{"CAPTION TAB5.png": "'\\n\\n[MISSING_PAGE_POST]\\n\\n'", "CAPTION FIG3.png": "'\\n\\n**Figure 3.** **Participants generate similar numbers of good ideas across conditions, adjusted for baseline fluency.**'", "CAPTION TAB3.png": "'Table 3. Machine-generated paths increase number of **idens relative to control condition. Means are adjusted for baseline fluency**.\\n\\n'", "CAPTION TAB4.png": "'\\n\\n**Table 4. External stimulation increases breadth of search by relative to control condition, for both problems.**'", "CAPTION FIG1.png": "'Figure 1. Screenshot of human labeling task (left) physical setup of human cluster-label task (right)\\n\\n'", "CAPTION FIG2.png": "'Figure 2: Screenshot of ideation interface. Actual inspirations from the machine-machine condition are shown. Participants enter ideas on the left, and view inspirations on the right. Participants can star inspirations they find useful.\\n\\n'", "CAPTION TAB1.png": "'\\n\\n**Table 1. Example solution paths extract by each method for both problems. For comparability, we select examples that are close in semantic meaning.**'", "CAPTION TAB2.png": "'\\n\\n\\\\begin{table}\\n\\\\begin{tabular}{l l} \\\\hline \\\\hline \\\\multicolumn{1}{c}{} & \\\\multicolumn{1}{c}{} & \\\\multicolumn{'"}