Gundapusunil at SemEval-2020 Task 8: Multimodal Memotion Analysis

Proceedings of the 14th International Workshop on Semantic Evaluation, pages 1112–1119 Barcelona, Spain (Online), December 12, 2020. 1112 gundapusunil at SemEval-2020 Task 8: Multimodal Memotion Analysis Sunil Gundapu Language Technologies Research Centre KCIS, IIIT Hyderabad Telangana, India sunil.g@research.iiit.ac.in Radhika Mamidi Language Technologies Research Centre KCIS, IIIT Hyderabad Telangana, India radhika.mamidi@iiit.ac.in Abstract Recent technological advancements in the Internet and Social media usage have resulted in the evolution of faster and efficient platforms of communication. These platforms include visual, textual and speech mediums and have brought a unique social phenomenon called Internet memes. Internet memes are in the form of images with witty, catchy, or sarcastic text descriptions. In this paper, we present a multi-modal sentiment analysis system using deep neural networks combining Computer Vision and Natural Language Processing. Our aim is different than the normal sentiment analysis goal of predicting whether a text expresses positive or negative sentiment; instead, we aim to classify the Internet meme as a positive, negative, or neutral, identify the type of humor expressed and quantify the extent to which a particular effect is being expressed. Our system has been developed using CNN and LSTM and outperformed the baseline score. 1 Introduction According to Wikipedia article on Internet Memes, “A meme is an idea, behavior, or style that spreads from person to person within a culture often with the aim of conveying a particular phenomenon, theme, or meaning represented by the meme”. Meme is not only about the humorous picture, the Internet culture, or the sentiment that passes along, but also about the richness and distinctiveness of its language: it is often greatly structured with unusual written style. The Internet Memes often include superimposed text description with broken grammars and spellings variations. Nowadays, Internet memes come in almost every form of media, with modish formats continuously expanding. Initially, they work as a medium for humor to be shared, using cultural themes. However, they can also be manipulated to further political ideals, company promotions, and social media marketing. In this paper, we present a deep learning based multimodal approach for SemEval 2020: Task 8 on Memotion Analysis -The Visuo-Lingual Metaphor! (Sharma et al., 2020). This task has three Subtasks which describe as follows: (A) Sentiment Classification: From the given input Internet meme extract the sentiment and classify it as a positive, negative, and neutral meme. (B) Humor Classification: In this sub-task, the system has to recognize the type of humor expressed in a given Internet meme. The humor classes are sarcastic, humorous, offensive, and motivational. A meme can have more than one class. (C) Scales of Semantic Classes: In the third sub-task, quantify the extent to which a particular effect is being expressed in a given emotion, namely humour, offensive, sarcasm or motivational. Details of such quantification are reported in Table 1. This work is licensed under a Creative Commons Attribution 4.0 International Licence. Licence details: http:// creativecommons.org/licenses/by/4.0/. 1113 All our models are trained using only the corpus supplied by Memotion analysis organizers. The evaluation metric for subtask …

Content truncated to 500 words