INFORMATİON TECHNOLOGİES

A database (DB) is defined as a collection of related data.

Data models can be grouped as conceptual or high-level data model, low-level or physical data models and implementation or representational data models.

Conceptual design provides conceptual schema of DB. It allows representing entities and relationships between them in high-level view models. Entity-Relationship model is one of the popular data model used for conceptual design. It represents data as entities, attributes and relationships.

A database schema is the description of a database including structure and constraints that the database should obey.

Physical design specifies database in terms of physical configurations of chosen DBMS, such as indexes, physical file storages.

Data cleaning includes processes of filling missing values, correcting inconsistencies, identifying outliers and getting rid of them. Missing values can be filled with attribute mean, attribute mean of all tuples belonging to the same class of the given tuple containing missing values, overall system mean, and a prediction produced by a learning algorithm.

Data integration merges data from different sources such as databases, data flats etc. into a compatible data store.

Classification is the process of understanding data classes or concepts. Class labels of data are given for purpose of classification.

Association is referred as patterns frequently purchased together. Mining association is the process of discovering relations among data.

Metadata is generally referred as data about other data or catalog for a data. It describes, identifies, explains, manages, and uses a data resource.

Data can be grouped as qualitative data and quantitative data. Quantitative data can be referred as data in numerical form whereas qualitative data can be referred as data different than numerical form. A quantitative or numeric data can be continuous or discrete. A continuous data can be any value from the specified range. A discrete data can be any one of the specified discrete numerical values.

Data can be referred as raw, unorganized facts and concepts which need to be processed. Information is organized, categorized, contextualized data. Data becomes meaningful with information. Information needs to be analyzed especially for making strategic decisions. Information does not need to be useful. Knowledge is simply defined as useful information. Wisdom is related with judging which knowledge is true and applying it to real life.

Managing data is important and provides some advantages described as follows:It provides mechanisms to become data into accessible, reliable, authentic and complete form; it provides mechanisms to securely share data between information technology systems, applications, departments or employees for an enterprise; it helps decision makers to make strategic decisions by allowing mining and analyzing data; it allows organizing, tracking, deleting, creating, and reporting data; it disallows redundancy and loss in data.

Master data is a data that describes core entities for an enterprise such as customers, assets, employees and so on. Master data management is a set of tasks, such as governing, controlling master data and so on. Master data management provides sharing data among departments and employees.

An attribute can be any type of composite, derived, simple and multivalued. A simple attribute is the attribute which is not divisible. To illustrate, an id property for a book is simple attribute.

Relational schema have some restrictions such as domain constraints, key constraints, entity integrity constrains and referential integrity constraints. According to entity integrity constraints, a primary key cannot be NULL value.

Mapping 1:1 binary relationship types: There are three possible ways to be followed while mapping 1:1 binary relationship type R relating two entity types concept to relational design. One of the choices is foreign key approach where primary key of one of the entities is inserted to the other entity type’s relation as foreign key. Merging relation approach is valid when both entity types are total participated in R. This time, two different entity types are merged into one relation. As a third choice, a relationship is created where primary keys of two entity types are used as foreign keys. One of those foreign keys are chosen as primary key of R.

Data cleaning includes processes of filling missing values, correcting inconsistencies, identifying outliers and getting rid of them. Missing values can be filled with attribute mean, attribute mean of all tuples belonging to the same class of the given tuple containing missing values, overall system mean, and a prediction produced by a learning algorithm.

Bayesian classifier is one of the mostly known classification techniques. It is based on some probabilistic computations such as computing probability of an object belonging to a class. Bayesian classifiers use Bayes’ theorem. According to Bayes’ theorem, let X be an object and H be a hypothesis that X belongs to class C. The probability of H holds is shown as P(H|X) and it is calculated as follows: P(H|X) = P(X|H)*P(H)/P(X).

Metadata can be grouped as descriptive, structural and administrative metadata. Descriptive metadata identifies a data resource. Administrative metadata defines the ways of managing a data resource. Structural metadata describes relationships between parts of a composite object.