Open event extraction method and system based on multi-channel hierarchical graph attention network

Abstract

The invention provides an open event extraction method and system based on a multi-channel hierarchical graph attention network, and the method comprises the steps: obtaining a given document, and building a document syntax dependency tree according to the given document; the method comprises the following steps: based on a document syntax dependency tree, sequentially adjusting a dependency structure, associating default components, adding a sequence sequence and associating ancestor nodes and descendant nodes to construct a bidirectional dependency graph; based on the original graph attention network, node types are embedded and hierarchical adjustment is carried out so as to modify the feature convergence process of the original graph attention network, and then a hierarchical graph attention network is constructed; constructing to obtain a three-channel graph attention network; and inputting words in a given document into the three-channel graph attention network to update the embedded representation, and performing four-classification judgment through the full-connection network to realize extraction of the open event. The feature coding capability of the graph structure can be effectively enhanced.

Description

technical field [0001] The invention relates to the technical field of information extraction, in particular to an open event extraction method and system based on a multi-channel hierarchical graph attention network. Background technique [0002] As a key basic task of information extraction, event extraction aims to extract valuable event information from vast text data. There is no doubt that event extraction is of great significance and can help investment decision-making, risk analysis, understanding of people's conversation methods, case push, and auxiliary decision-making. [0003] At present, some of the existing research works on open event extraction use remote supervision and confrontation techniques to detect unrestricted types of events; some use pattern matching methods to extract triplet events. [0004] However, the above research has great limitations. Event detection only needs to identify the trigger word of the event, and does not discuss the extraction ...

AI Technical Summary

Problems solved by technology

[0004] However, the above research has great limitations. Event detection only needs to identify the trigger word of the event, and does not discuss the extraction of event arguments; the method of pattern matching usually has high accuracy, and its recall rate is generally low, that is, only applicable in some corpora and scenarios

Images