Multimodal Representation Learning For Textual Reasoning Over Knowledge Graphs