Open Source Semantic Web Tools Platform

1
Open SourceSemantic Web Tools / Platform

• Allan Spale
• Kaleidoscope

2
My Background

• Data visualization
• User interface design
• Videoconferencing systems
• Collaboration tools
• Teleimmersion

3
My Inspirations

• Databases are the core that can drive just about
  anything
  
• Scientific American article (1999)?
• The emergence of XML
• Tim Berners-Lee's vision of the semantic
  web
  
• Dr. Dobb's Journal (2000)?
• Mismatch between relational databases and
  artificial intelligence systems

4
One Moment

• Friend had to collect data from people in the
  field
  
• He knew nothing about databases and did not
  particularly care about them
  
• Ended up importing data from a PDA into an Excel
  worksheet
  
• Solving the problem
• How could I abstract the complexities of the
  database?

5
Another Moment

• Working in a university data mining research lab
• Friend recommended Jeff Hawkins' book On
  Intelligence
  
• The brain stores a basic invariant form of
  information
  
• Higher level structures of meaning are built
  above this basic structures
  
• Connections among the structures

6
Current State of Web 2.0

• Mashups (Google Maps)?
• Web services (Amazon Web Services)?
• User-generated content (Flickr, YouTube)?
• Social networking (MySpace, Facebook)?
• Rich internet applications (AIR, Silverlight)?
• RSS
• And more

7
Shortcomings of Web 2.0

• The web is still mostly
  document-driven
  
• End-users cannot easily remix data from different
  sites
  
• Seamless integration among web services and
  devices of any size
  
• Remote, centralized data repositories

8
Solution

• Separate the content data
  from the original application
  
• Protect data and migrate data
  where it is needed
  
• Provide a way for users to acquire, manipulate,
  and share static and live data
  
• Allow any device and any content
  to interoperate within this platform

9
Kaleidoscope

• Database abstraction
• Users store data without having to know
  about relational databases
  
• Collaboration system
• Messaging facility for humans and machines
• Programmable
• Symbolic language, supports Unicode names
• End-user environment
• Embrace widgets, toolbars, and canvases

10
Database Abstraction

• Relational database back-end
• Big as PostgreSQL or as small as SQLite
• Provide a few basic table structures
• Set of items, relations between items
• Encode data at various levels of detail
• Scale from individual words to documents
• React to changes in the data
• User-defined and system-level triggers

11
Collaboration System

• Social networks of people
• Processing networks of machines
• Gather data from people and machines
• Process the data
• Discuss the data
• Stream data to people and machines
• Repeat

12
Programmable

• Sequential or event-driven
• Uses symbolic operators and keywords to allow for
  internationalization
  
• Unicode names for everything
• Access local and remote data stores
• Creates distributed pipelines that enable
  peer-to-peer applications
  

13
End-User Environment

• Lots of widgets that interoperate with each other
  (reusable parts)?
  
• Move away from large-scale applications with huge
  feature sets
  
• Pick and choose the functions needed
• Blur the distinction between internet and desktop
  application
  
• Move toward a toolbox and canvas metaphor

14
System Parts

• Databases
• PostgreSQL (disk), SQLite (memory)?
• Python
• Stackless, SQLAlchemy, PyXMPP
• Jabber / XMPP
• ejabberd
• Utilize server plug-in architecture to integrate
  other components
  
• XML-RPC

15
Integration of System Parts

• Run both database and Jabber server on the target
  device
  
• Database will refuse remote connections
• All database commands will originate from users
  connected to the local Jabber server
  
• Running a Jabber server allows for integration of
  local devices
  
• PDA / smart phone, home automation
• Support assistive / accessibility technologies

16
Integration of System Parts

• Jabber provides contact management and security
• Jabber contact list is imported as individual
  users that are also organized into user groups
  
• Content can be encrypted and checked for
  tampering using checksums
  
• Use of Jabber will overcome dynamic IP
  addressing problems

17
Managing Contacts

• Remote users will register on the Jabber server
  to gain access
  
• Remember, this differs from ones own personal
  contact list.
  
• Registration will trigger a Jabber server
  component that will update the user list
  on the databases

18
Managing Devices

• Jabber provides the ability to add services
• This will allow for external programs and device
  drivers
  
• Through Jabber, one can manage the availability
  of these devices and restrict who can access them

19
Securing Data

• Table Access
• Restricted access using user ownership and group
  ownership
  
• Access can be further restricted by applying
  read, write, and execute permissions for each
  user and group owner
  
• Database Encryption
• PostgreSQL natively supports encryption
• SQLite will require an external encryption
  module
  

20
Importing / Exporting Data

• Text is the most important part of the system
• Word and phrase tables used by all text
  encoders/decoders
  
• External programs will encode and decode files
  and data streams of various formats and map them
  into user tables and system tables
  
• Use multi-level encoders
• Start with mechanical encoders (few
  heuristics)?
  
• Feed that encoding into the next higher-level
  encoder
  
• Repeat

21
Constructing Functions

• Bead programming language
• Direct support for Stackless tasklets instead of
  threads
  
• Acts like a database object relational mapper
• Remote data access and function calls occur using
  XML-RPC
  
• Function code will reside in user tables
• External programs can be connected as Jabber
  server components (better performance) or through
  XML-RPC

22
Building Widgets

• Use existing widget-building models
• AJAX, Flash/AIR, Silverlight, etc.
• Make widgets Kaleidoscope-compatible
• Data resides Kaleidoscope containers
• Mapping of user actions to Bead scripting
  functions

23
Using Environment

• Desktop is just a surface for widgets
• Canvas widgets for displaying data
• Text editing, drawing, 3-D modeling
• Widget tools for interacting with elements on the
  canvas widgets
  
• Font chooser, shape chooser, paintbrush, etc.
• Import widgets and related functions from other
  users or run them remotely

24
Empowered by Kaleidoscope

• Separates data from application
• Data from a program does not become obsolete
• Data can be annotated with various layers of
  meaning
  
• Text document has formatting, sentence structure,
  reviewer notes, etc.
  
• Functions can be remixed from various programs
  without relying on a single
  application
  
• Pick and choose what is needed

25
Empowered by Kaleidoscope

• Widget environment allows the user to operate on
  data with a toolbox
  
• Choose a set of tools to manipulate data
• Data is not dependent upon the tools
• User can choose ones own interface to operate on
  the data

26
Computational Semantic Platform

• Encode data with meaning
• Write programs that utilize both data and
  meaning
  
• Collaboration among
  communities of users
  
• Customizable and portable interfaces

27
Discussion

• What open-source parts can be assembled to deploy
  your semantic web application?
  
• How do you encode your data?
• How do you import / export data?
• How does the user interact with the semantic web
  application?
  
• Think beyond the web and think about the
  separation of data from an application...

28

• Process data your way.