Building consumer confidence index based on social media big data

This study takes microblog texts as the original data, builds Consumer Confidence Index (CCI) user dictionary, and calculates the CCI through different experimental conditions of the target words' frequency analysis. We explored the better way between "exact matching" and "fuzzy matching," and tested the necessity of degree words and negative words. The main findings are as follows: (a) The numbers of principal components extracted in the six different matching methods and weight settings were similar, which were nine or 10 principal components, and the explanatory variance of composition was above 96%; (b) Comprehensively considered, the principle of "fuzzy matching target word + degree word" was the best calculation method; (c) When the validation set data to predict the CCI in next month which based on the optimal method was used, the difference between the index estimate and the official value was within 10%. To a certain degree, CCI based on microblog can be on behalf of consumers' subjective feelings as well as attitudes towards China's current and future economic situations.