Event atlas-based journey prediction method, system, device and storage medium

Abstract

The invention provides an event atlas-based journey prediction method, a system, a device, and a storage medium. The method comprises the steps: generating event nodes and directed edges according to network travel text information, and obtaining a travel event atlas; inputting the node sequence with the weight into a network modeling tool, and constructing a weighted directed graph; inputting the weighted directed graph into a graph neural network model, and generating vector representation of each node in the tourism event graph; constructing a sub-graph related to each event chain according to the weighted and directed graph, and obtaining a corresponding sub-graph adjacency matrix A; inputting the samples into the graph neural network model in batches to obtain vector representation of nodes; inputting the sub-graph adjacency matrix A and the vector representation into a gated graph neural network, and outputting a sub-graph event representation; and determining a candidate event according to the candidate event node with the highest correlation score. According to the method, information extraction and fusion can be carried out on the tourism field data, the tourism event map is constructed, accurate journey prediction is carried out, and the labor cost is greatly saved.

Description

technical field [0001] The present invention relates to the field of deep learning reasoning, in particular to a method, system, device and storage medium for travel prediction based on event graphs. Background technique [0002] In recent years, with the continuous development of deep learning, reasoning and cognitive computing based on deep learning have attracted more and more attention from the industry and scholars. The evolution rules and patterns that occur successively between events in time are very valuable knowledge. Tourism event graphs can provide strong support for revealing and discovering event evolution rules. Many downstream tasks of text reasoning rely on A deep understanding of the logical knowledge of events. However, it is still challenging for machines to grasp a large amount of logical knowledge and perform cognitive reasoning, especially in the field of tourism where user experience is the goal, and the accuracy rate is still low. [0003] At prese...

AI Technical Summary

Problems solved by technology

Although the above methods have shown certain effects, they ignore the rich connection information between event nodes, and do not learn the representation of events from the structural level of the graph, so as to perform correlation analysis and prediction.

Images