Algebraic Graph-Oriented = Category-Theory-Based Manifesto of Categorizing DataBase Theory Third version

Foreword. Category theory is well known as a high-level polymorphic framework suitable for specifying complex structures and formalisms. Quite unexpectedly, however, it turned out that even elementary category theory notions can be valuable in practice of software engineering (see [18, 9, 12]), and in the summer of 1994 we wrote a manifesto ([7], the first version of the present one) to claim our belief in the extreme fruitfulness of incorporating the category theory (CT) machinery into the database (DB) area. That time we did not know about works by Johnson with coauthors (see [18] for references) which seem to be the first publications on the real influence of CT machinery in the DB field. So, a real mutual liking between CT and DB is about two years old. We cannot say that their relations were unclouded during these years, and now we recognize quite clearly that theoretical achievments do not directly lead to revolutions in the field of information (and any other) technologies: the normal way of changes in the latter is the way of more or less rapid evolution only more or less modulated by theoretical ideas. Nevertheless, we consider the last year as a confirmation of our previous predictions: the list of (CT+DB)-publications is actually growing (though not so rapidly as we hoped), and at a recent International Workshop ([6]) there was a special section on (CT+DB), and methodologies being developed in Sydney ([18]) and Riga ([9, 12]) seem to meet some (methodological, of course, but partly also technological) challenges in the DB area. As before, we forecast an intensive growth of CT and DB interaction and present below the third version of our manifesto in which the text is slightly revised and the bibliography list is updated.