the content outlines the concepts of semantic transition trees (stt) and transition frames in the context of artificial intelligence and language processing. stt is a linear branching structure that represents possible sequences of word classes based on semantic properties. each stt is paired with a semantic translation, allowing the system to map input strings (like english sentences) to output strings (like database queries). for example, the sentence "what size is the m16" can be represented as "what ?noun-group1 is ?noun-group2", with semantic translations mapping "the size" to a database query. the content also discusses event structures, which are used to represent the meaning of sentences. event structures include multiple subrepresentations, such as thematic representations (which identify the main actors in a sentence), trajectory frames (which represent motion along a path), and state transition frames (which represent changes in quantities over time). the content provides an example of how to construct a trajectory frame for the sentence "the army grew larger before declining and then reached a constant size before disbanding." finally, the content describes how to build an stt that outputs trajectory representations instead of database queries. this involves writing an stt that admits english-like trajectory patterns and outputs a trajectory representation. the content provides a template for defining trajectory trees, including the structure of trajectory frames, path frames, and place frames. the template includes predefined prepositions and their meanings, as well as examples of how to represent different types of things, such as troops, enemies, and locations.the content outlines the concepts of semantic transition trees (stt) and transition frames in the context of artificial intelligence and language processing. stt is a linear branching structure that represents possible sequences of word classes based on semantic properties. each stt is paired with a semantic translation, allowing the system to map input strings (like english sentences) to output strings (like database queries). for example, the sentence "what size is the m16" can be represented as "what ?noun-group1 is ?noun-group2", with semantic translations mapping "the size" to a database query. the content also discusses event structures, which are used to represent the meaning of sentences. event structures include multiple subrepresentations, such as thematic representations (which identify the main actors in a sentence), trajectory frames (which represent motion along a path), and state transition frames (which represent changes in quantities over time). the content provides an example of how to construct a trajectory frame for the sentence "the army grew larger before declining and then reached a constant size before disbanding." finally, the content describes how to build an stt that outputs trajectory representations instead of database queries. this involves writing an stt that admits english-like trajectory patterns and outputs a trajectory representation. the content provides a template for defining trajectory trees, including the structure of trajectory frames, path frames, and place frames. the template includes predefined prepositions and their meanings, as well as examples of how to represent different types of things, such as troops, enemies, and locations.