Figure 1: Graph visualization for the lemma “run”. Each sense is displayed as a rectangle (a sense node) and is labeled with the corresponding sense id. Each resource is assigned a distinct color that is used for color-encoding the resource membership of a sense. Senses that are aligned to each other are linked to an alignment node (red nodes) which itself is linked to the root node respresenting the lemma of interest (black node in the middle). Similarly, all nodes that are not part of an alignment are directly linked to the root node.
Figure 2: Experiment with a more advanced graph layout algorithm (stress layout with manual adjustments); again the lemma “run” is shown. Compared to Fig. 1, we adapted the colors in a way that the different resources are assigned colors with a similar perceptual difference (to avoid the undesired effect of grouping) and removed unnecessary labels.
Figure 3: A grid-based design (here for the lemma “align”) stressing the fact that alignments form clusters. The view contains one column per sense cluster and one row per sense.
Figure 4: In the final cluster-based design each sense is represented as a colored circle (where color = resource). Senses within the same cluster are grouped together and enclosed with a border line. Within each cluster the senses are ordered according to their resource membership which eases an estimation of the distribution of resources.
A poster paper describing the iterative and highly collaborative design process and summarizing lessons learned can be found here: Is there a Value in Detours? – Experiences with Designing a Visual Browser for the Linked Lexical Resource UBY