{"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/traversing-knowledge-graphs-in-vector-space","title":"Traversing Knowledge Graphs in Vector Space","arxiv_id":"1506.01094","date":"2015-06-03","proceeding":"EMNLP 2015 9","authors":["Kelvin Guu","John Miller","Percy Liang"],"abstract":"Path queries on a knowledge graph can be used to answer compositional\nquestions such as \"What languages are spoken by people living in Lisbon?\".\nHowever, knowledge graphs often have missing facts (edges) which disrupts path\nqueries. Recent models for knowledge base completion impute missing facts by\nembedding knowledge graphs in vector spaces. We show that these models can be\nrecursively applied to answer path queries, but that they suffer from cascading\nerrors. This motivates a new \"compositional\" training objective, which\ndramatically improves all models' ability to answer path queries, in some cases\nmore than doubling accuracy. On a standard knowledge base completion task, we\nalso demonstrate that compositional training acts as a novel form of structural\nregularization, reliably improving performance across all base models (reducing\nerrors by up to 43%) and achieving new state-of-the-art results.","url_abs":"http://arxiv.org/abs/1506.01094v2","url_pdf":"http://arxiv.org/pdf/1506.01094v2.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":"traversing-knowledge-graphs-in-vector-space","repo_url":"https://worksheets.codalab.org/worksheets/0xfcace41fdeec45f3bc6ddf31107b829f","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":null},{"paper_slug":"traversing-knowledge-graphs-in-vector-space","repo_url":"https://github.com/nec-research/dccg","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"knowledge-base-completion","task_name":"Knowledge Base Completion"},{"task_slug":"knowledge-graphs","task_name":"Knowledge Graphs"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1506.01094","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}