Information Extraction in Semantic Web

1
Information Extraction in Semantic Web

• Amreek Singh
• (02329025)

2
Motivation

• The www is growing at an exponential rate.
• User will be lost in jungle of information.
• Need for proper classifier.
• Currently we depend on humans for classification
  and maintenance.

3
Motivation (Cont.)

• Yahoo spends millions of dollars to classify web
  sites.
• Html is designed for displaying data but is not
  suitable for extracting useful information.
• This manual way of classification of www is not
  scalable.

4
The semantic web approach

• So we need machine to understand these pages and
  be able
• Classify
• Extract information

5
Extracting Information

• Information extraction can be done by any agent
  in between client and server (hosting source
  pages)
• These agents can be viewed as Semantic Agent

6
Challenges to Developing Semantic Agent

• So we need an agent that crawls the web and
  classify the web page to extract information on
  its own.
• Even with natural language understanding it is
  very difficult, if not impossible to make the web
  agent read and understand the web.

7
One step towards next Internet

• HTML XML
• HTML XHTML XML
• Imagine best parts of html and mixing them all
  with great features of xml.

8
What Is XHTML?

• XHTML is Extensible HTML standard developed to
  help efficient extraction of information from
  web-pages
• Link http//www.xhtml.org/

9
Issues in XHTML

• Documents must be well formed i.e. must have
  closing tags and nested properly.
• Elements and attributes must also be in lower
  case.
• Attributes must be quoted.
• For empty tags either tags can be used.
• ltbrgtlt/brgt or ltbr /gt
• Inline style sheets and java scripts must be in
  separate files.

10
Benefits of Xhtml

• Xhtml documents are well formed so leads to
  quicker and smaller parsers.
• A significant limitation of the HTML is its FORM
  field.
• Pre built-in functions remove the need to use
  javascript as heavily as in the past.
• Allowing to add voice or other input methods.
• Data is transmitted in xml format.

11
Benefits of Xhtml (cont.)

• Xhtml is now standard.
• Xhtml documents are backward compatible.
• Xhtml can utilize application (scripts and
  applets) that rely either upon html DOM or xml
  DOM.
• The next internet world is of XML, and Xhtml is
  the step towards.

12
Real Power of Xhtml

• Xhtml can be directly transformed into another
  xml structure. But html can not.
• Xpath gives you a way to perform complex queries
  on the nodes and extract data at any level of
  complexity.
• Traditionally index servers use a brute force
  method to index a site, recording the position of
  given words in their respective files.

13
(Cont.)

• User agents that access Xhtml documents served as
  text/html can use HTML DOM.
• User agents that access Xhtml documents served as
  text/xml via DOM can also use XML DOM.

14
Two approaches to embed Xhtml in HTML pages

• Authors manually annotate their web pages w.r.t.
  one or more ontologies provided.
• Automatic annotation of web pages with machine
  learning techniques.

15
Issues with first approach

• First method is of no use, if authors do not
  annotate their web pages.
• The information provided is totally dependent on
  the ontology used by author.
• While updating author may only update html
  document and forgets to update the annotations.

16
Issues with the second approach

• Machine learning techniques have proven to be of
  great practical value in many applications like
  data mining, speech recognition etc.
• However, it is very difficult to guess what the
  other person is thinking.

17
Semantic Web

• Web in which information has well defined meaning
  and enables computers and people to work in
  cooperation.
• Makes machines better understand the data rather
  than just displaying it.

18
Approaches adapted by Semantic web

• How to add logic to web.
• Define language that expresses both the data and
  the rules for reasoning about the data.

19
Simulation of Semantic Web

• Given a set of html web sites that are annotated
  with special tags.
• First program understands the meaning of the
  Search String.
• Extracts the domain of web pages that is given.

20

• Program extracts the required information from
  the web pages, in a way that it extracts the
  related information also.
• Eg. If search is given for address,
• Program will search for address as well as
  location, home, house, etc.

21
References

• http//www.javascriptkit.com/howto/xhtml_intro2.sh
  tml
• http//www.xhtml.org/
• http//www.w3c.org/
• http//research.nii.ac.jp/collier/papers/LREC-200
  2.3b.pdf
• http//infomesh.net/2002/augmeta/sec-xhtml
• http//216.239.39.100/search?qcacheizh-0ZZoha8C
  research.nii.ac.jp/collier/papers/
• Programming Perl, OReilly

22

• Thank You