184103.fig.0012
Figure 12: Transactions between human user and IMAGES-M while text understanding, map composition and question-answering on the map (At headers: “u” = human user, “s” = IMAGES-M).