Extracting%20Semistructured%20Information%20from%20the%20Web

1
Extracting Semistructured Information from the
Web J. Hammer, H. Garcia-Molina, J. Cho, R.
Aranha, A. Crespo from Stanford University
Presented by Wei Mao
2

• Introduction
• Background
• Fast growing of WWW
• Semistructured data in web pages
• Difficulty with manipulating web data
• One solution
• A configurable extraction program
• Extraction result in OEM
• A wrapper is used for query

3

• A detailed example
• Weather table

Can we query What is the forecast for Vienna for
Jan. 28, 1997??
4

• Extraction process
• HTML file
• Specification file
• Commands
• variables, source, pattern
• Package result into an OEM object

5
The HTML for weather table
6
A sample specification file
7
Extraction result
8
Customizing the extraction result
9
Additional capabilities

• Extract_table construct
• Case operator
• Get(url) operator

Query the extracted result

• Use existing wrapper generation tool
• Only simple interface is required

10
Advantages

• Manipulate web data efficiently
• Flexible
• Easy to use
• Reuse the existing systems
• (OEM, Lorel, HTML parser)

11
Disadvantages

• Depends on outside input
• Requires prior knowledge of the
• structure of HTML file
• Have to use specification file