PROBLEM: Although images play a vital role in the spread of information,but at same time they can also be used to disseminate misinformation on a wide scale. Viral social media posts and message forwards usually consist of an image and an associated title. Although the image might depict one a particular phenomenon or event, the text could purposefully be chosen to incite the viewer that could end up having fatal consequences. The danger of fake information propagated through images is very high in our country.
SOLUTION: It is a methodology and program to assist a person to identify if an image and associated title are trustworthy. The fundamentals of our approach are based on corroboration of news, i.e, the image must be present in credible news websites with associated title in news articles similar to description of image provided So It uses word mover distance to find the difference in the actual meaning of image caption and title of news articles. If not, we recommend that the user perform a human verification procedure.
TECH USED:
css3
google-cloud-vision-api
google-nlp-api
html
python
sass
word2vec
HOW TO RUN:
download GooogleImageVector Add it
download libraries
run my1.py