Domain concept extraction method based on Deep Learning

Abstract

The invention discloses a domain concept extraction method based on Deep Learning. The method includes extracting samples in a training corpus, adopting word frequency, document frequency, inverse document frequency, word length, word frequency variance and domain consensus as feature vectors, training and acquiring a deep network model, which is capable of representing the complex mapping correspondence between the word-type filed concept multi-dimensional feature vectors and class labels, on the basis of the Deep Learning technology, and finally comparing the deep network model established on the basis of the Deep Learning technology, an optimized BP neural network model and mainstream KNN and SVM models in the testing step. According to the tests, the optimal test effect is acquired through the deep network model established on the basis of the Deep Learning technology.

Description

technical field [0001] The present invention relates to domain concepts, automatic extraction of domain concepts, artificial neural network, Deep Learning and deep belief network technical fields, specifically a feature extraction method based on Deep Learning that is suitable for the characteristics of word-type domain concepts. Background technique [0002] Domain concept is a manifestation of domain knowledge. People use domain concept to describe certain objects in the domain and to disseminate domain information. For example, "SMS" and "CRBT" are concepts in the field of mobile communications, while "data structure" and "computer network" are concepts in the field of computers. In a sense, the domain concept is the abstraction of things in the process of human cognition, a form of expression of domain knowledge in texts, and reflects the development and changes of the domain to a certain extent. Domain concepts are usually used more frequently in specific domains and l...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a deep learning-based domain concept extraction method for the traditional unsupervised method with weak learning ability and poor domain concept extraction effect, and transform the domain concept extraction problem into a binary classification problem. Richer statistical features, using Deep Learning's domain concept extraction algorithm, combining Deep Learning and domain concept extraction tasks, conducting unsupervised pre-training by building a deep belief network, and then performing supervised adjustments with traditional neural network models , compared with the KNN and SVM models, the finally trained deep network model achieved the highest F value on the test data set

Images