Text this: MDCKE: Multimodal deep-context knowledge extractor that integrates contextual information