Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Geog 480: Principles of GIS Guofeng Cao CyberInfrastructure and Geospatial Information Laboratory Department of Geography National Center for Supercomputing Applications (NCSA) University of Illinois at Urbana-Champaign What we have learned • Database concepts: • Database and Database Management System (DBMS) • Elements of a DBMS • Transaction Management: Recovery and Concurrency • Relational Database o Relations (tables) o Operations on relations: select, project, join o Relational database and spatial data: • Structure of spatial data does not naturally fit with tables • Performance is impaired by the need to perform multiple joins with spatial data (spatial join) • Indexes are non-spatial in a conventional relational database Conceptual data model • A conceptual data model provides a model of the proposed system that is independent of implementation details • An effective conceptual model will o provide a means for communication between analysts, designers and users o aid the design of the system o provide basic reference material for implemented system Entity relationship model #1 • The entity relationship model is a conceptual data modeling technique where o An entity type represents a collection of similar objects o An entity instance is an occurrence of a particular entity o An attribute type is a property associated with an entity attribute type entity type • An attribute type that serves to uniquely identify an entity type is called an identifier /key o Identifiers/keys are usually underlined identifier Entity relationship model #2 • Entity types are connected using relationships o A relationship type connects one or more entity types o A relationship occurrence is a particular instance of a relationship • Relationships may have their own attributes independent of entities • Entity, attribute, and relationship types are shown in an entity relationship diagram (E-R diagram) relationship type Entity relationship model #3 • Relationship types may be o many-to-many: e.g., a town may have many road, which in turn may pass through many towns o many-to-one: e.g., a town may have many cinemas, but a cinema can be located in at most one town o one-to-one: e.g., a cinema may have one manager who manages only one cinema • These constraints constitute cardinality conditions Entity relationship model #4 • In addition to cardinality conditions, relationships may also have participatory conditions: o optional or mandatory (indicated with a double line) • A relationship from an entity to itself is called involutory Buddies Drinkers • A relationship connecting three entities is called a ternary relationship Design Guidelines • Avoid redundancy. • Don’t use an entity when an attribute will do. 8 Avoiding Redundancy • Redundancy occurs when we say the same thing in two different ways. • Redundancy wastes space and (more importantly) encourages inconsistency. o The two instances of the same fact may become inconsistent if we change one and forget to change the other, related version. 9 Example: Good or Bad? name Beers name ManfBy addr Manfs manf This design states the manufacturer of a beer twice: as an attribute and as a related entity. 10 Example: Good or Bad? name Beers name ManfBy addr Manfs This design gives the address of each manufacturer exactly once. 11 Example: Good or Bad? name manf manfAddr Beers This design repeats the manufacturer’s address once for each beer; loses the address if there are temporarily no beers for a manufacturer. 12 Entity Versus Attributes An entity should satisfy at least one of the following conditions: • o o It is more than the name of something; it has at least one nonkey attribute. or It is the “many” in a many-one or many-many relationship. 13 Example:Good or Bad? name Beers name ManfBy Manfs 14 Example: Good or Bad? name Beers name ManfBy addr Manfs •Manfs deserves to be an entity set because of the nonkey attribute addr. •Beers deserves to be an entity set because it is the “many” of the many-one relationship ManfBy. 15 Example: Good or Bad? name manf Beers There is no need to make the manufacturer an entity type, because we record nothing about manufacturers besides their name. 16 Extended entity relationship model • The extended entity relationship model (EER) adds further features: o An entity type E1 is a subtype of E2 if every occurrence of E1 is also an occurrence of E2. In this case, E2 is a supertype of E1 o The operation of forming subtypes is called specialization; the inverse operation of forming supertypes is called generalization • For specialization (and conversely for generalization) o A subtype has the same identifying attribute(s) as the supertype o A subtype has all the attributes of the supertype, and possibly some more o A subtype enters into all the relationships in which the supertype is involved, and possibly some more. • Subtypes and supertypes are organized into an inheritance hierarchy Extended entity relationship model • Subtypes may be: o disjoint: where no occurrence of one subtype is an occurrence of another o overlapping: subtypes are not disjoint • EER uses an extended diagrammatic notation to represent specialization/generalization constructs supertype subtype disjoint overlapping EER for spatial information #1 • E-R or EER can be used to model spatial entities • Most vectorbased GIS use a similar structure (Coverage file or Geodatabase of ArcGIS) node directed arc area EER for spatial information #2 Relational database design: From E-R model to Database Schema • An E-R model can be transformed into a relational database scheme • Advantageous features for a relational database scheme are: o Lack of redundancy (redundant data wastes space and causes integrity problems) o Fast access to data • There usually exists a balance between space (lack of redundancy) and speed (fast access to data) o Many relations leads to lower redundancy, but more joins (slower speed) o Fewer relations leads to fewer joins (slower speed), but greater redundancy (and integrity problems) Example Star Cast Name Birth year Film M N Title gender Role director year length Redundancy • For example, the following relation and relation scheme will be able achieve fast access but involves considerable redundancy Removing redundancy Building relational schemes • Another guideline is to ensure relations are in first normal form, a process known as normalization • A first pass at building a relational scheme from an E-R model is to: o Convert each entity into a relation o Convert each relationship into a relation • However, not all relationships will require a relation (combining relations) o For entities in a mandatory many to one relation, we can always opt to define a single joined relation in the relation scheme, known as posting the foreign key Example: Relationship -> Relation name addr name Drinkers Likes manf Beers husband 1 2 Favorite Buddies Likes(drinker, beer) wife Married Favorite(drinker, beer) Buddies(name1, name2) Married(husband, wife) 26 Combining Relations • It is OK to combine the relation for an entity E with the relation R for a many-one relationship from E to another entity. • Example: Drinkers(name, addr) and Favorite(drinker, beer) combine to make Drinker1(name, addr, favBeer). 27 Risk with Many-Many Relationships • Combining Drinkers with Likes would be a mistake. It leads to redundancy, as: name Sally Sally addr 123 Maple 123 Maple beer Bud Miller Redundancy 28 • How to represent the following spatial data set in relations? Object-orientation Foundations of object-orientation • The object is at the core of object-orientation • Objects have attributes that model the static, data-oriented aspects of a system (similar to tuples in a relation) o The totality of attribute values constitutes the state of an object • Objects also have operations that model the behavior of a system o Behaviors are also called methods • Objects with similar behaviors are grouped into classes o The set of behaviors for a object form an interface object = state + behavior Example of object-orientation Features of object-orientation • The four main features of object-orientation from a modeling perspective are: o Reduces complexity: decomposes complex phenomena into simpler objects o Combats impedance mismatch: object-orientation can be applied at every level of system development o Promotes reuse: System development is more efficient if constructed from collections of well-understood components o Metaphorical power: Objects in object-orientation are metaphors for physical objects, making the modeling process easier • In addition, four key constructs are closely associated with object-orientation: identity, encapsulation, inheritance, and association Identity and encapsulation • An object has an identity that is independent of its attribute values o Even if an object changes all its attribute values, it retains its identity o Identity is immutable, created with an object and destroyed only when that object is destroyed • Objects hide the internal mechanisms of their behavior from the external access to that behavior, called encapsulation o What behaviors an object exhibits are separated from how those behaviors are achieved o Encapsulation promotes reuse, because changes to an object’s internal mechanisms will not affect the object’s external interface Inheritance and polymorphism • Classes may be organized into an inheritance hierarchy that allows objects to share common properties o A class that provides more specialized behaviors is a subclass o A class that provides more generalized behaviors is a superclass • Inheritance allows objects to perform different roles within specific contexts, termed polymorphism o Inclusion polymorphism is where a subclass is substituted for a superclass o Overloading is where subclasses implement their own specialized versions of general behaviors • There exists two types of inheritance: o Single inheritance: each class may have zero or one superclasses o Multiple inheritance: each class may have zero or more superclasses (requires some protocol for resolving behavioral conflicts) Class diagram superclass behavior (single) inheritance subclass overloading (polymorphism) Association • An association groups objects together to in order to model phenomena with complex internal structure • Aggregation is a type of association concerned with part/whole relationships (e.g. a wheel is “part of” a car) o Aggregation relationships will form a hierarchy often referred to as a partonomy • An association is homogenous if it is formed from objects all of the same class. E.g., a soccer team is a homogenous association (aggregation) • An association is ordered where the ordering of component objects is important. E.g., a polyline might be a linear ordering of points Object-oriented modeling #1 • Object-oriented modeling comprises defining the classes, attributes, behaviors, associations, and inheritance for a system o Attributes for a class can be defined in a similar way to E-R modeling • Behaviors for a class fall into three categories o Constructors are behaviors that are activated when an object is created, while destructors are activated when an object is destroyed o Accessors are behaviors that may be used to examine the state of an object o Transformers are behaviors that change the state of an object Object-oriented modeling #2 • Defining associations and inheritance relationships is an iterative and application-dependent process • As a rule of thumb: o Inheritance relationships can be detected by using the connection “is a” in a sentence with two classes. E.g., ‘a car “is a” vehicle’ o Aggregation relationships can be detected using “part of” in a sentence. E.g., ‘a steering wheel is “part of” a car’ Class diagrams transformer association aggregation constructor accessor attribute Object-oriented DBMS • A DBMS that utilizes an object-oriented data model is called an objectoriented DBMS (OODBMS) • In addition to OO constructs, several other features are needed by OODBMS o o o o Scheme management (ability to create and change class schemes) Automatic query optimization Storage and access management Transaction management • There exists technical problems with achieving these features: o System complexity means that there are no longer a few simple operators, like in relational systems o Encapsulation means that internal state may be hidden from DBMS • As a result, performance for OODBMS is lower that for RDBMS • Hybrid object-relational DBMS (ORDBMS) use a combination of relational data management and object-oriented “shell” for mediating user access to the DBMS Reading • Chap. 2 • http://www.spatial.maine.edu/~max/oomodeling.pdf Hands on • Connecting to Server o o o o o Use openssh client (Start → All Programs → OpenSSH) hostname = geog480.cigi.illinois.edu username = netid port = 22 Enter your netid passwd when prompted • If successful, you just logged in a Linux system (Ubuntu): o Out of your comfortable zone Unix Basics • Folder and directories Unix Basic Commands Directory command: pwd cd Exercise1 mkdir Exercise2 rmdir temp Print the name of the working directory Change the working directory to Exercise1 Make a new directory and call it Exercise2 Delete the (empty) directory temp Basic file command: ls cat File1 mv File1 File3 cp File1 File3 rm File4 less File1 List the files and directories in the working (current) directory Display the contents of the file Change the name of (move) file File1 to File3 Make a copy of File1 and call it File3 Erase (remove) the file File4 Display the contents of File1 a page at a time, q to stop displaying Connecting to Database • psql -U username -d database_name o username = geog480 o database_name = tutorial o Enter passwd when prompted (same as username) • Postgres Commands o o o o \l List all accessible databases \dt List all the tables in current DB \? Help \q Quite Operating Database • Create Table o create table REPLACE_ME_your_netid (key int, attr varchar(20),value float); • Insert a row o insert into your_netid values(1, 'attr0', 100); o insert into your_netid values(2, 'attr1', 101); o insert into your_netid values(3, 'attr1', 102); • List contents of table (Notice that the select statement allows you to view contents in the table and the where clause allows you to filter what the records you what to view) o o o o select * from your_netid; select * from your_netid where attr='attr1'; select * from your_netid where key=2; select key, value from your_netid limit 5; • Update table contents o update your_netid set attr='attr1' where key=1; o update your_netid set value=105 where key=1; • Sorting o select * from your_netid Order by key asc; o select * from your_netid Order by key desc; • Counting o select count(*) from your_netid; o select count(*) from your_netid where attr like '%1'; • Max/Min/Avg o select max(value) from your_netid; select avg(value) from your_netid where attr ilike '%1%'; • Delete Rows o delete from your_netid where key=1; • Copying a CSV file (postgres specific) o \COPY your_netid FROM 'your_file' with CSV HEADER o You may use /srv/cigi/code/test.csv for your_file • Drop Table o drop table your_netid; • End of this topic