Transitioning web application frameworks towards the Semantic Web (master thesis progress report)

Chapter Copyright 2007 Digital Enterprise Research Institute. All rights reserved. www.deri.org

Digital Enterprise Research Institute www.deri.ie

Transitioning web application frameworks towards the Semantic Web

(Master thesis progress report)

Benjamin Heitmann

1


Benjamin Heitmannof 42

Overview

Part 1: Introduction About me Example use case and Problem Statement Methodology

Part 2: Contributions Survey of 54 Semantic Web applications and papers Analysis of requirements based on application types In Progress: Evaluation of current framework approaches

Part 3: Outlook

2



Introduction

3



About me

Research topic: software engineering for the Semantic Web

Helped with ActiveRDF Wrote the SIOC explorer (based on ActiveRDF) Started packaging components for reuse Participation:

Paper about using ActiveRDF to support building a complete Semantic Web application

4


Example use case: tracking a music community

?

5



Solution: Reusing Community Content

web siteexporters

linked data:

microformats (contact, location, event)

after aggregation:

6



Problem statement

How can we build such an application? manual implementation is required Problems:

– high learning curve– no reuse of existing infrastructure and experience

Question: Why is there no framework for building a complete Semantic Web application? Answer: unknown requirements What common application types? What components do they share?



Methodology

8

RequirementsApplication types and

components

Literature review

prescriptive, analyse vision and potential

Surveydescriptive,based on existing

implementations

Evaluationof current

approaches

General idea: extract framework from real applications



Preliminary results of Semantic Web Application Survey

9



Working Definition: Semantic Web Application

10

1. Every application allows the user to perform a specific task

2. This leads to the requirements: User: requires an interface to perform: requires application logic the task: is represented by data

3. To solve the task by utilising the Semantic Web the support of Semantic Web standards is required e.g. RDF, RDF Schema, SPARQL or OWL



The survey

11

FlinkMuseumFinland

DbinNASA

Semantic Organizer

Annotea

Bibster MusiDB Swoogle Foafing the music

Semantic Media Wiki

Paper Puppy

RDF Homepage

FOAF Map Confoto

Fungal Web

Personal Publication

ReaderAnno Terra CS Aktive

SpaceSeco DOPE

http://activerdf.org/survey/54 applications, 34 properties

http://activerdf.org/survey/

http://activerdf.org/survey/



Sources of the survey

Semantic Web Challenge (at ISWC) 90 % of entries fit definition 41 entries in total results from 2003 to 2006 available

Semantic Scripting Challenge (at ESWC) all entries fit definition 13 entries without overlap from Sem. Web Challenge results from 2005 to 2007 available

Both challenges required: description of implementation and capabilities published as paper

12



Excluded source

Sweet Tools List contains 542 “tools” privately maintained by Michael K. Bergman

Reasons for exclusion time constraints most entries have no academic publication more then 50 % of them do not fit application definition all of them are general purpose tools majority can only be used to build parts of an application

13



Acquiring the data

source for data: papers, code mostly unavailable iterative process:

look at applications create new properties

many projects did not specify implementation at all incomplete descriptions or properties from

literature review lead to “unspecified” properties Disclaimer:

Validation of data by authors of the papers in the future

14



Application Types

15

55

5

6

7

31

semantic portalsemantic annotationsemantic repositorynot an applicationsemantic authoringsemantic desktop application

majority: semantic portal

only one third support editing



Semantic Viewer/Browser

Example: FOAF Map Reads FOAF Files puts locations of People on Google Maps

16



Semantic Portal

Example: Museum Finnland

Integrates data about artefacts from multiple museums

allows faceted browsing

recommends similar items

17



Semantic Annotation

Example: Confoto

Add meta data about people and events to photos

18



Semantic Repository

Example: NASA Semantic Organizer Add documents and binary files Annotate them with semantic metadata

19



Semantic Authoring

Example: Semantic MediaWiki

Add semi-structured information

Add schema View

information

20



Semantic Desktop Application

Example: DBin enables

sharing of data about specific topics

provides tools to build a community around the topic

21



Semantic Web Standards and Schemas

22

DAMLRDFS

SPARQLother query langs

OWLRDF

9 18 27 36 45

44

24

8

7

6

3

RDF support: 80% OWL support:

50% SPARQL support

as high as previous query languages

Many projects use custom schema

Even Standard schemas used mostly once (“others”)

privateunspecified

FOAFRSS 1.0

Dublin Coreothers

9 18 27 36 45

34

4

7

9

12

14



Implementation

23

3

16 19

25% of all projects did not specify implementation

103

42 112

23 5

5

ScriptingJavaProlog

RubyPythonPerlPHP

SesameRAP (PHP)JenaActiveRDF (Ruby)RDFLib (Python)RedfootRedland

Languages Scripting Languages RDF Libraries



Architecture Types

24

63

6

832

centralised serverdecentralised server networkpeer to peerstand alone appother

majority: centralised second biggest:

decentralised and peer to peer architecture



Application Domains

25

ArtsCollaboration

SearchSocial Network Analysis

HistoryLife Sciences

Geospatial InformationResearch

PublishingInfrastructure

Knowledge Management

4 8 12 16 20

17

9

9

7

7

3

3

2

2

2

2



Degree of openness (1)

26

Editing

Can use External Data

Data Updates

Schema Updates

Alignment/Integration

yes no unspecified offline

majority: no

majority: yes

majority: yes

majority: no

59

12 13

15

semi-automaticmanualnot neededunspecifiedautomatic



Degree of openness (2)

27

Unstructured text supported

Semi-structured data supported

inferencing of new knowledge

data export / reuse of app as new data source

yes no unspecified

no real priority

equal distribution

no real priority

majority: yes, good for Linking Open Data



Unsupported at the moment

mash-ups / visualisation of data multimedia content collaboration multilingual content tracking authorship and provenance trust semantic transport layer

(unsupported by definitive majority)

28



Requirements

29



Application types

30

Semantic Annotationadd only meta data

Semantic Portalintegrate and view multiple sources

Semantic Viewerview one data source

Semantic Repositoryadd binary and structured data

Semantic Authoringadd semi-structured data

and schemas

Semantic Desktop

Application

an application can have multiple types at the same time



Semantic Viewer

Example: FOAF Map Capabilities:

very simple: just view and navigate

one source with one schema

only read access, no editing

Requirements: RDF data support Navigation and

Visualisation of the data domain

31

Components: Data interface: to local

or remote store User Interface:

provides navigation and visualisation

Application logic: binds data to interface

Search



Semantic Portal

Example: Museum Finnland Capabilities:

Uses multiple different data sources

Requirements: Integration of multiple data

sources Alignment of multiple

schemas Object Consolidation support for Semantic Web

data and schema formats

32

Components: Crawler: aggregate

data from different (remote) sources

Integration Service: integrate data and schema

Persistence layer: materialise the integration results



Semantic Annotation

Example: Confoto Capabilities:

annotate resources with meta data but do not add new ones

Requirements: Write support in User

Interface Write support in persistence

33

Components: Annotation Interface:

guides user in resource annotation

Persistence layer: storage for user annotation



Semantic Repository

Example: NASA Semantic Organizer

Capabilities: add new binary resources add new structured data annotate existing data

Requirements: Integration of user provided

data:– binary data– structured data

34

Components: Repository Interface:

allows adding of resources

Content or document management service: provides versioning and permissions

Binary repository: stores binary data



Components: Authoring interface:

guides user in creating semi-structured data

Semi-structured persistence: no enforcing of schema, runtime updates possible

Semantic Authoring

Example: Semantic MediaWiki

Capabilities: add and edit semi-structured

data add and edit schemas

Requirements: supports semi-structured

data runtime evolution of data and

schema

35



Components: Rich Interface: locally

executed, low latency Desktop Integration:

use and integrate local data and schema

Semantic Desktop Application

Example: DBin Capabilities:

Rich user interface Desktop Integration

Requirements: supports Semantic Web data local execution of application

36



In progress: evaluation

37


Benjamin Heitmann and Eyal Orenof 18

First Evaluation

38

ActiveRDF together with Ruby on Rails

ActiveRDF fits transparently into the Ruby on Rails architecture

View

Controller

Model

ActiveResource

ActiveRDF

ActiveRDF bridges the gaps between data models Allows reusing the Rails eco-system:

libraries and components



Supported Application Types

Semantic Viewer: low effort Data Source: ActiveRDF provides remote or local RDF

source Visualisation: Rails supports AJAX Application logic: object oriented access to RDF triples

Semantic Portal: medium effort Crawler: manual implementation, reuse Ruby libraries Integration Service: manual implementation Persistence Layer: ActiveRDF supports read-write access

39



Unsupported Application Types

High effort for implementation: all other application types Semantic Annotation:

– Annotation Interface must be bound to RDF data

Semantic Repository: – Management of documents and structured data has to be

integrated into semantic part of application

Semantic Authoring: – semi-structured capabilities of user interface must be

implemented– online updates of schema and data have to be visible in user

interface

Semantic Desktop Application:– possible by using Joyent Slingshot: Rails Desktop Integration

40



Outlook

41



ToDo for Master Thesis

42

Validate results of survey: ask paper authors for corrections

Evaluate more approaches to support semantic web application development possible candidates: about 10 depends on availability of papers or source code

Writing: Write down the results :) maybe publish something Should be finished at end of September



Summary

Question: How can we support the development of Semantic Web applications?

Answer: look at existing Sem Web App publications which describe:

– capabilities– implementation

derive requirements from those apps

Results: 6 types of semantic applications Requirements and components for each type Allows evaluation of current approaches Recommendations for future Sem Web App frameworks

43

Transitioning web application frameworks towards the Semantic Web (master thesis progress report)

Technology

semantic web applications

application logic

fit application definition

web challenge results

iesemantic web challenge

common application types

entries fit definition

ielinked data