Event extraction method based on encoder-decoder

Abstract

The invention discloses an event extraction method based on Encoder-Decoder. S1, text preprocessing, text preprocessing is performed on open domain data, required text content is extracted, and word vector training is performed on the preprocessed data; S2, Mark the text, mark the preprocessed text, and identify the event type corresponding to the sentence; S3, combine the GRU+Attention mechanism to build and train the Encoder-Decoder model for determining the event type, and obtain a lightweight deep learning Encoder-Decoder The Decoder network model is continuously trained, and the event extraction task is abstracted and represented to obtain the prediction result; S4, based on the trained lightweight deep learning Encoder-Decoder network in step S3, the event extraction task is completed, and the corresponding event type is obtained, etc. The present invention has the advantages of strong domain adaptability, high abstract learning ability, simple model, and the calculation process needs to consume a small amount of resources and time.

Description

technical field [0001] The present invention relates to the field of natural language processing, and more specifically, to an Encoder-Decoder-based event extraction method. Background technique [0002] With the rapid development of the Internet, especially the mobile Internet, resources on the Internet have exploded. How to quickly and accurately extract valuable information from such a large amount of Internet resource information has become a major problem faced by people. In this context, people put a lot of energy into information extraction. Event extraction is a complex form of information extraction, which can provide a higher level of content processing abstraction capabilities, mainly refers to extracting events of interest to users from unstructured data, and storing them in the database in a structured way for users use. [0003] With the vigorous development of event extraction technology, early research methods are mainly carried out through pattern matchin...

AI Technical Summary

Problems solved by technology

[0004] Existing technologies are often aimed at a specific field, and there are problems such as domain adaptability, poor abstract learning ability, complicated models, and the calculation process consumes a lot of resources and time.

Images