The geographic map is an intuitive visualization of spatial datasets, for example the distribution of news articles or social media postings. Triggered by a search query, the system analyzes the set of matching documents to determine all referenced locations.
While the notion of a Geospatial Web  has inspired our work for the better part of the last decade, the March 2017 release Mangrove Monitor takes webLyzard’s geospatial analytics capabilities to a new level. We rebuilt the geographic map from scratch for optimized performance and a more effective information design. Custom base layers and adaptive tooltips adapt the map to specific tasks. Flexible export options enable its reuse in external applications.
Try out the dashboard at asap.weblyzard.com
Layout and Data Representation
Dynamic updates triggered by a user interaction help understand the geographic context of a query without interrupting the user’s workflow. Layout and color scheme of the geographic map depend on the chosen base layer. Circles of different size and color represent single documents or document clusters:
- The position of circles mark the coordinates – i.e., longitude and latitude – of locations (cities, countries, landmarks, etc.) that were identified in the documents.
- The size of the circles is proportional to the number of documents referring to a specific position.
- The color indicates either one of the selected topics, or metadata attributes such as sentiment. In the case of sentiment, for example, the color ranges from red (negative) to grey (neutral) and green (positive). Colors vary in saturation, depending on the degree of polarity. Vivid colors hint at emotionally charged issues, less saturated ones a more neutral coverage.
- Optional Arcs connect referenced locations with the origin of the documents (i.e., the locations of authors or publishers). Accessible via the menu in the lower left corner, this feature allows analysts to explore frequently mentioned locations in a particular country’s media channels.
The full potential of the geographic map unfolds in conjunction with the drill-down capabilities of adaptive tooltips. Based on the user’s current context (country shape, point of interest, etc.), the tooltip displays the most relevant information in a local context. It also includes an option to restrict or extend the search.
- Hovering over a circle activates a document preview and a tooltip with a line chart and top associations with this particular location. All arcs pointing to this location are displayed with higher opacity. This highlights in which countries this location is being discussed.
- Clicking on a circle or country displays an extended tooltip with context-specific actions: (i) Focus on this Point to show results within a 100 km radius, automatically extended to 1000 km in case of sparse coverage, (ii) Replace the current search with a search for mentions of this location, and (iii) Restrict or Extend the search via Boolean operators (AND, OR).
- Zooming is available via the mouse wheel or double click, while clicking and dragging enables a seamless panning of the entire display.
Details on Demand
Advancing the State of the Art
Developed as part of the ASAP and PHEME research projects, the geographic map supports very large datasets and has been tested with search queries returning more than 100 million documents. The screenshot above shows a street-level display with an adaptive tooltip for on-the-fly query refinements. The blue markers presents anonymized cell tower activity data for the City of Rome. The system then overlays this information with geotagged Twitter postings (green = positive; red = negative). With such an overlay, analysts can easily identify communication hotspots during an event. This shows how semantic technologies in conjunction with advanced visual tools transform statistical data into valuable repositories of actionable knowledge .
- Scharl, A. and Tochtermann, K., Eds. (2007). The Geospatial Web – How Geo-Browsers, Social Software and the Web 2.0 are Shaping the Network Society. London: Springer.
- Bostock, M., Ogievetsky, V. and Heer, J. (2011). “D3: Data-Driven Documents”, IEEE Transactions on Visualization and Computer Graphics, 17(12): 2301-2309.
- Brasoveanu, A.M.P., Sabou, M., Scharl, A., Hubmann-Haidvogel, A. and Fischl, D. (2017). “Visualizing Statistical Linked Knowledge for Decision Support”, Semantic Web Journal, 8(1): 113-137.