Interest point duplicate removal method and device based on graph neural network, equipment and storage medium

Abstract

The invention relates to an interest point duplicate removal method and device based on a graph neural network, computer equipment and a storage medium. The method comprises the following steps: acquiring all interest points in a target territorial range to be subjected to duplicate removal, and constructing an interest point diagram based on a geographic position according to all the interest points; screening out multiple pairs of interest point repeated pairs from all interest points, and labeling the multiple pairs of interest point repeated pairs to obtain multiple pairs of seed interest point repeated pairs; iteratively training the graph neural network model according to the interest point graph and the multiple pairs of seed interest point repetitive pairs to obtain a trained graph neural network model; processing the interest point diagram through a trained graph neural network model, and determining all interest point repeated pairs in all interest points according to a processing result; and deleting any one interest point in each interest point repeated pair. According to the embodiment of the invention, the method can improve the deduplication effect of the points of interest, is efficient and reasonable, and is wide in application range.

Description

technical field [0001] The present application relates to the field of electronic maps, in particular to a method, device, computer equipment and storage medium for deduplication of points of interest based on a graph neural network. Background technique [0002] Point of Interest (POI for short), which generally includes information such as name, address, latitude and longitude, category, etc., is the most important content of network electronic maps and the foundation of Internet location services. Because there is a lot of redundant data in the points of interest on the Internet, that is, two points of interest may have different text descriptions, but they may correspond to the same point of interest in the real world. When users use map services, their experience will be seriously affected. Therefore, on the basis of ensuring data richness, how to remove duplicate data and present more concise and pure map POI data to users has become a current research hotspot. At pre...

AI Technical Summary

Problems solved by technology

[0012] (1) Based on the unsupervised similarity score method, the matching effect is poor for the scene where the two interest points are actually repeated data, but the text is quite different.

[0013] (2) Although the two interest points are very close in text, it is not actually repeated data that will cause a mismatch

[0014] (3) The threshold of the similarity score is not easy to set

[0016] (1) A lot of feature engineering work is required to construct features, and the process is cumbersome

[0017] (2) The model is shallow, the expression ability is limited, and the judgment effect is average

[0018] (3) This method assumes that the interest points are independent of each other. However, there is a certain spatial relationship between the actual interest points, so the relationship information between the interest points is not used to judge the weight, and the use of less information is not effective. it is good

[0020] (1) Pre-trained deep models generally input plain text information, which is poorly compatible with non-text features

[0021] (2) This method assumes that the interest points are independent of each other. However, there is a certain spatial relationship between the actual interest points, so the relationship information between the interest points is not used to judge the weight, and the use of less information is not effective. it is good

Images