OODB Support for WWW Applications: Disclosing the internal structure of Hyperdocuments

1
OODB Support for WWW ApplicationsDisclosing the
internal structure of Hyperdocuments

• Ramprasadh Ramanarayanan

2
Basic Idea

• WWW servers use the OS native file system
• Documents -gt files
• WWW server is a document server
• Use of a OODB as a storage system for WWW

3
Introduction

• WWW -gt Open, distributed system for sharing
  Information
• Links -gtJump from one info to another
• the web is one large hyperdocument of nodes and
  links

4
Access to WWW

• Follow links!

Boon or bane?
5
Associative retrieval

• Searching for keywords not possible.(Depends
  on power of search engines)
• Require Indices

6
Earlier attempts..

• Glimpse Indexing at server level.
• Y!, tradeway galaxy hierarchical menu.
• All use textual contents to produce their index
  databases gt Ignore the internal structure.

7

• Access to information under control of a certain
  server, through a query, should be provided by
  that server, not by search engines or remote
  overviews.

8
Strategy

• Jump from server to server rather than node to
  node.
• Search engines analyze contents and link them to
  keywords.
• Search engines query each server.
• Ability now limited to known servers.

9
Approach

• Traditional web servers treat each document as
  independent -gt Not aware of links.
• Treat entire info base as one hyperdocument.
• The position of the document has no effect.
• Info can be got from specific parts.

10
Approach

• Build on top of OODB.
• Ability to store hypertexts as objects
• Offers a gradual migration from WWW servers to OO
  databases.

11
Design Structure
12
Index databases?

• Space overhead -gt Separate
• Must be explicitly updated.

13
Current search programs

• Determines amount of support
• Complex support -gt Complex queries.

14
Issues..

• Use query facilities of databases
• Storage model is transparent to the user.
• -gt Exact document is to be recovered.
• -gt Same URL usage.
• -gt Questions on title, structure and header.

15
Proposed structure
16
Ode Database system

• Programming language O.
• Uses EOS as storage manager.
• Small objects (lt4KB) in slotted pages.
• Large objects in many pages -gt transition handled
  transparently.

17
Data structure

• Document is an object -gt metadata about size,
  status and a data part.
• Data part has content -gt large object.

18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
Parser for HTML data

• SGMLS parser -gt document to fully comply with
  DTD.
• Identify references on the basis of HREF and
  NAME- attribute.

22
Using CGI scripts

• Preserve original URL -gt transparent use of CGI
  scripts.
• User should see the same view!!
• Modify server code where is it present?

23
WWW server-Ode server coupling
24
Conclusion

• Use of OODB for WWW
• Changes needed are minimal.
• Advantages of OODB incorporated.
• Persistence, authorization and concurrency
  control.
• Problem Is this fast enough? Will people modify
  servers?

25
(No Transcript)