In an exploratory context, users often have no specific goal in mind and aren't seeking a particular outcome. However, when they discover something intriguing, it's important for them to retrace their steps and understand how they arrived at those results. Additionally, validating hypotheses may require branching out to compare data across different views.
MGExplorer is designed to facilitate data exploration with a flexible visualization approach. Utilizing the concept of chained views, MGExplorer allows users to visualize their analytical journey through a sequence of views. This feature supports various visualization techniques across multiple datasets, enabling users to explore different scenarios, retrace their steps, and refine their analysis as needed.
The chained views methodology in MGExplorer lets users seamlessly integrate multiple visualization techniques during their exploration process. It records every step of the journey, providing a detailed map of the analytical path. This not only enhances understanding but also supports comprehensive analytical provenance studies.
With MGExplorer, you can unlock the full potential of your data, uncover hidden insights, and navigate your exploratory path with ease.
The node-link diagram represents nodes as items and the edges between them as relationships, offering a clear overview of any network in the dataset based on criteria such as keywords or co-publications. It is the primary technique used by the application, serving as the foundation for all other visualization techniques.
Users can interact with the diagram by hovering over nodes to highlight them and by right-clicking a node to generate a new visualization based on its data. For example, if nodes represent authors and edges indicate co-authored works, selecting a node will filter the data to focus on that author. The diagram also includes filters to help locate specific nodes in large networks and allows users to adjust settings like distance, gravity, and repulsion to refine the network view.
The Cluster View technique visualizes clusters based on relationships among data items. It features a multi-ring layout: the innermost ring displays the data items as circles, while the outer rings represent various data attributes using rectangles. Items within the same cluster are connected by curved lines.
By default, it displays four rings, but this can be adjusted to fit the number of data attributes and rings needed. Users can interact with the view by hovering over the connecting lines to highlight different clusters, making it easy to explore and understand the relationships within the dataset.
The Bar Chart technique visualizes the distribution of data attributes' values over time or across categories. By default, the x-axis represents temporal information, while the y-axis shows the count of associated items per time period.
The data is presented as either a single bar for each time period or multiple colored bars to represent different categories or attributes. This approach allows for a clear comparison of attribute values across different time periods or categories, making trends and patterns easy to identify.
The Listing View technique presents a detailed list of elements that define the relationships between two or more nodes in a graph. Each item in the list is linked to a descriptive web page within the dataset, allowing users to access additional information about each element. This view facilitates a deeper exploration of the relationships by providing direct access to further details about each listed item.
The Pairwise Relationship View technique visualizes clusters using a matrix where both rows and columns represent data items. Each cell in the matrix contains a glyph that encodes attributes describing the pairwise relationships between the items.
The default glyph resembles a star plot, with multiple axes extending from a central point. Each axis represents a different data attribute and is used to encode its corresponding value.
Hovering over a glyph enlarges it, allowing users to view detailed information about the data attributes. This interaction facilitates a deeper understanding and analysis of the relationships within the cluster.
The technique isolates a data item of interest at the center of a circular view, displaying the remaining related data items around it. Inspired by the eye's iris, which focuses on a limited field of view, this visualization technique places the selected item at the center and arranges its related items in a way that those within the central field of view appear larger than those further out.
Data attributes for these relationships are represented by bars connecting the central item to each related item, with the height and color of these bars encoding different attributes. Users can shift the focus by clicking on any item, moving it to the center and updating the view to highlight new relationships.
The visualization tool includes a querying process that enables users to retrieve datasets from multiple SPARQL endpoints and explore them within a single dashboard. This is managed through a query panel, which lists available SPARQL endpoints and supported queries. Queries are created and managed through the LDViz interface
When a query is selected, the panel displays custom parameters to filter the data based on criteria like publication period or research institution. Users can then choose a visualization technique and click "Run" to retrieve data, apply the query, transform the data, and generate the selected visualization. The tool uses a cache to store query results temporarily, which helps reduce the frequency of data server access. Users can clear the cache to ensure that the next query retrieves updated data from the server.
The query panel provides an interface for selecting SPARQL endpoints and queries. It includes:
*This work is developed by the Wimmics team at the Centre Inria, University Côte d'Azur. It results from collaborative efforts involving several researchers and students.