Linked (Data) Scientometrics Keynote

Linked Science Semantic Web Value Proposition Scientometrics
Linked (Data) Scientometrics
Linked Science 2015 Keynote
Krzysztof Janowicz
STKO Lab, University of California, Santa Barbara, USA
Linked Data Scientometrics K. Janowicz

What is Linked Science?
What is Linked Science?
Scientific dissemination traditionally relies heavily on scholarly ar-
ticles and presentations at conferences. However in the past
few years, we have seen an increasing trend towards the publi-
cation of raw research data to facilitate verification and reuse.
Linked Science champions the process of publishing, sharing
and interlinking scientific resources and data along with com-
plete experiment context, which is critical for understanding, reusing
and verifying scientific research. Semantic Web technologies pro-
vide a promising means for achieving this practice.
(From the Linked Science 2015 call)
What are the research questions of Linked Science, what are the bottlenecks?

What are Scientometrics?
The ﬁeld of scientometrics is concerned with measuring and analyzing
the impact of science in its broadest sense.
(Raw) data by example
Publications
Authors
Aﬃliations
Keywords
Themes
Funding sources
Citations
...
What is meant by measuring and analyzing?

Scientometrics Research Questions
Research questions by example
Simple and boring
Number of papers at ISWC 2015
Boring
Number of Papers by a speciﬁc W. Zhang in 2015
Simple and interesting
What goes here?
Interesting
Is the Semantic Web as a research area growing or shrinking?
Are Linked Data and Semantic Web the same community?
Are the research interests of a researcher changing?
What are the new research trends in Artiﬁcial Intelligence?
To which university should I go to study geo-semantics?
Who are good reviewers for a certain paper?

Simple and boring
Boring
Simple and interesting
∅
Interesting

Simple and boring
Boring
Should be Simple and interesting
How does a change in aﬃliations impact a researcher’s interests?
Is there a relation between spatial proximity and citations?
Interesting

Whyare interesting scientometrics questions not simple?

Retrieval
Key Limitations: Data Retrieval
Even the major data hubs such as Data.gov still rely on keyword-based search
and have unreliable, incomplete, and missing metadata. For this type of
retrieval problems, even ‘a little semantics goes a long way’ (Hendler 1997).

Sensemaking
Key Limitations: Sensemaking and Fitness for Purpose
There is no shortage of data, but
finding data that is fit for a certain
purpose is difficult.
Data as statements not as truth,
e.g., according to Springer I am at
WSU not UCSB.
Heterogeneity is caused by cultural
differences, progress in science,
viewpoints, ...; e.g., associate
professor versus senior lecturer
Lack of provenance information
Sensemaking requires more
powerful semantic technologies and
ontologies (compared to IR).

Interoperability
Key Limitations: Meaningful Analysis and Synthesis
Ensuring that data is analyzed and
combined in a meaningful way is far
from trivial.
What if the information on how to
use the data would come together
with these data?
Focus on smart data instead of
(merely on) smart applications.
The purpose of ontologies is not to
agree on the meaning of terms but to
make the data provider’s intended
meaning explicit.

Smart Data
The Smart Data Argument
One of the key arguments underlying the Semantic Web and
Linked Data paradigms is to make data smart, not applications.
Instead of developing increasingly complex software, the
so-called business logic should be moved to the (meta)data.
The rationale is that smart data will make all future applications
more usable, ﬂexible, and robust, while smarter applications
fail to improve data along the same dimensions.
(http://goo.gl/FMXOZT)
Why the Data Train Needs Semantic Rails. (2015) K. Janowicz, F. van Harmelen, J. Hendler, P. Hitzler

Semantics-Enabled Linked-Data-Driven Scientometrics
Howdoes this relate to scientometrics?

Integrates data from a variety of sources, e.g., Semantic Web Dog Food, SWJ.
Example: http://stko-exp.geog.ucsb.edu/lak/

ISWC Installation Based on New Deployment Framework
http://scientometrics.geog.ucsb.edu/iswc/
Smart Data: ﬁrst scientometrics installation (for SWJ) took months to develop and
deploy, now we are down to hours at least when leaving semantic lifting and data
cleaning aside (!) and by using a reduced number of modules (8/30)

Value Proposition
Why do we use Semantic Web and Linked Data for Scientometrics
Federated queries over multiple data sources
Unique global identifiers easy conflation and deduplication
Transparent data model; reduces the need for guessing
No data silos, no API restrictions
Many pre-defined lightweight vocabularies (ontologies)
Smart data reduces the need for smart applications
Machine reasoning support
So do we still need a deeper knowledge representation beyond
surface semantics?

Web@25
Web@25 Installation: Timeline
Keyword frequency for Semantic Web; WWW conference series (1994-2013)
http://stko-exp.geog.ucsb.edu/web25portal/index.html

Web@25
Web@25 Installation: Timeline
Keyword frequency for Linked Data; WWW conference series (1994-2013)
http://stko-exp.geog.ucsb.edu/web25portal/

An Interesting Question
Given the keyword timeline, is the Semantic Web as a research ﬁeld
disappearing, diversifying, radiating, ...?

Letthe data speak for themselves

Web@25
Community detection
Colors: community membership, node size: frequency, line width: co-occurrence strength

Web@25
Web@25 Installation: Self-organizing Map
Landscape analogy: counties, mountains, and valleys

Web@25
Web@25 Installation: Mapping
Location of top institutions that published on Semantic Web between 2009-2013.

Web@25
Similar pattern for Linked Data keyword between 2009-2013.

Web@25
Dissimilar pattern for Search Engine keyword between 2009-2013.

A Tale of Three Papers
Three Papers That Shaped the Semantic Web
Citations peaked 2009 for the Ontology and Semantic Web papers.
More interestingly, why would you still cite these papers today?
http://stko-testing.geog.ucsb.edu/ios/

Top keywords: {Ontology, SW},{Semantic Web, Ontology}, {Linked Data, Semantic Web, Ontology}

If a paper makes impact beyond its own home community, we should see an
increase in keyword variability (entropy).

And Now?
Sois the Semantic Web
disappearing, diversifying, radiating,...?

And Now?
Sois the Semantic Web
disappearing, diversifying, radiating,...?
No amount of data analytics is going to answer such questions before
we precisely deﬁne and communicate what we mean by those terms.

And Now?
Where Do We Go From Here?
Using Linked Data, ontologies, and basic reasoning capabilities, allows us to
rapidly deploy scientometrics installations
Getting basic bibliographic data into (or as) Linked Data is becoming a
trivial task
Conflation, data enrichment, lack of rich metadata remains a major problem.
Discovering owl:sameAs links is just a subtask of conflation
Conflation race between academic publishers, libraries, ...
Generate and enrich the data where it is created or first processed
We need a rich but simple ontology that goes beyond academic publishing
but includes the related processes and roles
Revive Semantic Web Dog Food; ISWC really needs better metadata!
These slides and existing scientometrics systems are about embarrassingly
simple analysis, everything else will needs substantially stronger conceptual
models and machinery (combining inductive & deductive methods)

Linked (Data) Scientometrics Keynote

More Related Content

What's hot

Viewers also liked

Similar to Linked (Data) Scientometrics Keynote

More from kjanowicz

Recently uploaded

Linked (Data) Scientometrics Keynote