan Shur-Fen Gau and Chi-Chun Lee
Autistic children are less able to tell a fluent story than typical children, so measuring verbal fluency becomes an important indicator when diagnosing autistic children. Fluency assessment, however, needs time-consuming manual tagging, or using expert specially designed characteristics as indicators, therefore, this study proposes a coherence representation learned by directly data-driven architecture, using the forget gate of long short-term memory model to export lexical coherence representation, at the same time, we also use the ADOS coding related to the evaluation of narration to test our proposed representation. Our proposed lexical coherence representation performs high accuracy of 92% on the task of identifying children with autism from typically development. Comparing with the traditional measurement of grammar, word frequency, and latent semantic analysis model, there is a significant improvement.
This paper also further randomly shuffles the word order and sentence order, making the typical child's story content become disfluent. By visualizing the data samples after dimension reduction, we further observe the distribution of these fluent, disfluent, and those artificially disfluent data samples. We found the artificially disfluent typical samples would move closer to disfluent autistic samples which prove that our extracted features contain the concept of coherency.
Behavioral Signal Processing, Lexical Coherence Representation, LSTM, Autism Spectrum Disorder, Story-telling
Joint Modeling of Individual Neural Responses using a Deep Voting Fusion Network for Automatic Emotion Perception Decoding
Wan-Ting Hsieh and Chi-Chun Lee
In the era underlying grouping life, affective computing and emotion recognition are closely bonding with daily life, and impose great impact on social ability. Understanding the individual differences is significant factor that should not be ignore in fMRI analysis while most of the brain studies on fMRI seldom truly deal with it, we carry out a system considering individual variability to recognize the emotion to the vocal stimuli with BOLD signal. In our work, we propose a novel method using multimodal fusion in a voting DNN framework, where we utilize a mask on weight matrix of fusion layer to learn an individual-influenced weight matrix and realize voting in this network, and achieve 53.10% in UAR for a four-class emotion recognition task. Our analysis shows that the multimodal voting net is an effective neural network encoding individual differences and thus enhances the ability to emotion recognition. Further the join of audio feature also boosts the result to 56.07%.
Keywords: Individual Difference, fMRI, Vocal Emotion, Perception, Deep Voting Fusion Neural Net