Bottom-up algorithms or reasoning begin with the raw data combining individual items or units and successively constructing more complex ones from them. For example, in vision one might move from edge detection, to boundaries of shapes, to object recognition. In natural language processing one might start by combining succesive words into small phrases, and then see how these make more complex clauses or sentences. It is the opposite of top-down reasoning.
Used in Chap. 3: page 26; Chap. 12: page 186; Chap. 13: page 192; Chap. 14: page 211
Also known as antecedent-driven reasoning, bottom-up algorithm, consequent-driven reasoning, bottom-up