Knowledge graph-based entity co-reference detection data processing system

Abstract

The invention relates to a knowledge graph-based entity co-reference detection data processing system, which comprises a pre-constructed knowledge graph, a pre-trained encoder, a pre-trained spatial conversion matrix W, a processor and a memory in which a computer program is stored, and is characterized in that the encoder is used for converting a text entity into a Y-dimensional vector; the knowledge graph comprises M pieces of graph entity name coding information {R1, R2,... RM}, Rm is the mth graph entity name coding information, Rm is a Z-dimensional vector, and the value of m ranges from 1 to M; and the spatial conversion matrix W is a Y*Z-dimensional matrix and is used for converting the text entity code into a Z-dimensional vector. According to the invention, the accuracy and efficiency of entity co-reference detection are improved.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a data processing system for entity coreference detection based on a knowledge map. Background technique [0002] In order to avoid repetition, pronouns, titles and abbreviations are used in many texts to refer to the full names of the entities mentioned above. For example, "XX Industrial University", "XX Industrial University" and "工大" are used in the same text to refer to the same school name. In the same text, different words may also be used to express the same meaning, for example, "singer" and "singer". The above phenomenon is called the coreference phenomenon. Although humans can distinguish different titles of entities in text without difficulty, it is still a very difficult task for computers. In application scenarios such as using text information for natural language artificial intelligence model training, it is necessary to identify entity coreferences and r...

AI Technical Summary

Problems solved by technology

Although humans can distinguish different titles of entities in the text without difficulty, it is still a very difficult task for computers

In application scenarios such as artificial intelligence model training using text information for natural language, it is necessary to identify entity coreferences and resolve coreferences, otherwise, the accuracy of the model cannot be guaranteed.

In the existing technology, machine learning is used to detect entity coreference, but due to limited samples and text diversity, it will cause missed detection, the recognition result is comprehensive enough, and the accuracy of entity coreference detection is low

And in the face of different types of text or text information updates, different machine models need to be rebuilt, resulting in low efficiency of entity coreference detection

Images