Research And Implementation Of Situation Awareness System For Intelligent Home Industry Based On Text Data

At present,society has made great progress and people's living standard is improving day by day,and smart home is gradually integrated into daily life.With the rapid increase in the value of the industry,the development status and prospects are more worthy of study.At present,smart home enterprises do not have a mature research system,which hinders practitioners and investors from understanding the smart home industry to a certain extent.The key information of enterprise development is often interspersed among various data,and the types and quantities of data are large.Integrating data into text is conducive to retaining value to the greatest extent and reducing the degree of distortion of processing results.Therefore,based on the existing enterprise text data,this paper uses text mining algorithms to analyze the development level and potential risks of enterprises from the perspective of situational awareness,and build a complete smart home industry research system.This paper is mainly studied from the following stages,namely the acquisition stage of situational elements,the stage of situational understanding and the stage of situational prediction.In the element acquisition stage,crawler technology is used to obtain industry information,linear regression is used to fill in the missing information,the kettle integration tool is used to store multiple data sources uniformly,and discrete data and continuous data are transformed using One-Hot and Box-Cox respectively.,using principal component analysis to reduce the dimension of text data.The preprocessed data is more suitable for training the model under the condition of ensuring the original characteristics of the data.In the situation understanding stage,this paper builds a complete set of enterprise index scoring system according to different types of text data.This paper uses jieba word segmentation,word2 vec and LSTM models to analyze sentiment data;uses word segmentation,TF-IDF and BERT models to perform similarity analysis on document material data,and puts the sentiment analysis and similarity analysis results into the enterprise index system.In order to make the score more credible,this paper mainly uses entropy method to calculate the weight,and the comprehensive score of the enterprise is finally obtained by combining the scores of the indicators.In the situation forecasting stage,this paper uses the ARIMA forecasting model to analyze the future development trend according to the data changes of the enterprise over the years,and compares the forecasting accuracy of the two models.At the same time,situation forecasting makes it possible for enterprises to prevent risks in advance.According to the above content,in order to express the research process and results more intuitively,the paper visualizes the text data and analysis results to realize a complete situational awareness system.