NUST Institutional Repository

Multimodal Islamophobic content detection on social media using Deep learning

Show simple item record

dc.contributor.author Nawaz, Mehram
dc.date.accessioned 2023-09-15T07:26:15Z
dc.date.available 2023-09-15T07:26:15Z
dc.date.issued 2023-09-15
dc.identifier.other 00000318212
dc.identifier.uri http://10.250.8.41:8080/xmlui/handle/123456789/38859
dc.description Supervised by Prof Dr. Hammad Afzal en_US
dc.description.abstract Islamophobia or anti-Muslim antagonism is one prepotent yet dilapidated form of racism in today’s world. The last couple of years has witnessed an immense surge in Islamophobic hate speech on social media nurturing and progressing violence and prejudice against Muslims and Islam. A growingly frequent expression of online hate speech is multimodal (text + image) in nature and known as a meme. Despite ample literature on hate speech detection on social media, there are only a few papers on Islamophobic hate speech detection. Our target is to automatically detect and classify the content of those memes that are hostile to Islam and transfer extremist thoughts against Muslims. As detecting memes is a multimodal (relying on both textual and visual cues) problem thus requiring a holistic understanding of photos, words in photos, and the context around the multimodal content conveys messages through a combination of images and text, demanding a need for multifaceted reasoning that encompasses both visual and linguistic comprehension. Identifying Islamophobic content that employs multiple modes of communication is inherently complex and remains an open challenge. When we encounter a meme, for instance, we naturally process the words and images in tandem, grasping their collective significance. For machines, this presents a formidable obstacle since they cannot simply analyze the text and images separately. They must instead synthesize these diverse modalities and discern how the meaning evolves when presented together. In this work, we seek to advance this line of research and develop a multimodal framework for the detection of Islamophobic memes. The Data shall be collected from Facebook and Instagram and shall be manually annotated to train the system to automatically classify the Islamophobic cyber hate instances into the given categories. Specific keywords that refer to Islamophobic content shall be considered as a search criterion, considering different manifestations of hatred against Muslims, such as terrorists, extremists, stereotyping, objectification, destruction, and violence en_US
dc.language.iso en en_US
dc.publisher MCS en_US
dc.title Multimodal Islamophobic content detection on social media using Deep learning en_US
dc.type Thesis en_US


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace


Advanced Search

Browse

My Account