Graph Merge
Prov Viewer has a feature to merge two provenance graphs in order to generate a single unified provenance graph. The merge process combines the current displayed graph with another graph (chosen by the user) to generate a single unified graph for visualization. The merge process is composed of four steps: (1) vertex matching, which selects pairs of vertices from the graphs; (2) similarity verification between vertices, which receives two vertices from the first step and informs if they are similar. Vertices are considered similar if they belong to the same vertex type and have the same properties (or similar numeric values) within a configurable margin of error; (3) merge vertices that were considered similar in the previous step; and (4) creation of the unified graph for visualization, which only occurs after the matching process is over.
The figures below illustrate the graph merge of two distinct graphs from the same domain, generating the third graph. Red vertices represent vertices from the first graph, green vertices represent vertices from the second graph, and gray vertices represent vertices common to both graphs.
Red vertices in the merged graph belong exclusively to the first graph (left), while grey vertices represent common vertices (i.e., merged vertices) from both graphs, and green vertices belong exclusively to the second graph. This graph merge feature is useful when analyzing multiple sessions or trials by detecting common sections. Merged vertices from this feature also provide similar summarized information using the five-number summary.
To be described in the future.
To be described in the future.
Not supported yet. Currently, Prov Viewer uses the default merge configuration, which matches vertices that has 95% of similarity.
A merge configuration file will be created in the future, allowing the user to define the threshold and/or error margin for each type of attribute and inform acceptable values when dealing with attributes that uses String instead of numbers.