Mining Graph Data
Karel Vaculík and Luboš
Popelínský
Knowledge Discovery Lab
Faculty of Informatics, Masaryk University Brno
Czech Republic
popel@fi.muni.cz www.fi.muni.cz/~popel
Is there a need for mining in
graphs?
or exist already tools
that can manage it?
Movie information as a graph
IMDb
Movie information as a graph
What commonalities can we find about movies in
IMDb?
by frequent subgraph discovery:
Movies receiving awards (Oscars, Golden Globes)
come from the same small set of studios
Certain director/composer pairs work frequently
together
Movie information as a graph
Movie information as a graph
What common relationship can we find between object in the
db?
Movies made by the same studio also have the same producer.
An emerging film star may be characterized by a sequence of
successful movies.
Will a movie make more than $2 million in its opening weekend?
Will be the movie nominated for an award? but also
for inferring missing links in a movie graph
Mutagenesis data
mutagenic vs. non-mutagenic
substances
What are commonalities for each of those classes?
What are commonalities for each of those classes,
e.g. subgraphs, that distinguish
mutagenic and non-mutagenic substances?
Mutagenesis data
Inductive logic programming
can help.
atom(Id, Element, AdditionalInfo, ….).
bond(Id1, Id2, Arity, AdditionalInfo, …).
ring(…).
Mutagenesis data
Web
Web mining =
web usage mining
web structure mining
web content mining