Design of an GIS-based Investment Heatmap System using Topic Classification and NER

In recent years, Vietnam has received a significantly increasing Foreign Direct Investment (FDI) year on year. It has lead to the creation of a large number of social news that reflect to a certain extent the investment activities. Quantitatively extracting such information would be meaningful in analyzing market's direction. The objective of this study was to design a social listening system to identify key investment activities and trends over time using historical news data. First, we present the first-of-its-kind manually annotated investment domain-specific dataset for Vietnamese. Particularly, our dataset is annotated for join-tasks of 1) topic classification and 2) named entity recognition (NER) with newly-defined entity types. Second, empirical experiment was conducted using strong baselines on our dataset and show potential results with $\mathrm{F}1=82.43$ for topic classification task, and $\mathrm{F}1=92.15$ for NER task. Finally, we demonstrate the results on a Geographic Information System (GIS)-based heatmap system for the analysis of real-world social listening problem.