Annotation and Browsing Annotation and Browsing on the Semantic Web on the Semantic Web

blaredsnottyAI and Robotics

Nov 15, 2013 (3 years and 8 months ago)

136 views

Annotation and Browsing
Annotation and Browsing
on the Semantic Web
on the Semantic Web
(experiences from KnowledgeWeb and Magpie use cases)
(experiences from KnowledgeWeb and Magpie use cases)
Martin Dzbor
Knowledge Media Institute
The Open University, UK
http://kmi.open.ac.uk

2

Feeding the Semantic Web

acquiring annotations and metadata

from hyperlinks & correlations to semantic relationships

Navigating the Semantic Web

working with and using annotations

familiarity of interacting with Semantic Web tools

bootstrapping semantic tools with the Web tools

Lessons & experiences

what we learned about annotating & navigating the Semantic
Web in the KnowledgeWeb project
Overview
Overview
3

A bit of a cliché:

people are “overwhelmed by the amount of information”

In fact, size is not the major issue here

most intelligent behaviour relies on capability to see & make
connections, links & associations
[Vannevar Bush, 1946]

this includes sciences, engineering, learning…

whereas current Web is largely about ‘singular’ entities with
fairly simple and limited relations
Issues…
Issues…
Publications
Sources
Centres
Projects
co-occur
Languages
Atomic Concepts
Authors
Technologies
OWL <extends> XML
OWL <is_a> Markup_Lang

Class’ represents a
collection of entities
co-occur
co-occur
publish
co-occur
Research
Issues
Semantic Web view
Semantic Web view
xyz:Author
foaf:Person
abc:Institute
abc:University
skos:Document
dolce:Activity


prj:Task

skos:Language
xyz:Relation
dolce:Activity


discussed_in
situated_in
expert_in
implemented_in
has_src
relates_to
relates_to
criticizes
coauthor
investigates
has_key
active_in
active_in
active_in
researched_by
5

User tasks are typically more than queries

few activities/tasks can be directly & uniquely translated to
(formal) queries

multiple queries need to be connected & data from them
interpreted

Interpretations are often imprecise

queries are hard to (re)formulate & expand by the users

Where Semantic Web can help

embed initial queries into potential exploratory paths

don’t just respond to the queries

give alternatives, suggest what next/else can be done
and why/why not

lead the user through a problem space
Beyond queries
Beyond queries
Semantic Web annotation
Semantic Web annotation
What & why?
What & why?
Part I of the talk
7
Annotation
s in portals
Annotations in portals

Annotation is essentially an attachment of some
descriptive text or data to resources

Mostly working with text documents

Mostly focusing on more clever information retrieval

Example applications

Large scale: Yahoo! directory or deli.cio.us / Flickr tagging

Medium scale: specialized repositories like REASE

Small scale: iTunes
8
REASE in detail
REASE
in detail
Annotations
help with data
filtering
Annotations
help with
browsing &
navigation
Annotations
help with
result ranking
http://rease.semanticweb.org
9
Annotation categories
Annotation cat
egories

What would you like to have in a portal about SW?

Could you fill in the short feedback form to enable us to
extend and improve REASE coverage?

Online version

http://www.l3s.de/kweb/REASE-industry
10
Acceptance?
Acceptance
?

Feedback on semi-automatically annotated portals

‘…
REASE is more targeted and finer-grained than large
generic libraries…’

‘…
REASE tells me something about the domain, doesn’t only
list 1000s of documents…’
0
20
40
60
80
100
120
Jul-04
Sep-04
Nov-04
Jan-05
Mar-05
May-05
Jul-05
Sep-05
Nov-05
Jan-06
Mar-06
May-06
Jul-06
Sep-06
Nov-06
#Learning units
#LUs for Industry
0
50
100
150
200
250
300
350
400
Jun-04
Sep-04
Dec-04
Mar-05
Jun-05
Sep-05
Dec-05
Mar-06
Jun-06
Sep-06
Dec-06
# users
# Institutions
11

Annotations used in problem exploration

semantic, syntactic & semiotic proximity

What else is useful to know when analyzing ‘query languages’?

Can I explore emerging correlation between ‘rules’ and ‘logic’?

filtering, summarization

Show me only educational materials from a specific viewpoint
(e.g. OWL ‘and’ constraints ‘but not’ toy examples)

Summarize retrieved materials in terms of communities of
practice that are knowledgeable about a given topic

User may not want the single best answer

guidance vs. prescription

Support for
homing in
and
selection
of the appropriate
information

rather than immediate retrieval
Why
bother?
Wh
y bother?
12
Paths & g
uides in
Paths & guides
in
the Magpie f
ramework
the Magpie framework

A simple example of
applying
a semantic relation

getting beyond
Y: Y
coauthor_of
X
P:
Y: (Y
coauthor_of
X) and (X
author_of
P) and…

Case of Magpie & CORDER
(both from KMi)
13
Acquiring annotation
s
Acquiring annotations

Semantics based on lexicons and proximity relations

web mining = high scalability

selectable corpora = constrained mining (or deliberate bias)

lexicon = interactive & customizable serialization of ontology
Semantic annotations in Magpie
From correlations to annotations
Magpie + third-party services
14

Annotating the manual or the ‘CORDER’ way?

acquiring XML/RDF annotation datasets from the Web

working in standard web browsers, with standard web pages

adding hyperlinks in a role ‘relationship surrogates’

Let’s generalize this question:
How novel/radical shall the user interface
for the Semantic Web application be?

Three different take-ons

familiarity of use interaction

legacy data & knowledge bootstrapping

users’ expectations & rewards
Use
r’s position?
User’s position?
15


but it’s so tedious
but it’s so tedious

In fact, novelty has drawbacks

hard to sustain over longer period

creates resistance to the proposed change

Technology (Semantic Web) is not the sole new thing
the user has to cope with!

many tools assume new user roles
16

Semantic Web offers qualitatively new way for
information processing

hence, we need new tools & user interaction methods to
automatically ‘feed’ the Semantic Web

only novelty assures sufficient visibility of the new
technology, method,…
Automating = scaling
Automating = scalin
g
Martin Dzbor
KMi, NeOn, ESTC
Richard Benjamins, York Sure
Interoperability, problem solving
Web documents
ESpotter
Entity
data
CORDER
Associative
relations
Bias corpus
17
Novelty vs. familiarity
Novelty vs. familiarity

New technologies are ‘cool’ in short term,but…

they’re hard to sustain over longer period

may create resistance to the proposed changes

Technology (Semantic Web) is not the sole new thing
the user has to cope with!

many tools assume new user roles

many tools assume new interaction modalities

Try a different view…

instead of pushing new technology, we need to improve the
overall use experience


how can Semantic Web make task I’m doing different,
easier, faster, simpler,…’
Semantic Web browsing and navigation
Semantic Web browsing and navigation
How and with what?
How and with what?
Part II of the talk
19
Reusing & interpreting
Reusing & inter
preting

We can use existing annotation datasets

e.g. by interpreting results to certain types of queries
20
Annotation as data re
use
Annotation as data reuse

Reusing (often non-semantic) data to produce
semantic annotations

previous examples reuse non-semantic data (SQL DB)

while data content in DB is not semantic, certain
composite
queries
have a well-defined semantic interpretation

hence, such queries act as if they were feeding semantic
annotations on the singular Web resources

More importantly, this approach to semantic annotation
takes advantage of the Web

information is captured redundantly

reliance on the principle of authority/provenance

law of big numbers & statistical correlations

But would it scale?

cf. the CORDER example
21
Goin
g the Web scale
Going the Web scale

Large repositories can be mined for annotations

Automatic acquisition of metadata is a low-hanging fruit
18k mined resources
automatically annotated
with a given term
22
Annotations and the use
r
Annotations and the user
Tabulator (RDF browser)
Faceted mSpace (RDF browser)
PiggyBank (RDF scraper)
23
Semantic Web navigation
Semantic Web
navigation

Who is the user I’m targeting, and in which role?

e.g. ordinary users of web browsers, tagging tools,…

What underlying technology I’m championing & why?

e.g. XML/RDF annotated phrases, documents, resources,…

How do I want the users to interact with technology, what tools
am I offering?

e.g. browse and navigate using semantic annotation
24
The ‘big’ question
The ‘big’ question

How radical is the introduction of a Semantic Web
technology (annotation)?

for developers
moderately
; they are used to structured
languages, but not on this level of granularity

for web surfers
substantially
if we only offer them raw
semantic structures

Using technology
seeing technology

people are resistant against unknown features, roles,…

unknown #1: RDF is unfamiliar to many end users

hard to imagine what exactly it describes
<?xml version="1.0"?>
<popup id="kmiPopup" titlebar = "false"

xmlns="http://www.mozilla.org/keymaster/gatekeeper/there.is.only.xul">
</popup>

Highlights of Magpie technology

dynamic integration of distributed
computational resources with
information found on the web

generic framework for developing
Semantic Web applications

framework applied to a range of
domains: climate science outreach, UN
public service, e-learning,…

Magpie in practice

annotations equal to user
choosing a lexicon, a focus

semantic browsing intertwined
with ‘classic’ browsing
Semantic proximity link
Web link

Association service

based on semantic proximity

populated by web mining

shows non-authoritative
semantic
relationships

Relationship types

topic to topic similarity

person’s familiarity with topic

person A is an alternative
source to person B

community of practice in a
specific topic

All hidden

behind menus or links
28
Navigation & legacy
Navigation &
legacy

The best way to prove a new technology

is next to the existing one…

Intertwining legacy & new technologies

on the level of user interfaces

e.g. Magpie, KIM, VieWs, PiggyBank – all ‘in’ web browsers

on the level of interaction protocols

e.g. SOAP over HTTP, XML + XSL = ‘HTML’

on the level of languages

e.g. OWL importing UML modelling
Conclusions
Conclusions
Part III of the talk
30
Experience reward
Experience re
ward

Broad visions need to be instantiated into specific
(small) benefits

this is even more important if we are introducing change in a
user’s role

document reader

document annotator (
author)

can we show off the annotations to anybody?

and can we do it instantly?

Semantic Web needs to provide instant rewards


Low hanging fruit’ approach

applying a novel tool in one specific task

e.g. discovery, collection and association of Web data for the
purpose of analyzing domains

offering one specific benefit the user values

e.g. help with homing in on a smaller subset of analytic
references or partial results using fewer clicks
31
Final points
Final points

Semantic Web applications come in many different
shapes and sizes

although RDF is important for Semantic Web, one doesn’t
have to edit it or see it to carry out semantic annotations

annotations are useful if effort to acquire them is reasonable

greater benefit when applied (e.g. in reasoning or browsing)
rather than merely listed (e.g. as RDF triples)

Hiding complexity is not a sign of weakness

the point of Semantic Web is to make the use experience
better, not more complex


comprehension by machines’ is good; however, at the end of
any operation on the Semantic Web is a human user

Reuse & openness rule

more reusable components
(annotations, services,…)

open up your captured knowledge to others
Thank you; Q&A
Thank you; Q&A
(also your experiences, suggestions,…)
Tools/services shown can be found on:
* http://rease.semanticweb.org
* http://neon-project.org/aspl-v2
* http://kmi.open.ac.uk/projects/magpie
http://simile.mit.edu/piggy-bank
http://www.mspace.fm
http://www.w3.org/2005/ajar/tab