ECNU at SemEval-2018 Task 2: Leverage Traditional NLP Features and Neural Networks Methods to Address Twitter Emoji Prediction Task

Proceedings of the 12th International Workshop on Semantic Evaluation (SemEval-2018), pages 433–437 New Orleans, Louisiana, June 5–6, 2018. ©2018 Association for Computational Linguistics ECNU at SemEval-2018 Task 2: Leverage Traditional NLP Features and Neural Networks Methods to Address Twitter Emoji Prediction Task Xingwu Lu1, Xin Mao1, Man Lan1,2∗, Yuanbin Wu1,2 1Department of Computer Science and Technology, East China Normal University, Shanghai, P.R.China 2Shanghai Key Laboratory of Multidimensional Information Processing {51174506023, 10131530334}@stu.ecnu.edu.cn {mlan, ybwu}@cs.ecnu.edu.cn Abstract This paper describes our submissions to Task 2 in SemEval 2018, i.e., Multilingual Emoji Pre- diction. We first investigate several traditional Natural Language Processing (NLP) features, and then design several deep learning model- s. For subtask 1: Emoji Prediction in English, we combine two different methods to repre- sent tweet, i.e., supervised model using tradi- tional features and deep learning model. For subtask 2: Emoji Prediction in Spanish, we on- ly use deep learning model. 1 Introduction Visual icons play a crucial role in providing infor- mation about the extra level of social media infor- mation. SemEval 2018 shared task for researcher- s to predict, given a tweet in English or Spanish, its most likely associated emoji (Barbieri et al., 2018, 2017) (Task 2, Multilingual Emoji Predic- tion), which is organized into two optional subtask (subtask 1 and subtask 2) respectively in English and Spanish. For subtask 1, we adopt a combination model to predict emojis, which consists of traditional Natu- ral Language Processing (NLP) methods and deep learning methods. The results returned by the clas- sifier with traditional NLP features, by the neural network model and by the combination model are voted to get the final result. For subtask 2, we only use deep learning model. 2 System Description For subtask 1, we explore three different methods i.e., using traditional NLP features to learn a su- pervised machine learning-based classifier, learn- ing a deep learning model to make prediction and combine features captured by neural networks with traditional NLP features to train a supervised machine learning-based classifier. For subtask 2, we simply implement deep learning method to make prediction. 2.1 Traditional NLP Features In this task, we extract the following three types of features to capture effective information from the given tweets, i.e., linguistic features, sentiment lexicon features and tweet specific features. 2.1.1 Linguistic Features • N-grams: We extract 3 types of Bag-of- Words features as N-grams features, where N = 1,2,3 (i.e., unigram, bigram, and trigram features). • POS: Generally, the sentences carrying sub- jective emotions are inclined to contain more adjectives and adverbs while the sentences without sentiment orientation would contain more nouns. Thus, we extract POS tag from the sentence as features with the Bag-of- Words form. • Correlation Degree: For each word appear in training data, the ratio of the number of oc- currences under each class and the total oc- currences is counted as the correlation degree of the word to a certain class. When the fea- ture is created, the sum of the correlation de- gree of words in …

Content truncated to 500 words