Advanced Databases Introduction dr Toon Calders prof dr

Advanced Databases Introduction dr. Toon Calders prof. dr. Jan Paredaens

Outline • Motivation for the course • Other DH courses • Practical organization • Course topics • Project • Overview of changes

Motivation for the Course • Database = a piece of software to handle data: − store, − maintain, and − query • Most ideal system situation-dependent • • data type: simple / semi-structured / complex / … types of queries: simple lookup / analytical / … type of usage: multi-user / single-user / distributed / … …

Motivation for the Course • Relational databases are tuned towards: • simple data • simple, ad-hoc queries • multiple users • Other models are more suitable for other types of data • • Object-Oriented, Deductive, Semi-Structured Databases, Data warehouses

Motivation for the Course • Study different data models • Advantages, disadvantages • Conceptual level − what are the important notions? • What’s underneath? • In a scientific way • exact, not just claims

Motivation for the Course • Student knows: • different database models • Understands: • why they are introduced • conceptual notions • Is able to: • quickly master vendor-specific products

Outline • Motivation for the course • Other DH courses • Practical organization • Course topics • Project • Overview of changes

Other DH Courses • Relational database systems (2 ID 05) Databases and Data Modelling (2 ID 35) Database Technology transations, indexing, query optimization, distributed DB • Other database models (2 ID 45) Advanced Databases • (2 II 15) Data Mining • (2 ID 25) Information Retrieval • (2 ID 99) Capita Selecta DH

Outline • Motivation for the course • Other DH courses • Practical organization • Course topics • Project • Overview of changes

Practical Organization In principle … • Wed 8: 45 10: 30 Practical session M 1. 46 − no new material − opportunity to practice, ask questions − together solve exercises • Fri 10: 45 12: 30 Lectures − XML : Paredaens (6 lectures) − other parts: Calders HG 6. 09

Practical Organization • Important information http: //wwwis. win. tue. nl/~tcalders/teaching/advanced. DB/ • Subscribe to 2 ID 45 on studyweb ! • messages to the whole class group − lecture postponed, room changes, … • t. calders@tue. nl

Practical Organization • Course material • Book: Silberschatz, Korth, Sudarshan. Database system concepts 5 th edition. Mc. Graw-Hill International • Lots of additional material on course webpage − papers − slides − solutions to exercises −…

Practical Organization • Grades: • 70% written exam • 30% group project • No project = no grade • Grade for the project can be transfered to August, similar for grade for the exam • Grades expire in August

Outline • Motivation for the course • Other DH courses • Practical organization • Course Topics • Project • Overview of changes

Course Topics • Limitations of the relational model • Deductive databases • Object-Oriented Databases • Data Warehousing & OLAP • Semi-Structured data

Limitations of the relational model • Not every query can be expressed • Transitive closure cannot be expressed in Relational Algebra − Give all cities reachable from Antwerp by plane − Give all smallest components of a part − Give all decendants of person X • Not even if you’re very smart … − proof • Extension to other relational query languages

Deductive Databases • Motivation is two-fold: • add deductive capabilities to databases; the database contains: − facts (intensional relations) − rules to generate derived facts (extensional relations) Database is knowledge base • Extend the querying − datalog allows for recursion

Deductive Databases • Datalog as engine of deductive databases • similarities with Prolog • has facts and rules • rules define -possibly recursive- views • Semantics not always clear • safety • negation • recursion

Deductive Databases g(a, b). g(b, c). g(a, d). reach(X, X) : - g(X, Y). reach(X, Y) : - g(X, Y). reach(X, Z) : - reach(X, Y), reach(Y, Z). node(X) : - g(X, Y). node(Y) : - g(X, Y). unreach(X, Y) : - node(X), node(Y), not reach(X, Y).

Deductive Databases • In this topic we study: • How to handle negation and recursion in the same program • How to efficiently evaluate Datalog queries

OO Databases • Many applications require the storage and manipulation of complex data • design databases • geometric databases • … • Object-Oriented programming languages manipulate complex objects • classes, methods, inheritance, polymorphism

OO Databases • Very simple example: • Class book − set of authors − title − set of keywords Extremely simple to model in OO language Hard in relational database!

OO Databases • In many applications persistency of the data is nevertheless required • protection against system failure • consistency of the data • Mapping: object in OO language tuples of atomic values in relational database is often problematic

OO Databases • Either we ignore the multivalued dependencies Title Author Keyword Database System Concepts Silberschatz Database System Concepts Korth Database System Concepts Sudarshan Database System Concepts Silberschatz Storage Database System Concepts Korth Storage Database System Concepts Sudarshan Storage • This table is in 3 NF, BCNF

OO Databases • Or we go to 4 NF Title Author Database System Concepts Silberschatz Database System Concepts Korth Database System Concepts Sudarshan Title Keyword Database System Concepts Storage

OO Databases • Basically OODB = persistent OO programming language • Very important concept • rather uninteresting scientifically • This topic will mainly be self-study • Reading bookchapter + Q & A session

Data Warehousing & OLAP other Metadata sources Operational DBs Extract Transform Load Refresh Monitor & Integrator OLAP Server Analysis Query/Reporting Data Warehouse Serve Data Mining Data Marts Data Sources Data Storage ROLAP Server OLAP Engine Front-End Tools

Data Warehousing & OLAP Transaction processing Flight reservations • Operational setting • Up-to-date = critical • ticket sales • do not sell a seat twice • Simple data • reservation, date, name • Simple queries; only « touch » a • Give flight details of X small part of the database List flights to Y

Data Warehousing & OLAP Decision support • Off-line setting • « Historical » data • Summarized data • Integrate different databases • Statistical queries Flight company • Evaluate ROI flights • Flights of last year • # passengers per carrier for destination X • Passengers, fuel costs, maintenance info • Average % of seats sold/month/destination

Data Warehousing & OLAP • In this topic we will study: • Conceptual models for decision support • Database explosion problem • Efficient implementation strategies − indexing, view materialization

XML • Why is XML important? • simple open non-proprietary widely accepted data exchange format • XML is like HTML but • no fixed set of tags − X = “extensible” • no fixed semantics (c. q. representation) of tags − representation determined by separate ‘stylesheet’ − semantics determined by application • no fixed structure − user-defined schemas

XML <Person. List Type="Student" Date="2004 -12 -12"> <Title Value="Student List"/> <Contents> <Person> <Name>Jan Vijs</Name> <Id>11</Id> <Address> <Number>123</Number> <Street>Turnstreet</Street> </Address> </Person> <Id>66</Id> <Address> <Street>Hole Rd</Street> </Address> </Person> </Contents> </Person. List>

XML • In this topic: • XML • XQuery, XSLT • Li. XQuery • Taught by prof Paredaens

Outline • Motivation for the course • Other DH courses • Practical organization • Course Topics • Project • Overview of changes

Project • Pick one of the 4 topics: • • deductive databases / rule-based systems object-oriented databases data warehouses semi-structured databases • Formulate your own project • illustrating the different course concepts • showing you mastered the technology

Project • Make a project proposal ( WEEK 10 ) • examples of last year will be given • fulfilling certain constraints • listing technologies to be used • Status report ( WEEK 15 ) • Final report ( WEEK 20 ) • Project presentations ( WEEKS 21 & 22 )

Outline • Motivation for the course • Other DH courses • Practical organization • Course Topics • Project • Overview of changes

Overview of Changes • First some facts and figures regarding Spring 2008 • Heterogeneous group − Outside NL, HBO, BSc TU/e CSE BIS

Overview of Changes • Some suggestions I decided to act upon: 1. Start with the difficult material: − expressiveness of RA − Gaifman locality 2. Too much time is being spent on XML − (5+5) (6+3) & topic (XSLT) has been added 3. Disproportional weight given to XML in exam − project no longer exclusively XML

Overview of Changes • Some suggestions I decided to act upon: 4. Some materials and instruction just too hard − extra exercices will be added; more modular 5. The course was split up in lots of individual subjects, with no apparent relation to one another − tried to handle that in the course motivation

Overview of Changes • Some suggestions that were ignored: A google for 'advanced databases' returns quite some courses from other universities that look interesting to me. Perhaps the lecturers could take a look at those. − When (re-)constructing the course last year other universities’ ADB courses were surveyed. Many of the interesting topics are already handled in other courses (Data Mining, Information retrieval, Database technology)

Overview of Changes • Some suggestions that were ignored: Don't discuss prerequisite knowledge too much, it is prerequisite. Heterogeneous group. Balance the course subjects more, TC was discussed very specific while the other 3 subjects where treated in global. Time spent on TC is justified by its difficulty and its importance for database theory + motivates OODB & Deductive DB

Overview of Changes • Take-away message • (some? ) lecturers do act on questionnaires • filling out the questionnaires is useful

Summary • Relational model has limitations • simple queries • simple data • OODBs allow complex data types • Deductive databases, datalog complex queries • Somewhere in-between: datawarehouses and OLAP • special requirements, special datastructures • Semi-structured data can be stored in XML • Project complements theoretical lectures • Instructions for clarification

!! See you on Friday !!