Collection/item metadata relationships

In information organization systems, metadata is often attached to both collections and items. Collection metadata and item metadata are related: one can infer facts about items from descriptions of collections, and facts about collections from descriptions of items. This sort of reasoning, which is important to finding, understanding, and using information, is guided by specific, if usually only implicit, inference rules. This dissertation explores the general nature of these rules and develops a logic-based framework of categories for collection/item metadata rules. The resulting framework has 28 rule categories related by two logical relationships. This framework has practical applications in metadata vocabulary development, metadata-enabled search and retrieval, and metadata quality and completeness. A number of foundational questions are also discussed, including the ontological nature of collections, the logic of the collection membership relationship, the semantic and logical nature of collection/item inference rules, and difficulties in the translation of colloquial metadata records into a logic-based knowledge representation language.