Abstract: Big Data processing is well-known in computer science. It requires efficient data processing. But there are also many small data sources, like bank account data, private files, etc. Here, the main problem is not efficiency, but complexity and diversity. In our opinion, the problem can be solved only by a universal end-user computer language. All people learn in school operations like addition, multiplication, subtraction, division, sqrt, sin, sum, average, etc. Therefore, a computer language should be based on operations. But the objects of these languages should be numbers, words, texts, and more general objects built from this by a pair (tuple) and an addition operation. By pairing two objects horizontally together, and by adding an element to a collection. In addition to sets and lists, bags (multisets) are also required. Empty collections are starting points for the generation of corresponding finite collections. Very often, the elements of a collection are of the same type. They can be represented in a compact table format with a few columns. Matrices can be considered as tables, too. Matrices with many zeros can be represented by tables with repeating values (groups). The join is one of the most important operations of the Relational data model. Because the join of two flat tables has to be an object of the Relational data model, it must be in first normal form, too. Therefore, it also has several disadvantages, like high redundancy. The paper starts with methodological issues. Then, a more user-friendly “join” is discussed. A few questions of query optimization are presented in Chapter IV. The last chapter generalizes a transpose operation, known from matrices, to structured tables.

Keywords: data model, end-user programming language, structured table, comment, join, query optimization, transpose

Pages: 170-178