Vorträge und Posterpräsentationen (ohne Tagungsband-Eintrag):
G. Li, S. Nickel, M. Nöllenburg, I. Viola, H. Wu:
"World map of recipes";
Poster: Graph Drawing Contest,
unknown;
19.10.2019.
Kurzfassung englisch:
This poster visualises the Meal Ingredients dataset with 151 international food recipes and their corresponding ingredients. The underlying graph layout in the image is automatically generated using a new multi-level force-based algorithm developed by the authors, but not yet published. The background flags were added manually to identify the countries from the data set. The algorithm aims to untangle mutually nested subgraphs by harmonizing the available space for the labels and improving edge visibility by duplicating high-frequency ingredient nodes. Ingredients occurring in multiple countries also receive at least one node per country. The idea is inspired by map diagrams, which often show the semantics enclosed by country boundaries. In our diagram, countries are represented by octolinear polygons, and are placed next to each other if they share many ingredients in their recipes. The actual placement of the countries by the algorithm is entirely data driven. As we can see, this design naturally gathers countries that are located on the same continent, due to the accessibility of the ingredients. The names of recipes are visualized using textual labels with sharp corners, and they are enclosed by the country polygon they belong to. Contrarily, ingredients are represented by textual labels with rounded corners. Moreover, ingredients are visually classified into common (pink) and special (blue) ingredients based on their frequency in the dataset. For visually analyzing the data set, we can generate smoothed spanning trees along the boundaries of an (invisible) Voronoi diagram of all textual labels to connect identical nodes to visually integrate all copies of one ingredient. For example, we highlighted the ingredient "soy sauce", one of the most commonly used ingredients in Asia, to discover that it has spread to the UK as well. We can also perform visual queries for related recipes based on sharing rare ingredients. For example, the British dish "steak and kidney pie" is highlighted in green together with three blue spanning trees connecting all recipes related to that dish via at least one of its special (blue) ingredients.
Erstellt aus der Publikationsdatenbank der Technischen Universität Wien.