Design And Implementation Of Investment Information Software For Startups

Since the reform and opening up,the huge role played by SMEs in China's economy has increasingly been valued by people from all walks of life,and the biggest problem facing SMEs is the difficulty of operation,especially financing.In the entrepreneurial stage,SMEs often do not have professional data analysts,which leads to unclear trends in the capital market.There are often many detours in the entrepreneurial process.In the data analysis work of the entrepreneurial stage,the most important and most basic is the industry hotspot analysis and competitive product analysis.The basis of industry hotspot analysis is to first accurately classify each company into an industry that meets its actual situation,and then conduct hotspot mining on the business direction of each company.There are many domestic investment and financing information portal websites,but there are industries.The problem is that the division is too rigid and the business direction is too general;the basis of competing product analysis is to be able to accurately calculate the similarity between companies.The premise of calculating similarity is to effectively vectorize the company,and the correlation between vectorization at home and abroad.This topic collects relevant data on the Internet,and reclassifies the enterprise by multilabel classification technology.Each label corresponds to a business direction and solves the problem that the business direction is too general;through vectorization related technology,the company is Vectorization indicates that the inter-company similarity calculation performance is improved;the company is clustered based on the company vector,and each cluster is used as an industry to divide the industry,solving the problem that the industry division is too rigid.Combined with the above technologies,the industry hotspot analysis and competitive product analysis are solidified by software,which is convenient for SMEs to quickly complete data analysis in the startup stage.The main research and contributions of this thesis are as follows:(1)For the actual business scenario,a multi-label classification model based on Text-CNN and SGM model is designed to improve the classification performance by converting the multilabel classification problem into sequence generation problem,so as to better label the company.Classification,each label corresponds to a business direction,through a set of labels to describe the company's business direction in more detail,to solve the problem of the business direction is too general.(2)Designed a vector vectorization representation method based on multi-label classifier to obtain feature vectors more suitable for actual business scenarios to improve the performance of inter-company similarity calculation;and for company clustering,for each company Reasonable industry division,solving the problem of traditional industry division is too rigid.(3)With the designed multi-label classification model and company vectorization method as the core,a software module including data acquisition,model calculation and data analysis is designed and implemented,which can be used for industry hotspot analysis and competitive product analysis.