OCR Computing A Level - Unit 04 - Exchanging data

Lossy compression

Non-essential data is permanently removed

JPEG and MP3 use lossy compression

JPG removes data permanently to reduce file size then tries to reconstruct without the missing data

MP3 removes the sound in frequency ranges that are outside of human hearing or least affect the perceived playback quality, and quieter notes played at the same time as louder sounds are also removed

Lossless compression

Patterns in the data are spotted and summarised in a shorter format without permanently removing any information

Works by recording patterns in the data rather than the data itself, meaning a new file can be replicated without any loss of data

PNG and GIF use lossless compression

Run Length Encoding (RLE)

A basic method of lossless compression that summarises consecutive patterns of the same data

Works well with image and sound where data could be repeated many times

A sound recording could have thousands of samples taken a second, so the same sound or note played for a fraction of a second could result in hundreds of identical samples, so RLE records one example of the sample and how many times it consecutively repeats

Dictionary compression

Spots regularly occurring data and stores it separately in a dictionary

The reference to the entry in the dictionary is stored in the main file, reducing the original data stored

Even though the dictionary produces additional overheads, the space saving negates this problem

Encryption

A way of making sure data cannot be understood if you don’t possess the means to decrypt it

Encrypts and decrypts by performing an XOR operation with a shared key, bit by bit

Plaintext

The message to be sent before encryption

Ciphertext

The plaintext is encrypted using a cipher algorithm and key into equivalent ciphertext, which can then be decrypted back into plaintext using the same or different key

Caesar cipher

Most basic type of encryption and most insecure

Letters of the alphabet are shifted by a consistent amount

Brute force attack

An attack which attempts to apply every possible key to decrypt ciphertext until one works

Frequency analysis

Letters are not used equally often; in english, E is the most common letter (followed by T,A,O,I,N,S,R,H)

This attack analyses patterns in language to check possibilities and make substitutions of letters in ciphertext to break substitution ciphers

Vernam cipher

The encryption key, also known as a one-time pad, is unbreakable

The key must be a truly random sequence greater than or equal in length than the plaintext and only ever used once

Shared with the recipient by hand, independently of the message and destroyed immediately after use

One-time pad

Must be truly random, generated from a physical or unpredictable phenomenon (e.g. atmospheric noise, radioactive decay, snapshots of a lava lamp)

A truly random key will render any frequency analysis useless as it would half a uniform distribution

Computer generated ‘random’ sequences are not actually random

Database

A structured way to store data so that it can be retrieved using queries

Can hold data about several entities

Uses of a database

School use -- management information such as SIMS or iSAMS, learning platforms such as Google Classroom

Organisation or business use -- customer relationship management software, online calendars, backend of most websites to dynamically generate pages

Personal use -- social media data, multiplayer games

Flat file

Simplest kind of database

Consists of information about a single entity

Entity

An object, person, event or thing of interest about which data needs to be recorded

Entity descriptions can be written in format: e.g: Customer(__custID__, title, firstname, surname, email)

Each entity needs an identifier which uniquely identifies a particular record, known as the primary key

Natural primary key

A naturally occurring unique identifier

E.g. mobile number, national insurance number

Composite primary key

When 2+ attributes are needed to uniquely identify a record

Secondary key

In some databases, searches may often need to be made on other fields than just the primary key

If searches often need to be made on title or subject, either or both of these fields could be defined as a secondary key, and then would be indexed for faster lookups

Entity relationships

One-to-one (e.g: husband and wife)

One-to-many (e.g: mother and child, school and pupil)

Many-to-many (e.g: actor and film, recipe and ingredient)

Entity-relationship diagram

A graphical way of representing relationships between entities

Foreign key

An attribute that creates a join between 2 tables (relations)

It is the primary key in the first relation

It always goes on the ‘many’ side of a relationship

Referential integrity

No foreign key in one table can reference a non-existent record in a related table -- all data is consistent over entire database

Values in fields should be atomic (a single attribute cannot consist of 2 data items, e.g: first name and surname)

No 2 records can be identical

Each table needs a primary key

Normalisation

A process used to come up with the best possible design for a database

No duplication, consistent data, flexibility, complex querying

Partial dependency

When the fields are only dependent on part of the primary key (only occurs if primary key is a composite key)

Non-key dependency

Where the value of a field is determined by the value of another field that isn’t part of the primary key

First Normal Form (1NF)

A table contains no repeating attributes or groups of attributes

All field names must be unique

Values in fields should be the same domain

Values in fields should be atomic (lowest level of detail; a single attribute cannot consist of 2 data items)

No two records can be identical

Each table needs a primary key

Second Normal Form (2NF)

Ensure data is already in 1NF, all partial dependencies have been removed

(partial dependencies fixed by splitting table so that the part of the composite key becomes the primary key of a new table, with all fields solely dependent on the key put into the table)

If a field or group of fields can be inferred from a field that isn’t a primary key, then they can be split into a separate table

Third Normal Form (3NF)

Ensure data is already in 2NF, any transitive dependencies have been removed

(remove any non-key dependencies)

‘The fields are dependent on the key, the whole key and nothing but the key’

SQL

Structured Query Language

A declarative language used for creating, querying and updating tables in a relationship

Common data types in SQL

Everyday methods of data capture

Barcode scanners

Cookies

Contactless

Form/census

Automated methods of data capture

Smart card readers

Scanners

Optical character recognition (OCR)

Optical mark recognition (OMR)

Magnetic ink character recognition

Sensors

Methods of inputting data into database

Automatically, using a DBMS software

Typing manually into customised form

Importing from spreadsheet or file

Using EDI (Electronic Data Interchange) -- used to transfer data between one computer system and another

EDI (Electronic Data Interchange)

The computer-to-computer exchange of documents such as purchase orders, invoices, shipping documents between 2 companies or business partners, etc.

It replaces post, email or fax

All documents must be in a standard format so that the computer can understand them, to reduce errors and confusion in systems

Transaction processing

A single logical operation in databases is defined as a transaction

It may consist of several operations, all of which must be processed and checked appropriately

ACID

Atomicity, Consistency, Isolation, Durability

A set of properties to ensure that the integrity of the database is remained under all circumstances

Atomicity

This property requires that a transaction is processed in its entirety or not at all

In any situation, including power cuts or hard disk crashes, it is not possible to process only part of a transaction

Consistency

This property ensures that no transaction can violate any of the defined validation rules/referential integrity is upheld

Any changes to a database have to be reflected across the entire database (e.g: if a part of a table has been removed, all other instances of that part in other tables have to be fully removed as well

Isolation

This property ensures that concurrent execution of transactions leads to the same result as if transactions were processed one after the other

Crucial to a multi-user database

Any transactions that are happening at the same time cannot affect the quality and consistency of the data

Durability

This property ensures that once a transaction has been committed, it will remain so, even in the event of an issue such as a power cut or system crash

All information is stored, non-volatile and can be retrieved later

As each part of a transaction is completed, it is held in a buffer on disk until all elements of the transaction have been completed, only then will changes be made to database tables

Potential problems with multi-user databases

Multiple users simultaneously accessing a database could potentially cause one of the updates to be lost

Record locking

Prevents simultaneous access to objects in a database in order to prevent updates being lost or inconsistencies in the data arising

A record is locked when a user retrieves it for editing or updating; anyone else attempting to retrieve it is denied access until the transaction is completed

Issue: could result in a deadlock if 2 users are attempting to update 2 records, and neither can proceed as each has locked the other record

Serialisation

Ensures that transactions do not overlap in time and therefore cannot interfere with each other or lead to updates being lost

Techniques: Timestamp ordering, commitment ordering

Timestamp ordering

Each object in the database has a read timestamp and write timestamp, updated whenever an object is read or written

When a user tries to save an update, if the read timestamp is not the same as it was when they started the transaction, the DBMS knows another user has accessed the same object, so the transaction will be cancelled with an ‘Update unsuccessful’ message sent to the user

Commitment ordering

Ensures that no transactions are lost of 2 clients are simultaneously trying to update a record

Transactions are ordered in terms of their dependencies on one another as well as the time they were initiated

Can be used to prevent a deadlock by blocking one request until another is completed

Redundancy

Duplicate hardware, located in different geographical locations, mirrors every transaction that takes place on the main system, so that if it fails the backup system automatically takes over

Many organisations cannot afford to have their computer systems go down for even a short time (e.g. air traffic control) and so have a built-in redundancy in their computer systems