Large-scale entity alignment method based on reciprocal reasoning and progressive partitioning

Abstract

The invention discloses a large-scale entity alignment method based on reciprocal reasoning and progressive partitioning. The large-scale entity alignment method comprises the following steps: acquiring data of a source knowledge graph and data of a target knowledge graph; performing graph segmentation on the source knowledge graph and the target knowledge graph to obtain a group of sub-graph pairs of a source sub-graph and a target sub-graph; obtaining joint entity representations in all the sub-graph pairs by utilizing a representation learning model of the entity structure; performing reciprocal alignment reasoning on the sub-graph blocks in each sub-graph pair, and aggregating to obtain an alignment result of the sub-graph pair; and aggregating alignment results of all the sub-graph pairs to obtain an alignment result of the source knowledge graph and the target knowledge graph. According to the method, a more accurate entity alignment result can be generated through bidirectional segmentation and aggregation; compared with traditional direct alignment reasoning, two-stage reciprocal modeling of reciprocal reasoning can obtain a better alignment effect; by applying progressive partitioning, the memory and time cost of reciprocal alignment reasoning can be greatly reduced.

Description

technical field [0001] The present invention relates to the technical field of knowledge graphs, in particular to the technical field of large-scale knowledge graph fusion, and in particular to a large-scale entity alignment method based on reciprocal reasoning and progressive block. Background technique [0002] As an effective way to organize and store data, knowledge graph has received continuous attention in recent years. Knowledge graphs play an increasingly important role in organizing knowledge and downstream tasks (such as recommender systems, information retrieval, question answering systems, etc.). A knowledge graph is difficult to achieve completeness, because there are always new data and knowledge emerging. In order to increase the scale and coverage of knowledge graphs, a feasible way is to introduce data from other knowledge graphs. And entity alignment plays an important role in this process. Entity alignment aims to find the same real-world things in diff...

AI Technical Summary

Problems solved by technology

However, such methods (one-way segmentation methods) are often difficult to meet the first goal, because the segmentation of the second knowledge graph is limited by the constraint of retaining the seed entity pairs, which in turn destroys the knowledge graph to some extent. Structure

Images