The article deals with the methodology of creation a universal data storage for applications oriented to the specialists who participate in scientific researches. It is assumed that the storage structure is not rigidly bound to any scientific domain, and it will be demanded in many projects related to the data accumulation, data analysis and processing. The proposed approach to the organization of the storage system is based on fundamental principles, concepts and technologies that used in the software design process. This process is based on the subject domain analysis and the detection criteria of commonality and variability which guarantee high abstraction level as one of the main tools for working with data.
The creation of software systems for scientific researches is a difficult task. As
increasing their complexity, the appropriate software products building require more
and more time, and the development costs grow exponentially. It also related to
design process and implementation of the storage subsystems for applications, which are
used in scientific activity, and also provide researchers of diverse, but at the same
time actual and complete information. Many scientific laboratories or research centers
could work with applications based on the proposed data model because of its
universality and the high level of abstraction when describing the properties of objects and
their classification. For the specialists making an experiment it is necessary to obtain
and process the correct and complete information connected with research problems,
but during of such researches is negatively influenced some factors requiring change
the structures of data storage [
Objects of any type can be described in terms of some simplified data model for
which the basic concepts are entities and attributes. Attributes are the predefined
lexical units or descriptors needed to describe the basic semantic meaning of the key
concepts in the data domain.
When such concepts are defined, it is not difficult to move to a more specific
description. It is possible to extend semantics of the selected conceptual model to the entities
relationships and relationships attributes.
This example is a very simplistic, but at the same time it can support most
applications and provide adding data of any type without specific names of tables or fields
that are required when using a relational data model [
Classification is the most simple and at the same time, the most often solvable task of
the universal model of data storage. The result of classification is the detection of the
signs characterizing groups of the researched objects as classes to which is possible to
refer new object. It is important to keep common concepts that defines the
commonality and variability, and also other features already classified or again obtainable or
analyzable information. In fact, classification is any system distribution of objects,
phenomena or concepts by any essential signs selected for convenience of their
representation and processing. Thus, the classified information can be provided as the set
of the abstract or concrete entities ordered by some principle, which have similar
classification criteria (one or more properties). It is necessary for determination of
criteria of a commonality in the description, behavior or any other dimensions of
source unstructured data. Usually classification tools are separate by a method of their
impact on the classified element of information. For example, it is possible to
organize internal classification, which can be made on the essential signs characterizing a
commonality of objects, concepts or phenomena. The data domain, in this case,
describes by the elementary units of data, related to some primitives, which are
determined a priori. The main concepts are “meta-type”, “instance type”, “instance”,
“hierarchy type” and “relationship attribute”.
These concepts determine a set of tables that fixes the entity-relation structure in the
database. A unique feature of such approach is the possibility of arrangement of
unlimited number of objects over a limited set of concepts. For example, the concepts in
the materials science are connected to definitions of graph theory that allows one to
describe crystal-chemical data [
The storage structure is constructed by the principles of universal data model logically
can be separated into some interdependent components. These include: classification
tools (typification and categorization), entity instances (with a possibility of
assignment for instance type a specific property set), entity relationships (allowing to
specify the relationship types and values of the appropriate attributes), properties palette
(combining mechanisms to define primitive and composite properties).
This makes it possible to use the proposed universal data model in different fields of
scientific research. These problems can be in the scope of theoretical and
experimental chemistry connected with design and structural analysis of compounds and
new materials [
The following are code samples from the DDL-script for the database creation using the universal model of data storage. 1. Code sample of creation of the meta-types table, as a set of the elementary primitives related to some internal base concepts of data domain, which are defined a priori.
( row_id uuid NOT NULL DEFAULT uuid_generate_v4(), alias varchar(128) NOT NULL, description text, CONSTRAINT meta_type_pkey PRIMARY KEY(row_id),
CONSTRAINT meta_type_uindex UNIQUE(alias) ); 2. Code sample of creation of the instance types table, as a list of the concepts defining the criterion of identity that distinguish the corresponding entity from other entities, which aren’t corresponding to this criterion.
CREATE TABLE IF NOT EXISTS instance_type ( row_id uuid NOT NULL DEFAULT uuid_generate_v4(), meta_type_id uuid NOT NULL, alias varchar(128) NOT NULL, description text, CONSTRAINT instance_type_pkey PRIMARY KEY(row_id), CONSTRAINT instance_type_fkey
FOREIGN KEY(meta_type_id)
REFERENCES meta_type(row_id),
CONSTRAINT instance_type_uindex UNIQUE(alias) ); 3. Code sample of the table creation of category types, as a set of the elementary primitives related to some external concepts of data domain, which are determined a priori.
CREATE TABLE IF NOT EXISTS category_type ( row_id uuid NOT NULL DEFAULT uuid_generate_v4(), alias varchar(128) NOT NULL, description text, CONSTRAINT category_type_pkey PRIMARY KEY(row_id), CONSTRAINT category_type_uindex UNIQUE(alias) 4. Code sample of creation of the categories table as a tool providing necessary flexibility through the additional level of indirection in case of determination of criteria of commonality for entities independent of their typification.
CREATE TABLE IF NOT EXISTS category ( row_id uuid NOT NULL DEFAULT uuid_generate_v4(), category_type_id uuid NOT NULL, alias varchar(128) NOT NULL, description text, not_available boolean, CONSTRAINT category_pkey PRIMARY KEY(row_id), CONSTRAINT category_fkey
FOREIGN KEY(category_type_id)
REFERENCES category_type(row_id),
CONSTRAINT category_uindex UNIQUE(alias) YablokovDE,SaleevVA. Universal… 5. Code sample of creation of the categories hierarchy table for determining external criteria of commonality by the principle of similar behavior.
CREATE TABLE IF NOT EXISTS category_hierarchy ( category_id uuid NOT NULL,
parent_category_id uuid NOT NULL, CONSTRAINT category_hierarchy_pkey
PRIMARY KEY(category_id, parent_category_id), CONSTRAINT category_hierarchy_fkey_1
FOREIGN KEY(category_id)
REFERENCES category(row_id), CONSTRAINT category_hierarchy_fkey_1
FOREIGN KEY(parent_category_id)
REFERENCES category(row_id) 6. Code sample of creation the table of linked categories that contains data on the external of commonality by the principle from part to whole. CREATE TABLE IF NOT EXISTS category_link ( category_id uuid NOT NULL, linked_category_id uuid NOT NULL, CONSTRAINT category_link_pkey
PRIMARY KEY(category_id,linked_category_id), CONSTRAINT category_link_fkey_1
FOREIGN KEY(category_id)
REFERENCES category(row_id), CONSTRAINT category_link_fkey_2
FOREIGN KEY(linked_category_id)
REFERENCES category(row_id) 7. Code sample of procedures for creation the data row in the table meta_type.
CREATE OR REPLACE FUNCTION create_meta_type_row ( meta_type_id uuid, alias varchar(128), description text = NULL ) RETURNS VOID AS $BODY$ BEGIN
INSERT INTO meta_type(row_id, alias, description) VALUES(meta_type_id, $2, $3); Information Technologyand Nanotechnology(ITNT-2016) END; $BODY$
LANGUAGE plpgsql; 8. Code sample of procedure for update the data row in the table instance_type.
CREATE OR REPLACE FUNCTION update_instance_type_row ( instance_type_id uuid, bit_mask integer, meta_type_id uuid = NULL, alias varchar(123) = NULL, description text = NULL ) RETURNS void AS $BODY$ BEGIN
UPDATE instance_type t SET meta_type_id =
CASE bit_mask & 1
WHEN 1 THEN $3
ELSE t.meta_type_id
END, alias =
CASE bit_mask & 2
WHEN 2 THEN $4
ELSE t.alias
END, description =
CASE bit_mask & 4
WHEN 4 THEN $5
ELSE t.description
END
WHERE t.row_id == instance_type_id; END; $BODY$
LANGUAGE plpgsql; 9. Code sample of data reading from the table category_type.
CREATE OR REPLACE FUNCTION read_category_type_rows (
category_type_ids VARIADIC UUID[] ) RETURNS SETOF record AS $BODY$
RETURN QUERY SELECT t.* FROM category_type t
WHERE t.row_id == ANY(category_type_ids); END; $BODY$
LANGUAGE plpgsql; 10. Code sample of procedure for deleting data row from the table category_link.
CREATE OR REPLACE FUNCTION delete_catgory_link_rows ( category_id uuid, use_linked boolean = NULL ) RETURNS void AS $BODY$ BEGIN
DELETE FROM category_link t WHERE $1 =
CASE COALESCE($2, false)
WHEN true THEN t.linked_category_id
ELSE t.category_id
END;
The problem of creation of universal data storage becomes significant in the case of
implementation of unstructured data collection using an object-oriented approach [