The Center for Strategic Analytics and Big Data held an autumn series of webinars on the topic of human potential research
At the first webinar in the series held on October 7, 2022, “How Can Big Data Help to Identify Decision-Making Centres and Key Development Trends in Different Countries (China as an Example)?”, the speakers – HSE ISSEK Centre for Strategic Analysis and Big Data fellows Mariia Antasheva, Julia Isaeva, and Polina Lobanova – commented on the peculiarities of working with data sources in Chinese, outlined the main trends in the human capital development that can be identified based on big data analysis, and also highlighted the main measures to support the human capital implementation in China. The speakers noted that human capital development and various talent support measures have become extremely relevant to Chinese domestic agenda, along with the intensification of government activity in this sphere, including the widest possible usage of digital technologies. With the help of the iFORA big data mining system developed at the HSE ISSEK, the speakers identified a number of competent authorities and organizations related to the most influential decision-making centers in the PRC in the field of human capital, which include, for example, the National Development and Reform Commission of the PRC (国家发展改革委).
At the second webinar in the series held on November 2, 2022, “How Can Big Data Help in Determining Public Sentiment? On the example of the Chinese mass media”, speakers – HSE ISSEK Centre for Strategic Analysis and Big Data fellows Julia Isaeva and Mariia Antasheva, revealed the importance of sentiment analysis as an effective tool for identifying trends in public sentiment. The Chinese social network Zhihu (知乎), whose audience currently compounds more than 150 million people, was chosen as the source base for the current investigation, the results of which were presented at the webinar. The choice of this type of source – that is a social network – was determined by its specificity and a high degree of emotional coloring of the content published. As a major part of the study, the regression model of sentiment analysis for texts in Chinese had been trained, which allows "predicting" the semantic assessment of texts in the range (-1; 1), and the result of which is interpreted through the proximity of the semantic assessment received from it to one of values: "-1" = "negative", "0" = "neutral", "+1" = "positive". The results of sentiment analysis of a social network for key queries within the topic "Urban infrastructure" showed that the public assessment of the level of infrastructure development in China varies from the type of a source analyzed – media or social networks.
At the third webinar in the series held on November 18, 2022, “A ‘Machine’ View on the Human Capital Development”, the speaker – HSE ISSEK Centre for Strategic Analysis and Big Data leading expert Mikhail Zakharov, outlined colleagues’ experience in analyzing large amounts of textual data on inflation, unemployment, investment in education, and other aspects that are significant for studying the factors affecting the human capital development. To identify this kind of text data, special models are trained; the models’ training algorithm includes a number of processes, for example, data markup, hyperparameters selection, etc. The major part of the webinar was devoted to describing how machine learning is being conducted based on the Transformer model, and how the problem of the lack of labeled data is being solved. During the webinar, the speaker also commented on such issues as how machine learning is being actively applied on the example of real projects of the HSE ISSEK Centre for Strategic Analysis and Big Data that are performed using the iFORA system, how can the problem that needs to be solved be formulated as a machine learning problem, what are the methodological approaches used to train models.
In the final part of each of the three webinars, the audience asked the speakers some questions related to the topics of the webinars. Moreover, following the results of the webinar “How Can Big Data Help in Determining Public Sentiment? On the example of the Chinese mass media” an article summarizing the results of a study conducted by HSE ISSEK Centre for Strategic Analysis and Big Data fellows on the sentiment analysis of Chinese-language sources has been published on the HSE Daily university brand media website by Pavel Aptekar, columnist Media.