Interactive and iterative learning-based place name labeling corpus intelligent construction method

Abstract

The invention discloses an interactive and iterative learning-based place name labeling corpus intelligent construction method based on, and the method comprises the steps: generating a word vector matrix of a character and a disambiguation matrix of the character in a sentence in an initial corpus, splicing the word vector matrix and the disambiguation matrix, inputting the spliced matrixes intoa Bi-LSTM and CRF integrated model for training, and generating a place name recognition model; embedding the place name recognition model into a human-computer interaction type place name labeling platform, and carrying out human-computer interaction type correction; fusing the initial training corpus with the marked place name corpus, optimizing the parameters of the place name recognition modeluntil the constructed corpus meets the requirements, ending iterative training learning, and realizing intelligent construction and optimization of the place name corpus. According to the method, theproblems of lack of geographical name corpus data, slow updating, time and labor waste in manual construction of geographical name corpus, low efficiency and the like at present can be effectively solved, and the problem of intelligent updating of the geographical name annotation corpus for multi-source, dynamic, heterogeneous and exponentially growing Internet texts can be effectively solved.

Description

technical field [0001] The invention belongs to the technical field of geographic information processing, and specifically relates to a method for intelligently constructing a place-name labeling corpus based on interactive and iterative learning, which realizes maximum optimization of deep learning model parameters, improves the effect of place-name recognition, and realizes the intelligent construction of a place-name labeling corpus and optimization. Background technique [0002] With the rapid development of the Internet and the advent of the era of big data and artificial intelligence, the world today is entering the era of ubiquitous information society and big data (Zhou Chenghu, 2011; Li Deren, 2012; Goodchild, 2017). Location big data is an important part of big data, and 80% of the information in the world is related to location (Williams, 1987; Liu Jingnan, 2014). Geographical names are the proper names given by people to specific geographical entities in the uni...

AI Technical Summary

Problems solved by technology

[0006] Purpose of the invention: In view of the fact that the current large-scale general corpus for place name recognition is relatively small, the update is slow, and the manual construction of the place name corpus is time-consuming, laborious, and inefficient, etc., and the place name entity recognition cannot meet the needs of ubiquitous location information services. The purpose of the present invention is to provide An intelligent construction method of place name annotation corpus based on interactive and iterative learning, which can maximize the optimization of deep learning model parameters, improve the effect of place name recognition, and realize the intelligent construction and optimization of place name annotation corpus

Images