Knowledge vault: a web-scale approach to probabilistic knowledge fusion
By Xin Luna Dong, Evgeniy Gabrilovich, Geremy Heitz, Wilko Horn, Ni Lao, Kevin P. Murphy, Thomas Strohmann, Shaohua Sun. 2014
1,828 citations recorded by the source corpus.
Recent years have witnessed a proliferation of large-scale knowledge bases, including Wikipedia, Freebase, YAGO, Microsoft's Satori, and Google's Knowledge Graph. To increase the scale even further, we need to explore automatic methods for constructing knowledge bases. Previous approaches have primarily focused on text-based extraction, which can be very noisy. Here we introduce Knowledge Vault, a Web-scale probabilistic knowledge base that combines extractions from Web content (obtained via analysis of text, tabular data, page structure, and human annotations) with prior knowledge derived from existing knowledge repositories. We employ supervised machine learning methods for fusing these distinct information sources. The Knowledge Vault is substantially bigger than any previously published structured knowledge repository, and features a probabilistic inference system that computes calibrated probabilities of fact correctness. We report the results of multiple studies that explore the relative utility of the different information sources and extraction methods.
Subfields: Artificial Intelligence, Information Systems, Knowledge Graphs & Representation
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Mohit Bansal
- Jianfeng Gao
- Percy Liang
- Luke S. Zettlemoyer
- Yejin Choi
- Yoshua Bengio
- Noah A. Smith
- Quoc V. Le
In Information Systems
In Knowledge Graphs & Representation
- Bilal Abu-Salih
- Sebastian Riedel
- Tim Rocktaschel
- Xinyi Zhou
- Seyed Mehran Kazemi
- Partha P. Talukdar
- Yejin Choi
- Zhiyuan Liu
Citations in the corpus
- Mass-Editing Memory in a Transformer (2022, background): “Pioneering hand-curated efforts (Lenat, 1995; Miller, 1995) have been followed by web-powered knowledge graphs (Auer et al., 2007; Bollacker et al., 2007; Suchanek et al., 2007; Havasi et al., 2007; Carlson et al., 2010; Dong et al., 2014; Vrandečić & Krötzsch, 2014; Bosselut et al., 2019) that extr”
- COMET: Commonsense Transformers for Automatic Knowledge Graph Construction (2019, background): “, 2008) and unstructured text extraction (Dong et al., 2014; Carlson et al., 2010; Nakashole et al., 2011, 2012; Niu, 2012).”
- A Survey on Data Collection for Machine Learning: A Big Data - AI Integration Perspective (2018, background): “Knowledge Vault [137] also extracts from Web content, but combines facts from text, tabular data,”
- Complex Embeddings for Simple Link Prediction (2016, background): “Web-scale knowledge bases (KBs) provide a structured representation of world knowledge, with projects such as DBPedia (Auer et al., 2007), Freebase (Bollacker et al., 2008) or the Google Knowledge Vault (Dong et al., 2014).”
- Embedding Entities and Relations for Learning and Inference in Knowledge Bases (2014, background): “; Taskar, 2007), such as Markov-logic networks (Richardson & Domingos, 2006), which directly encode multi-relational graphs using probabilistic models; (2) path ranking methods (Lao et al., 2011; Dong et al., 2014), which explicitly explore the large feature space of relations with random walk; ”
- A Review of Relational Machine Learning for Knowledge Graphs (2015, methodology): “In [24], they show a similar result for the semantic embedding of relations using ER-MLP.”
- Representing Text for Joint Embedding of Text and Knowledge Bases (2015, methodology): “This task setting has been used in models for KB completion previously, e.g. (Dong et al., 2014; Gardner et al., 2014), even though it has not been standard in evaluations of distant supervision for relation extraction (Mintz et al., 2009; Riedel et al., 2013).”
- Traversing Knowledge Graphs in Vector Space (2015, background): “Many models have been proposed for knowledge base completion (Bordes et al., 2013; Yang et al., 2015; Socher et al., 2013), relation extraction (Riedel et al., 2013), and knowledge fusion (Dong et al., 2014).”
- Holographic Embeddings of Knowledge Graphs (2015, methodology): “In this case, negative examples can be generated by heuristics such as the local closed world assumption (Dong et al., 2014).”
- Convolutional 2D Knowledge Graph Embeddings (2017, background): “For example, in Freebase and DBpedia more than 66% of the person entries are missing a birthplace (Dong et al. 2014; Krompaß, Baier, and Tresp 2015).”
- Building a Large-scale Multimodal Knowledge Base System for Answering Visual Queries (2015, background): “In vision, there is now a small but growing amount of attention in building visual KBs.”
- Byte-Level Machine Reading Across Morphologically Varied Languages (2018, background): “It encompasses tasks like textual entailment (Rocktäschel et al. 2015), information extraction (Dong et al. 2014), semantic textual similarity (Kenter, Borisov, and de Rijke 2016), question answering (Fader, Zettlemoyer, and Etzioni 2013) and machine reading (Hermann et al. 2015).”
- Linked data quality of DBpedia, Freebase, OpenCyc, Wikidata, and YAGO (2017, background): “Thus, out of scope are KGs which are not openly available such as the Google Knowledge Graph4 and the Google Knowledge Vault [14].”
- Question Answering on Knowledge Bases and Text using Universal Schema and Memory Networks (2017, background): “However, a major drawback of this paradigm is that KBs are highly incomplete (Dong et al., 2014).”
- Representation Learning for Dynamic Graphs: A Survey (2019, background): “In this work, we mainly adopt the term KG.”
- A Survey of Fake News (2020, methodology): “• Local closed-world assumption [Dong et al. 2014] : The authenticity of non-existing triples can be determined by the following rule: let T ( s , p ) denote the set of existing triples in the KB (KG) for subject s and predicate p .”
- Structured Embedding via Pairwise Relations and Long-Range Interactions in Knowledge Base (2015, background): “…which is very useful for the incremental knowledge learning because we do not need to re-train all of the entities/relations when new (unobserved) entities or relations are added into a growing knowledge graph like NELL (Betteridge et al. 2009; Carlson et al. 2010) or KV (Dong et al. 2014).”
- Knowledge Graph Embedding for Link Prediction (2020, background): “For instance, it has been observed that over 70% of person entities have no known place of birth, and over 99% have no known ethnicity [12, 69] in FreeBase, one of the largest and most widely used KGs for research purposes.”