We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computation and Language

Title: Multi-Attribute Relation Extraction (MARE) -- Simplifying the Application of Relation Extraction

Abstract: Natural language understanding's relation extraction makes innovative and encouraging novel business concepts possible and facilitates new digitilized decision-making processes. Current approaches allow the extraction of relations with a fixed number of entities as attributes. Extracting relations with an arbitrary amount of attributes requires complex systems and costly relation-trigger annotations to assist these systems. We introduce multi-attribute relation extraction (MARE) as an assumption-less problem formulation with two approaches, facilitating an explicit mapping from business use cases to the data annotations. Avoiding elaborated annotation constraints simplifies the application of relation extraction approaches. The evaluation compares our models to current state-of-the-art event extraction and binary relation extraction methods. Our approaches show improvement compared to these on the extraction of general multi-attribute relations.
Comments: Preprint of short paper for the 2nd International Conference on Deep Learning Theory and Applications (2021)
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Machine Learning (cs.LG)
Journal reference: Proceedings of the 2nd International Conference on Deep Learning Theory and Applications, Vol. 1, (2021), P. 148 - 156
DOI: 10.5220/0010559201480156
Cite as: arXiv:2111.09035 [cs.CL]
  (or arXiv:2111.09035v1 [cs.CL] for this version)

Submission history

From: Lars Klöser [view email]
[v1] Wed, 17 Nov 2021 11:06:39 GMT (1381kb,D)

Link back to: arXiv, form interface, contact.