Loading...

Online Detection Of Multi-Modal Fake News

Ghorbanpour, Faeze | 2021

919 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 54387 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Rabiee, Hamid Reza; Fazli, Mohammad Amin
  7. Abstract:
  8. Today, social networks have provided an environment for disseminating all kinds of information and news among the people. One of the challenges of ex-panding social networks is reading and trusting fake news propagating against accurate information. Fake news is false information that the author intention-ally produces and publishes. Due to the destructive effects of spreading fake news, determining the trustworthiness of news is one of the critical issues in the so-cial, political, and economic fields. In this research, we worked on detecting fake news on social media using multimodal data. To solve the problem of fake news detection, we used the text and images of the information. Transfer learn-ing and pre-trained models have been used to extract the text and image fea-tures. A Recurrent Neural Network based on the attention mechanism has been used to extract the rela-tion between text and image features. We used the data collected from Twitter and Weibo to evaluate our proposed model. We compared it with previous re-searches, and the proposed model performed better than these models in terms of accuracy and F-measure. Due to the low number of labeled news, the lack of labeled data was another challenge we solved. We used the possible infor-mation of this unlabeled news to improve the performance of the proposed model by adding a clustering and label assignment phase. We evaluated the method by removing the labels of some news items and comparing them with related work. The results showed that adding information of this unlabeled news was efficient and increased the model’s accuracy
  9. Keywords:
  10. Deep Learning ; Social Networks ; Fake News Detection ; Multi-Modal Data ; Feature Extraction ; Text and Image ; Small Labeled Data

 Digital Object List

 Bookmark

...see more