The Tradeoffs Between Open and Traditional Relation Extraction

By Michele Banko, Oren Etzioni. Annual Meeting of the Association for Computational Linguistics, 2008

417 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/6c8898cda9a1f13607e24306f6f64f20e0ff2ae7

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Entity Linking: Finding Extracted Entities in a Knowledge Base (2013, background): “For instance, given two person names in news documents about crime and violence, identify the victim and the perpetrator [4, 44].”
StatSnowball: a statistical approach to extracting entity relationships (2009, methodology): “Finally, by using the MLN model, StatSnowball can perform joint inference, while the O-CRFs [4] treat sentences independently.”
Entity Disambiguation for Knowledge Base Population (2010, methodology): “The information extraction oeuvre has a gamut of relation extraction methods for entities like persons, organizations, and locations, which can be classiﬁed as open-or closed-domain depending on the restrictions on extractable relations (Banko and Etzioni, 2008).”
Multi-Task Transfer Learning for Weakly-Supervised Relation Extraction (2009, background): “Banko and Etzioni (2008) studied open domain relation extraction, for which they manually identiﬁed several common relation patterns.”
Knowledge-Based Weak Supervision for Information Extraction of Overlapping Relations (2011, background): “…which perform self-supervised learning of relation-independent extractors ( e.g. , Preemptive IE (Shinyama and Sekine, 2006), T EXT R UNNER (Banko et al., 2007; Banko and Etzioni, 2008) and W OE (Wu and Weld, 2010)) can scale to millions of documents, but don’t output canonicalized relations.”
Relational World Knowledge Representation in Contextual Language Models: A Review (2021, background): “resentations is not new in natural language processing (Banko and Etzioni, 2008; Fader et al., 2011).”
Structured Relation Discovery using Generative Models (2011, background): “However, less explored are open-domain approaches where the set of possible relation types are not ﬁxed and little to no labeled is given for each relation type (Banko et al., 2007; Banko and Etzioni, 2008).”
Stacked Ensembles of Information Extractors for Knowledge-Base Population (2015, methodology): “In relation extraction, a stacked classifier effectively combines a supervised, closed-domain Conditional Random Field-based relation extractor with an opendomain CRF Open IE system, yielding a 10% increase in precision without harming recall (Banko et al., 2008).”
Open Information Extraction: The Second Generation (2011, methodology): “Pre-emptive Information Extraction and Open Information Extraction are the first paradigms that relax the restriction of a given vocabulary of relations and scale to all relation phrases expressed in text [Shinyama and Sekine, 2006; Banko et al., 2007; Banko and Etzioni, 2008; Wu and Weld, 2010].”
Identifying Relations for Open Information Extraction (2011, unknown): “We then passed these labeled examples to TEXTRUNNER’s training procedure, which learns a linear-chain CRF using closedclass features like POS tags, capitalization, punctuation, etc.TEXTRUNNER-R uses the argument-first extraction algorithm described in Section 2.”
Seed-Based Event Trigger Labeling: How far can event descriptions get us? (2015, unknown): “One such IE paradigm, including Pre-emptive IE (Shinyama and Sekine, 2006), On-demand IE (Sekine, 2006; Sekine and Oda, 2007) and Open IE (Etzioni et al., 2005; Banko et al., 2007; Banko et al., 2008), focuses on un-supervised relation and event discovery.”
Recovering Semantics of Tables on the Web (2011, methodology): “In our implementation, we use the TextRunner open extraction system [2].”
Probabilistic Frame Induction (2013, background): “Open IE (Banko and Etzioni, 2008) reduces the manual effort to designing a few domain-independent relation patterns, which can then be applied to extract relational triples from text.”
Discrete-State Variational Autoencoders for Joint Discovery and Factorization of Relations (2016, methodology): “These methods extract surface or syntactic patterns between two entities and either directly use these patterns as substitutes of semantic relations (Banko et al., 2007; Banko and Etzioni, 2008) or cluster the patterns (sometimes in context-sensitive way) to form relations (Lin and Pantel, 2001; Yao”
Reducing Wrong Labels in Distant Supervision for Relation Extraction (2012, background): “…select entity pairs from a corpus when: (i) the path of the dependency parse tree between the corresponding two named entities in the sentence is no longer than 4 and (ii) the path does not contain a sentence-like boundary, such as a relative clause 1 (Banko et al., 2007; Banko and Etzioni, 2008).”