VLDB 2012 Summary

wonderfuldistinctAI and Robotics

Oct 16, 2013 (3 years and 7 months ago)

73 views

VLDB 2012 Summary

Bo Zhao

Day 1: QDB Workshop


People


Gerhard
Weikum

(MPI, creator of YAGO), Michael
Benedikt

(Oxford), Michael Ley (Trier, creator of
DBLP), Luna Dong (
At&t
), Erhard
Rahm

(Leipzig),
Ihab

Ilyas

(QCR), etc.


Two keynotes:


Entity Resolution by Erhard
Rahm


Data Cleaning by
Ihab

Ilyas


Paper from our group: truth finding on
numerical data


Observation: Data quality issues draw more
attention in today’s big data environment.


Day 2


Attended Talks:


An analysis of structured data on the web.


CrowdER
: Crowdsourcing Entity Resolution.


CDAS: A Crowdsourcing Data Analytics System


Entity Resolution Tutorial by
Lise

Getoor

http://www.cs.umd.edu/~getoor/Tutorials/ER_VLDB2012.pdf


Panel on big data (quote: hot term played by machine
learning people, but
VERY LARGE database
cares about big
data since the very beginning,
db

people need to catch up)


Observation: crowd sourcing becomes popular, one
important issue is still data quality. But there are some
new aspects, e.g. costs.


Day 4


Give two talks:


Marina’s pattern mining paper.


Truth finding paper.


Attended talks:


PARIS: probabilistic alignment of relations, instances and schema
(Matching of two knowledge bases)


Learning expressive linkage rules using genetic programming (Entity
resolution)


Supercharging recommender systems using taxonomies for learning
user purchase behavior (Utilizing taxonomies for recommendation)


Who tags what? An analysis framework (social tagging)


Whom to ask? Jury selection for decision making tasks on micro
-
blog
services (Crowdsourcing, considering cost)


Multilingual schema matching for Wikipedia info
-
boxes (schema
matching)


REX: explaining relations between entity pairs (efficiently finding graph
patterns linking two entities).

Interesting Findings


Quite a few machine learning style papers.


But the general audience is not very familiar with
machine learning, but they are very interested in
it.


e.g. In
Lise

Getoor’s

tutorial, she asked how many
people have heard of LDA? around 10 raised
hands; and Markov Logic
Networks?around

5
(including me and
Yizhou
..)


If you have a new machine learning method that
happens to be scalable, and has interesting
applications (to
db

community), consider sending
it to VLDB.