Unstructured text from user reviews of products purchased from amazon is used in order to create a Knowledge Graph. The system then creates Knowledge Graph Embeddings which will be trained with shallow embedding so that the embeddings are in correspondance with the semantics of the graph. These Knowledge Graph Embeddings are useful for tasks such as knowledge graph completion done for a recommendation system.
Open Information Extraction by: ClausIE
Knowledge Graph Embeddings by: TransE
Dataset: https://www.kaggle.com/datasets/snap/amazon-fine-food-reviews