1. The paper introduces a relational model of data as a superior alternative to existing network and tree models for large shared databases. It allows data to be described in its natural form without imposing additional structure.
2. The relational model provides a basis for treating issues like redundancy, derivability, and consistency of data relations. It also allows clearer evaluation of existing data systems and representations.
3. Existing systems still impose dependencies between data representation and user interaction models related to ordering, indexing, and access paths that limit changes without impacting applications. The relational model aims to reduce these dependencies.