As stated earlier, the COMPERE system provides the majority of the functionality of this supertask. The augmented text used as input is designed to circumvent the areas I have identified in the sentence processing supertask which COMPERE does not handle. These include things such as punctuation analysis and in-depth analysis of conjunction usage. As shown in Figure 32, COMPERE is used to parse the words of the phrase while punctuation analysis and tense analysis work on the markers. COMPERE produces a partial conceptual entity which is passed on to the integrator task. This task takes the partial conceptual entity along with the results from the other two tasks and synthesizes it all into a single conceptual entity. This is then passed back to the control supertask.
At the current level of implementation, neither punctuation analysis nor tense analysis are complete. For example, tense analysis is limited to identifying present, past, and future tenses, with no ability to detect more subtle nuances of the languages (e.g., He is walking to the store and He walks to the store are viewed the same; He walked to the store is the same basic concept already completed; and He will walk to the store is the concept placed in the future.). Punctuation analysis is more complete; the main element missing is the identification of capitalization, which could provide useful information to the reader. These elements were left incomplete in the current instantiation due to time constraints and the fact that their current level of implementation is sufficient for the model to demonstrate its creative reading capabilities.
