Affiliations: School of Information and Control Engineering, Xi’an University of Architecture and Technology, Xi’an, Shaanxi, China
Correspondence:
[*]
Corresponding author: Lili Dong, School of Information and Control Engineering, Xi’an University of Architecture and Technology, Xi’an, Shaanxi, China. E-mail: [email protected].
Abstract: There is a lot of Chinese information on the Internet, including many architectural texts. In the field of building information, how to effectively find and organize these architectural texts has become a hot research topic. In this paper, the parallel optimizing of Deep Belief Network (DBN) implemented by Spark is proposed to classify the architectural texts. Firstly, the architectural texts preprocessing is automatically parallelized by spark technology. Secondly, the Word2Vec tool is used to represent the architectural texts and construct the feature dictionary. Thirdly the architectural text classification model is constructed by the Deep Belief Network. Finally, architectural text classification is parallelized by the Spark cluster. From experimental results we can see that the training time of parallel DBN is shortened to a quarter of the original, and gets less errors and better performance than that of the traditional shallow learning method.
Keywords: Spark parallelization, Deep Belief Network, architectural texts classification
