Narrative Roads to Rebuild China's Global Image -Sentiment Analysis of Twitter Activities after China's Covid-19 Aid Activities in Italy

As the digital era unfolds, social media supported by mobile devices and AI technologies are rapidly changing the way people communicate. The impact of public opinion formed on the social network platforms often has an enormous and somehow long-lasting influence on a country's foreign policy and international relations, and international interactions in the form of Twiplomacy are gradually attracting the attention of researchers.

However, the attempts to analyze the impact that online public opinion has on actual policy or the inverse are still lacking or failed to explain the problem on the socioeconomic ground. There are very few theoretical explanations of how different pandemic narratives emerges and co-exists, and yet no article that analyzed how the positive and negative narratives about China spread and wrestled in their respective public opinion arenas during such a global event.

Under certain analogous sense, Virus like the rumor “China virus “and vaccines like the encouraging reports we get from the “China Aid Activities in Italy” are fighting each other in the social media battleground and thus in every individual's mind which finally aggregated as the ideology. Hence, to explain the underlying motives for the public to accept and spread different pandemic narratives, we combined the modified SIR model and Glaeser(2005) ' political economic dynamic to cast light on its deep route in the background of the U.S.-China trade conflict and the U.S. presidential election - all three Black swans landed in the same Lake.

Our research combined the sentiment semantic analysis methods from the AI-NLP field with the causal mechanism analysis to evaluate the differences in the attitudes of the Italian & international public towards China before and after the Chinese medical aid to Italy to fight the Covid-19 epidemic. Using the public data set from the GitHub project and other open-source scientific data, we applied the Bidirectional Encoder Representations from Transformers model (BERT) on individual tweets to include both emotional and non-emotional terms from a sentence sematic scale rather than word meaning scale which allows us to dig-in rich in topical information and estimate the international image trend of China during the 2020 February-March period. Here inspired by hybrid enrichment framework proposed by Gencoglu & Gruber (2020) but limited with the data, we choose the ARMA model instead to infer the sentimental level and the causality of medical aid event, empirical result shows that the positive change in Italy' public sentiment level is significantly affected by China's aid activities event here. The results of this article also provide insight into how China can improve its international image to promote the progress of many international regional economic and trade cooperation from the narrative perspective. Facts always speak louder than words, with which the adequate speed-up dissemination of fact is also important.