Data Management and Data Visualisation: What if we fuse them into one step?

Data visualization in the form of scientific figures can be seen as a "window" to the data space behind a scientific article. Most of the time, considerable effort goes towards designing scien-tific figures in such a way that the information they convey can be effectively understood [1]. As a consequence, the concise, well-presented data depicted in scientific figures is often of interest to the readership [2]. This is in fact where a reader will first be confronted with the research data management practices of the authors: How do I go about obtaining the data from a specific figure? In our contribution, we reflect on this reality and suggest that in the scientific workflow, the data visualization step can be easily enriched with an RDM compo-nent to automatically save the data presented in the figures, ideally together with easy-to-understand metadata in a human- and machine-readable format. We present Plot Serializer, a tool that we have developed for this purpose. Plot Serializer has been developed to lower the threshold for creating FAIR [3], comprehensi-ble datasets corresponding to the figures in a scientific publication. It enables effortless export of data plotted in scientific figures into interoperable datasets with customizable metadata for improved reusability, facilitating RDM practices. Realizing that data depicted in figures is often secondary rather than primary data, Plot Serializer offers the option of adding custom metadata, potentially linking to other existing datasets or software. This also enables customi-zability to a broad range of use-cases across disciplines. [4] Plot Serializer uses its own metadata model that loosely follows the well-known conventions of matplotlib. It is possible to differentiate between two kinds of metadata in the context of figures: semantic metadata that carry information about the content and meaning of the data (for example axis labels or plot title) and formatting metadata that describe the plot style (for example axis scaling, line thickness or colors). Plot Serializes prioritizes semantic information to formatting information, as its focus lies on supporting research data management (RDM). [4] As it is becoming more common that journals require scientific articles to contain a data avail-ability statement with a reference to openly available datasets, we believe that including an RDM component in the data visualization workflow can be beneficial for both authors and readers. For authors, it is because it requires little additional effort. Moreover, even if the pri-mary data needs to remain closed access (e.g. proprietary data), the visualized data can usu-ally be shared – they are depicted in the figure, after all. The benefit for the readers is obvi-ous: being able to find the dataset corresponding to a certain figure and vice versa while guaranteeing to include essential semantic and formatting metadata.