CS246: Web Information Systems

1
CS246 Web Information Systems

• Junghoo John Cho
• Fall 2007

2
Course Information

• Web page http//oak.cs.ucla.edu/cs246/
• Topic Web information management
• Time TuTh 1000 -- 1150 am
• Place 5419 Boelter Hall
• Instructor Junghoo John Cho
• office 3532E Boelter Hall
• email cho_at_cs.ucla.edu
• please use subject CS246
• office hours M 4-5 pm.

3
Todays Topics

• Overview of the course topics
• Central Indexing
• Dynamic Integration
• Data Extraction
• Course logistics
• Paper reading assignments
• Class project

4
Who is this class for?

• Strong interest in research
• Interest in Web information systems
• Time commitment
• Around 2-3 papers every week
• Typically one full day of paper reading
• One indepedent project
• Similar to paper writing
• In fact we read papers from past student
  projects!
• Or interesting application implementation

5
Prerequisite

• Introductory database, e.g., CS143
• e.g. query? SQL?
• Basic algorithms and data structures
• Predicate logic
• ?x, Taller (Fred, x)
• Design and implementation experience
• Basic C
• Quick test Grab a sample paper
• See if you can read, understand and build it

6
Tell Us About You

• Name
• Department Program
• Before coming to UCLA
• Brief history at UCLA
• Technical/research interests
• Expectation from the class

7
Information Galore
Biblio sever
Legacy database
Plain text files
8
Central Problem

• How to manage/access information on the Web?
• Three major approaches
• Central indexing
• E.g., Web search engine
• Dynamic integration
• E.g., comparison shopping services
• Data extraction
• E.g., spamming companies

9
Topic Web Search (Central Indexing)
Central Index
10
Topic Web Search (Central Indexing)

• Web collection of passive HTML pages
• Find Web pages relevant to a query
• Traditional Information Retrieval
• Web collection of HTML pages
• HTML page a bag of words
• More than that?
• Links, structure of the Web
• User access patterns
• HTML tags (markups)

11
Topic Dynamic Integration
Amazon.com
Cars.com
401carfinder.com
Apartments.com
12
Topic Dynamic Integration
Mediator
Wrapper
Wrapper
Wrapper
Source 1
Source 2
Source n
13
Topic Data Extraction
Structured data
WWW
Beatles 10 Madonna 20 NSync 20

• How can we extract structured data from free
  text automatically?

14
High-Level Goal

• Learn core ideas and techniques
• Some of the techniques can be useful for other
  fields
• Learn how to read papers
• Hopefully learn what it is like to do research
• Sometimes very frustrating but often very
  rewarding

15
Main Course Workload

• Paper reading
• Paper reading assignments
• Class discussion
• Independent projects

16
Paper Reading

• Why
• Something that you will do all the time as a
  researcher
• Learn to be critical and communicate well
• Acquire knowledge to conduct research/project
• About 20 papers from
• Conferences SIGMOD, VLDB, WWW, and
• Before the class
• Everyone read and review the paper
• During the class
• Instructor present his own understanding and
  lead class discussion
• Everyone participate!!!

17
How to Get Papers

• From the class homepage
• http//oak.cs.ucla.edu/cs246/
• Some of the materials password protected
• User name cs246
• Password papers
• Let me know if any problem

18
How to Read Papers

• Understand the Big Picture
• What is the problem?
• Why is it important?
• Why is it difficult?
• What has this paper done?
• What others have done?
• How do they support their claim?

19
Paper Reviews (1)

• Due by the preceding Sunday
• Submit through our Web submission interface on
  the class Web page
• Required components at most 5 paragraph
• Summary (1 paragraph) your own words
• This paper discusses how to optimize queries
  with...
• Comments/criticisms (2-4 paragraphs) the good
  the bad
• It addresses a real problem and the solution is
  interesting
• But I feel the experiments are not realistic
  because...
• Optional questions, as many as you want
• Why the authors assume that queries are
  independent?

20
Paper Reviews (2)

• May skip 3 paper summaries without penalty
• Graded by students (Excellent, Good, Poor)
• 10 Excellent
• 80 Good
• 10 Poor

21
Class Project

• Why
• Work on a specific problem and learn to find a
  solution
• 40 of the class
• Team of up to 3
• Topic any problem related to the general problem
• Open style
• Rigorous study of a research problem or
• Any interesting system implementation

22
Class Project Schedule

• Important Milestones
• Group formation 10/7 (Su)
• Project proposal 10/21 (Su)
• Project progress 11/06 (Tu)
• Final report 11/25 (Su)
• Project presentation 11/27, 11/29, 12/4, 12/6
• You are responsible to stay on track
• Make appointments with instructor as needed

23
Project Please Remember

• Put your aims high and be realistic
• Expect to read at least 4-5 papers along the way
• Start early
• Dont do it right before the deadline
• Always unexpected obstacles
• Most students could not finish in previous
  quarters
• Please, please start early
• You are responsible to be on track

24
WebBase dataset

• 70 million Web pages
• 600GB before compression
• C access interface similar to file interface
• repository open(), read(), seek(), tell()
• webcat
• Sunflower.cs.ucla.edu
• cho_at_cs.ucla.edu

25
Grading

• Midterm 40
• Paper reviews 20
• Project 40

26
Announcements

• First review due Sunday 09/30
• Two papers for class 3
• Graph structure in the Web
• Extracting Patterns and Relations from the Web
• Only for this week. All next week papers are due
  in the future
• Please sign up for the paper reviews!
• Everyone has to do it at least once
• Unassigned papers are assigned randomly at the
  beginning of second week