Web-Scraped Data for MSMEs Economic Analysis: Insights on Innovation and Bankruptcy

The use of non conventional data for economic analysis has proven to be relevant from several point of views. Not only unconventional data can fill informational gaps for entities not covered by conventional data, but they also provide additional information, different and complementary with respect to traditional sources. Among the unconventional sources, web-based indicators have been increasingly in use, with particular attention to the exploitation of text content. In this work we discuss instead the use of HTML code and illustrate two promising fields of application for the analysis of micro, small and medium enterprises: bankruptcy and innovation. We explore these possibilities through supervised and unsupervised learning, leveraging traditional data as a baseline for comparison and validation.