Image and associated text clustering based recommender system based on a novel similarity measure

We need to work in python notebooks (google colab). We take three public benchmark datasets which include images and the associated text from social media (like MVSA). The image and textual data are preprocessed separately. Then the relevant features are combined and a model needs to be built that considers the similarity between images (along with the respective text data) using a novel similarity measure. Post the clustering a recommendation system is built upon it depending on the value of similarity measure.