Please enable JavaScript.

Coggle requires JavaScript to display documents.

7330: Week 2 Jan 15 (Unit 2 BLT: Data Models (2.3 Data:…

- - - - architecture of typical relational database mgmt. system
      - how to use in organization
  - - - allows us to high level view structure without worrying about details
      - Collection of tools for describing:
        
        data
        
        data relationships
        
        data semantics
        
        data constraints
    - - relational model, basis of relational databases management
      - Entity-relationship data model;
        related to relational model;
        used for db design
      - object-based data models:
        object oriented, object-relational
      - semistructured data model:
        xml, json
      - network model: older
      - Hierarchal model: older
    - - Conceptual:
        
        getting closer to user
        
        objects or entities that we are storing information about
      - Logical
        
        adding of detailed infomation, attributes, constraints
      - Physical
        
        getting closer to hardware
        
        allows accessing/organizing of data: indexes
      - view represents group of localized users
    - - Sources of data:
        each has own data structure
        
        social media
        
        real time sales
        
        email archives
        
        CRM: customer relationship mgmt
      - Structure of Source Data
        
        Unstructured (raw data)
        
        eg raw text of emails
        
        Semistructured: xml/json: analysis, strict structure not important:not changing real time
        
        structured: relational: allows data to be consistent and accessible
  - - - Letters
      - emails
        
        ieee defined structure
        
        large amount of structured data "behind the scenes"
        
        Unstructured data makes up body of email
      - books
        
        Unstructured and semistructured
        
        how to make us of data?
        
        chapters
        
        pages
        
        index
      - blog posts
      - Images/videos
        
        hard to represent in relational databases
      - social media:
        
        twitter data
        
        search for trends usually involves unstructured/semistrctred data
      - system administrator log files
      - web search histories
      - Free-form text
        
        hard to analyze in relational manner
    - - Hierarchical Models
        
        xml
        
        json
      - May or not conform to strict model
    - - emails:
        tracking-dont need body;
        tracking based on content-need to make sens of body
    - - SQL: Language used to access relation databases: led to NoSQL
      - In reference to relational databases, there's rarely one tool that will solve all problems
        
        NoSQL really is more accurately not ONLY SQL
        
        Each category helps solve a problem not easily solved in a relational database
      - Development of NoSQL
        
        data had to be split onto multiple servers
        
        becomes expensive
        
        license based on server being used
        
        Organizations began developign alternatives to multiple servers
        
        Amazon, Google, Facebook
      - NoSQL Categories
        
        Key-value databases: data stored in key-value pairs
        
        Name=Mark Fontenot
        
        ID=1234
        
        Address=5901 Anywhere Street
        
        Document Databases eg mongo db:
        started from increasing complexity of key-value databases
        
        Column databases
        
        data stored in columns instead of rows
        
        way to distribute very large dataset among multiple nodes in database cluster
        
        Graph databses
        
        relational database systems don't translate well to graph data
    - - Time Series Data
        
        ordered sequence of values for a variable
        
        time stamped
        
        equally spaced intervals
        
        environment sensor data
        
        financial data
      - Geospatial data
        
        data with some spatial or location component
        
        synonymous with "spatial data" and geographic data"
        
        eg. system to find nearby restaurants
      - Multimedia Data: eg. youtube, flickr, mobile device's camera
        
        Produce multimedia data
        
        store in binary sequence
        
        picutures,images,videos
      - Data in Multiple Categories
        
        video file could be treated as a time series
        
        Pictures from camera may have gps coordinates
        
        Information can have auxiliary data connections
  - - - developed by edgar codd
        
        ibm researcher in 60s70s
        
        significant impact on data theory
        http://www.seas.upenn.edu/~zives/03f/cis550/codd.pdf
      - concept of relational model
        
        data are what can be conceptualized as a two-dimensional array eg table
        
        a relation is made up of attributes
        
        attributes have particular value
        
        sets of values are called tuples
      - Relational Model Table
        
        refer to a relation that has attributes and tuples make up data portion
        
        Attribute types
        
        columns are attributes with a domain; set of allowable values that the attribute can take on
        
        eg. final grade domain is letters
        
        Attribute values are (normally) required to be atomic, ie indivisible
        
        eg can/should separate 3 portions of a phone number?
        
        eg. can/should separate parts of a single address?
        
        tuples can't comprise multiple pieces of information
        
        one phone number per tuple
        
        special value null is a member of every domain
        
        can cause complications in the definition of many operations
        
        database systems implement null differently
        
        two null values aren't considered equivalent
        
        rows are tuple
        
        Relation Schema and Instance
        
        relation is a set of attributes related to a particular item
        
        Eg. R=A1, A2, ... An
        
        Instructor=(ID, name,dept_name,salary)
        
        Relation is a subset of Cartesia product
        
        eg. D1xD2xDn
        
        current values (relation instance) of a relation are specified by a table
        
        Properties of Relations
        
        ordering of tuples/attributes is irrelevant (may be stored in arbitrary order)
        
        Number of tuples/attributes in table is theoretically unbounded
    - - consists of multiple relations
        
        eg. instructors, students, advisers
        
        eg. products, customers, orders
      - a bad database example
        
        Univ (instructor_ID, name, dept_name, salary, student_ID,...)
        
        repetition of information (eg. two students have the same instructor)
      - Normalization theory (chapter 7) deals with how to design "good" relational schemas
    - - attribute or set of attributes in a relation that allow identification of tuples
        
        eg. social security number and student ID shoudl be able to uniquely identify a student
    - - Physical data independence
        
        Application programs and terminal activities remain logically unimpaired whenever any changes are made in either storage representations or access methods.
        
        E.g., changing data storage from single file to multiple files
      - Limited data redundancy
      - Data consistency (not stored redundantly)
        
        redundant data can take a while to update
    - - query processing
      - storage manager
      - disk storage