Improved script learning method and device based on event evolution diagram

Abstract

The invention relates to an improved script learning method and device based on an event evolution diagram. The method comprises the following steps: constructing an abstract event evolution diagram by utilizing a specific event network formed by specific event chains; mining event evolution knowledge contained in the event evolution diagram to learn initial embedded representations of context events and candidate events; inputting the initial embedding representations of the context event and the candidate event into a bidirectional LSTM network to obtain updated embedding representations of the context event and the candidate event, including time sequence information of an event chain and event evolution knowledge contained in an event evolution diagram; and carrying out similarity calculation on the context events and the candidate events by utilizing the updated embedded representation to obtain a similarity score of each candidate event, probabilizing the similarity scores, and selecting the candidate event with the highest probability as a final predicted event. The method has high accuracy for script event prediction, and can be used in the fields of privacy stealing attack event prediction and the like.

Description

technical field [0001] The present invention relates to a script learning (Script Learning) method, in particular to a script learning method and device based on an event evolution graph, and the application of the method in the prediction of privacy theft attack events. The method belongs to the natural language in the field of natural language processing Understanding (Natural Language Understanding) sub-fields, while intersecting with technologies related to privacy protection. Background technique [0002] In today's era, natural language understanding is one of the most important tasks in the field of artificial intelligence, in which the understanding of unstructured text is getting more and more attention. Because the real world is full of unstructured text, the most typical of which is newswire text. Accurately understanding newswire texts is not only important for dialogue generation, intent recognition, and decision-making, but also has broader uses if the vast am...

AI Technical Summary

Problems solved by technology

Two subsequent works by Pichotta and Mooney (Pichotta, K., and Mooney, R.J. 2016. Learning statistical scripts with lstm recurrent neural networks. InAAAI, 2800–2806. and Pichotta, K., and Mooney, R.J. 2016. Using sentence- level LSTM language models for script inference. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, ACL 2016, August 7-12, 2016, Berlin, Germany, Volume 1: Long Papers.), the long short-term memory network LSTM ( LongShor-Term Memory) introduces the temporal relationship between modeling event sequences to better encode script information, but may be plagued by overfitting problems

Images