Unlike humans, a computer cannot interpret the play on words, and infer that the predicate in the answer (i.e., "ajar") is being cleverly used as a noun (i.e., "a jar"). What the computer does in an AES context is to analyze the written text into its observable components. Different AES systems evaluate different numbers of these components. Page and Peterson (1995) referred to these elements as "proxes" or approximations for underlying "trins" (i.e., intrinsic characteristics) of writing. It is the observable components that automated essay scoring engines, identify, computationally, and subsequently use to compute essay scores. AES statistical models are developed by weighting the various observable components as they relate to intrinsic characteristics of writing. For example, a model in the PEG system, might be formed by taking five intrinsic characteristics of writing (content, creativity, style, mechanics, and organization)
annaabi.ee 
