Research On Extension Architectural Design Data Mining

Architectural cases include much of design knowledge that is significant for architects to study and refer.Especially when architects face design problems,learning related knowledge from architectural cases is a helpful approach for solving problems.However,more and more architectural cases have been accumulated in the internet with the advent of big data era.Resource surplus has already replaced resource shortage and become a new trouble for architects.While dealing with the design problems,architects have to spend much more time on searching helpful cases solving problems from a large amount of architectural cases in the internet.Since architects need to solve many problems during the design process,excessive time consumption has gradually become a burden,which takes up the time of design and causes negative impacts to the efficiency and quality of design.For this challenge,traditional retrieval method is unable to find the architectural cases and design knowledge needed by architectural within acceptable time.In order to help architects efficiently and accurately find the valuable cases and helpful design knowledge in the big data context,the Extension Architectural Design Data Mining(EADDM)was proposed as a brige between internet architectural cases and architects faceing with architectural design problems.Focusing on the dilemma of architectural design case searching in big data era,the idea of “Internet + Architecture” was introduced in this dissertation.Under the framework of Extension Architecture,by integrating the problems of Extension Architectural Design,the methods of Extension Data Mining,the architectural big data on the internet,and the design knowledge hidden in internet architectural design cases,the methodology of the EADDM was proposed.The main contents of this research include two parts,the theory and methods.In the theory part,the core concepts of the EADDM were defined first and its relationship with relevant fileds was sorted.Then,the core mechanism of the EADDM was introduced based on the principle of Extension Data Mining.According to the the adaptability of Extension Architectural Design theory and Extension Data Mining mthod to architectural big data and architects' needs,the advantage of the EADDM was demonstrated.After that,the eatures and essences of internet architectural design cases were summarized,including scattering,low-quality and variable.The diverse needs of architect facing with contradiction,quality and innovation problems were also analyzed.On the basis of these works,a five-stage process of EADDM was established,which includes data collection,data preparation,data storage,data analysis and image interpretation.Based on this process,the method of each stage was proposed in the method part as follow:In the data collection stage,items of important architectural design case data were selected as collection targets from the perspective of Extension Architectural Design,and their present reliable sources were also identified from the perspective of Extension Data Mining.According to a massive survey on internet architectural design cases,an architectural vocabulary was formed and optimized with extension transformations.The method of automatically collecting architectural design case data from the webpage source codes with web-spider was proposed based on this vocabulary.In the data preparation stage,the methods of form unifying,lack completing and mistake removing were proposed first to make up the defects of consistency,integrity and authenticty of regular data.Second,the methods of extracting valuable information,including keywords,features,abstract and recognition,from text data were proposed.Third,the matter-element from theory of Extension Architectural Design was introduced as a basic.By corresponding object,characteristic,and value of matter-element with name,parameter,and parameter's value of architectural design case respectively,case-element,the standard model of case data,was established.In the data storage stage,the basic framework of extension architectural design case database was established based on the needs of architects.In the data,program and interface levels,four fundamental modules of case storage and retrieval were designed for a large number of architectural cases input and output.The methods of how to apply the main functions of the database on architectural study,design and research were explained last.In the data analysis stage,the natures of three main kinds of Extension Architectural Design problem and the case and knowledge they need were analyzed first.A problem-leading comprehensive evaluation model of case value was proposed.By introducing the Extension Data Mining method,two superior modules about case analysis and recommendation were designed in this stage,and the architects can find the most valuable cases for actual problems efficiently and accurately.In the image interpretation stage,the design information and interpretation order of architectural case images were analyzed.Two technical routes of discovering knowledge solving design problems from the images were proposed to instruct architects to find design transformation solving design problems from case images.Then,according to the simulations of three typical problems of Extension Architectural Design,the application of image interpretation was illustrated.The practical effect of image interpretation and the EADDM were demonstrated by cross control test.With the help of EADDM proposed in this dissertation,architects facing with design problems can find valuable architectural design cases and helpful design knowledge in a short time according to the category and specific situation of problem.That is,architects would be able to get out of the heavy work of case searching.More time and energy can be devoted to architectural creation,which is the primary significance of this research.Furthermore,as a preliminary exploration of the new subject,“Internet + Architecture”,this research contributed to the scientific and intelligent development of architecture.Besides,it provided a reliable basic for future studies and applications which are more efficient,accurate and wide in related fields.Researchers interested in similar subjects can also take this dissertation as a reference.