Research On Big Social Data Analysis Based On Deep Learning

The construction of a smart city is a process of refinement of the city and a manifestation of people's pursuit of a higher quality of life.With the development of science and technology and urbanization,the construction of smart cities is an inevitable trend.In the process of building smart cities,it is inseparable from the acquisition of various types of social data and how to effectively analyze data.This paper studies the related work of social data processing methods and finds that the problems existing in this field at this stage are:(1)The existing algorithms can not be very effective in mining data information.(2)The construction and interpretation of social data sets are not accurate enough.In order to solve the above problems,this paper conducts deep research on deep learning,clustering algorithms,data visualization and demography,and proposes a social data analysis method based on deep learning.The method has achieved good results in the field of specific social data processing.The main contributions of this dissertation are as follows:(1)An algorithm based on deep learning and clustering is proposed and further improvements are made to the algorithm.In view of the existing problems,this paper proposes a combination of multiple Restricted Boltzmann Machines to extract features from the data and then use k-means clustering and improved.This improved algorithm and algorithm can more effectively obtain the characteristics of the data.And flexible clustering for different issues(2)Propose a new data construction method and a method of visualization and result interpretation.The current collection of data on urban development does not fully portray all aspects of a city.Through the study of demographics and other related fields,this paper has found a data set that can present a fuller view of the city.The features learned are an abstract expression of the original data.The current interpretation of the processing results is not very clear.The method of data visualization in this paper has simple and clear results in the interpretation of the results of the public data field.In this paper,the deep learning and clustering algorithms are deeply studied.The problems existing in the field of social data processing are analyzed.The corresponding solutions are proposed for these problems and verified in the actual data.Experimental results show that the proposed algorithm can analyze public data under different scenarios,and this method has advantages over current mainstream methods.