{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/embedding-multimodal-relational-data-for","title":"Embedding Multimodal Relational Data for Knowledge Base Completion","arxiv_id":"1809.01341","date":"2018-09-05","proceeding":"EMNLP 2018 10","authors":["Pouya Pezeshkpour","Liyan Chen","Sameer Singh"],"abstract":"Representing entities and relations in an embedding space is a well-studied\napproach for machine learning on relational data. Existing approaches, however,\nprimarily focus on simple link structure between a finite set of entities,\nignoring the variety of data types that are often used in knowledge bases, such\nas text, images, and numerical values. In this paper, we propose multimodal\nknowledge base embeddings (MKBE) that use different neural encoders for this\nvariety of observed data, and combine them with existing relational models to\nlearn embeddings of the entities and multimodal data. Further, using these\nlearned embedings and different neural decoders, we introduce a novel\nmultimodal imputation model to generate missing multimodal values, like text\nand images, from information in the knowledge base. We enrich existing\nrelational datasets to create two novel benchmarks that contain additional\ninformation such as textual descriptions and images of the original entities.\nWe demonstrate that our models utilize this additional information effectively\nto provide more accurate link prediction, achieving state-of-the-art results\nwith a considerable gap of 5-7% over existing methods. Further, we evaluate the\nquality of our generated multimodal values via a user study. We have release\nthe datasets and the open-source implementation of our models at\nhttps://github.com/pouyapez/mkbe","url_abs":"http://arxiv.org/abs/1809.01341v2","url_pdf":"http://arxiv.org/pdf/1809.01341v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"embedding-multimodal-relational-data-for","repo_url":"https://github.com/pouyapez/mkbe","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"embedding-multimodal-relational-data-for","repo_url":"https://github.com/pouyapez/multim-kb-embeddings","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"imputation","task_name":"Imputation"},{"task_slug":"knowledge-base-completion","task_name":"Knowledge Base Completion"},{"task_slug":"link-prediction","task_name":"Link Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1809.01341","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}