基于神经网络词转向量的数字鸿沟关注热点和词嵌入比较研究

igital divide refers to the inequality of access to digital technology among different countries, regions, industries and populations in the process of global digitisation, which is not only a technical problem, but also a social equity problem. word2vec is a neural network word embedding algorithm that learns features from a large amount of text, represents each word as a vector, and captures the semantic and grammatical features of the words. The digital divide has received wide attention in both academic and practical circles. In this paper, we conduct neural network word2vec modelling on the digital divide-related content on China National Knowledge Infrastructure (CNKI), an academic platform in China, and Weibo, a social media platform in China, to compare and analyse the similar and related words of the "digital divide", and through the dimensionality reduction process and visualization method, we conduct multi-dimensional feature display on the word embedding features of all the words. The multidimensional features of all words are displayed and compared by dimensionality reduction and visualisation methods. This study aims to promote the development of digital divide research. The study found the difference characteristics between our academic platforms and social media platforms in terms of digital divide\'s focus of attention and word embedding. Following the data-driven research paradigm, this paper innovatively uses neural network word2vec to compare the focus of attention and word embeddings of digital divide between China\'s academic platforms and social media platforms, which provides a new perspective for innovative and practical research on digital divide.