Sentiment Analysis of Fashion Related Posts in Social Media

Abstract

The role of social media in fashion industry has been blooming as the years\nhave continued on. In this work, we investigate sentiment analysis for fashion\nrelated posts in social media platforms. There are two main challenges of this\ntask. On the first place, information of different modalities must be jointly\nconsidered to make the final predictions. On the second place, some unique\nfashion related attributes should be taken into account. While most existing\nworks focus on traditional multimodal sentiment analysis, they always fail to\nexploit the fashion related attributes in this task. We propose a novel\nframework that jointly leverages the image vision, post text, as well as\nfashion attribute modality to determine the sentiment category. One\ncharacteristic of our model is that it extracts fashion attributes and\nintegrates them with the image vision information for effective representation.\nFurthermore, it exploits the mutual relationship between the fashion attributes\nand the post texts via a mutual attention mechanism. Since there is no existing\ndataset suitable for this task, we prepare a large-scale sentiment analysis\ndataset of over 12k fashion related social media posts. Extensive experiments\nare conducted to demonstrate the effectiveness of our model.\n