Visualizations of Natural Texts: A Survey on Text-to-Structured Data Transformations

Today’s world demands quicker access to data for timely decision-making. Automating the conversion of raw natural text into structured data is crucial for quicker decision-making and better accessibility. We have classified these conversion systems into two broad categories based on their representations which are visual systems and non-visual systems. Visual systems convert raw data into visual formats such as graphs, charts, diagrams, plots, or maps. In contrast, non-visual systems generate structured text output like SQL queries, regular expression (RegEx), first-order logic (FOL) and more. We have conducted a comprehensive survey of relevant related papers from various databases, including several state-of-the-art conversion systems, and analyzed their methods, architectures, output representations, datasets, and evaluation metrics. Finally, we have discussed our observations encouraging further research into these text-to-structured data conversion systems.