It has been recently found using graph theory that measures of network structure can predict ratings of dream complexity, where increases in connectedness and decreases in randomness are observed in relation to increasing dream report complexity. This approach proved to be useful to differentiate dream reports in the pathological population as well as NREM and REM dream reports but it has not yet been used to study the differences between different oneiric experiences. In this work we analyse dream reports that include non-lucid, lucid dreams and out-of-body experiences initiated from sleep paralysis. The reports are presented as directed graphs, where each different word plays the role of a node, and consecutive words are connected by a directed, unweighted edge. We analyse different network measures to compare the graphs. Preliminary results presented here suggest that both local measures, such as the degree of nodes, and global measures, such as clustering and the number of strongly connected components, allow for a categorization of different dream experiences.