Research On Topic Network Model Of Building Energy Conservation Based On Txt Mining

Building energy conservation is one of the key areas to combat climate change and energy shortage.Correspondingly,the number of building energy conservation researches is increasing exponentially.Meanwhile,the rich literature data bring challenges for researchers to assimilate the study topics and trends in a short time.Therefore,the systematic analysis of building energy conservation researches through text mining technology has theoretical and practical significance to promote the development of this area.In this paper,5,712 related literatures in web of science database are used as the data objects.First,those data objects are divided into 5 stages according to the relationship between building energy consumption and climate change.And data cleaning and structured have been done in this stage to established the corpus of building energy conservation researches.Secondly,research topics of building energy conservation has been analyzed systematically through keywords analysis and LDA topic model.Then,based on the inadequacy of the LDA topic model,word2 vec model has been used to add semantic relationships of research topics to optimize LDA topic model.Also social network model has been applied to build inter-topic associations.Based on the visualization platform of Gephi,a new topic network model of building energy conservation literature data has been proposed.Finally,based on the results of different stages in topic network model,the research topics and relationship also evolutions in building energy conservation from 1973 to 2018 has been expounded.Conclusions in this paper could be draw as follows:(1)Main research topics in building energy conservationThe top three influencing factor in building energy conservation are indoor thermal environment,lighting and user behaviors.The top three energy systems are HVACs,lighting and power systems.The top two important energy-saving measures are building envelope design and solar energy utilization.The top three important reasons to promote the development of building energy conservation are economic factor,environmental impact and government policy.(2)Correlation between research topics in building energy conservationThe requirements of indoor thermal environment and indoor lighting leads to the energy consumption of in HVACs,lighting and power systems.Building envelope is the key mean to reduce the energy consumption of HVACs and lighting,which could improve solar energy usage and thermal insulation of buildings through building enclosure structure and materials.The cost of energy consumption and environmental impact are important factors affecting government policy.(3)Main theme evolution in building energy conservationThe research on the affecting factors of building energy consumption,especially the indoor thermal comfort and HVACs,have always been the focus of research in building energy conservation.The heat of researches on lighting and power system tends to be flattened.The early researches in building energy conservation were focused on the economic impact of energy shortage,while the recent researches concerned more about the environmental impact of energy consumption.The future development trend of building energy saving researches would be building energy conservation transformation,green building and intelligent building.Based on text mining technology,this paper has established a new theme network model to realize the theme mining,theme correlation analysis and theme evolution analysis in building energy conservation researches.It could be a objective technical method for researchers and practitioners in the field of building energy conservation to grasp the present situation and future trend of industry research and development scientifically.At the same time,it also broadens data analysis of the construction industry.In the future,the proposed topic network model can be applied to other digital text data in the field of architecture to realize the scientific management and analysis of building text information.