Anzeige
Erschienen in:
2019 | OriginalPaper | Buchkapitel
A Relation Proposal Network for End-to-End Information Extraction
verfasst von : Zhenhua Liu, Tianyi Wang, Wei Dai, Zehui Dai, Guangpeng Zhang
Erschienen in: Natural Language Processing and Chinese Computing
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
Information extraction is an important task in natural language processing. In this paper, we introduce our solution on NLPCC 2019 shared task 3 Information Extraction which has provided with the largest industry Schema based Knowledge Extraction (SKE) data-set. Our proposed method is an end-to-end framework which first catches the relation hints in raw text with a relation proposal layer, then follows by an entity tagging design which is targeted to decode the corresponding triplet entities with the given relation proposal. Compared with previous works, our method is efficient and can well handle overlapping and multiple triplets in one sentence. With a simple model ensemble, our solution achieves 0.8903 F1-Score on final leaderboard which ranks forth among all participants.