I attended the 25th International Conference on Data Engineering (ICDE
2009) held in Shanghai from March 29 to April 2. It was almost 2
months ago. I finally find time to write down the trip report and
share with you. During the conference, I presented our paper "Top-k
Exploration of Query Candidates for Efficient Keyword Search on Graph-
Shaped (RDF) Data". I think it is well-organized and there are several
interesting topics. Some important messages have been conveyed in
ICDE2009.

1. ICDE is becoming more and more like WWW. This is the first time for
social networking and Web as seperate research sessions in ICDE.

2. User Interface is becoming a new trend in the database community.
They have realized the importance of usability. In particular, during
the opening, the general chair emphasized that there are 2 regular
papers accepted by ICDE 2009 focusing on user interface while there is
one short paper.

3. Semantic Web and RDF data management are still not the mainstream
in the DB conference. Besides ours, there is another two work
belonging to the SW scope (i.e. MPI's work on steiner tree
approximation over relationship graph (Yago in the experiment) and
XOntoRank: Ontology-aware keyword search on electronic medical
records).

4. Large scale graph search and mining is still a hot topic.

5. Keyword search will keep popular in the next 3 years.

6. XML is no longer so popular compared with when it was proposed.

7. Each regular paper is required to prepare for a poster during the
poster session. Yesterday the organizer has collected the feedback of
attendees and announced it warmly welcomed. I also think it is quite
useful if you cannot listen to the presentation or you still want to
learn more about the work.

8. Cloud computing is everywhere. It seems that everyone is talking
about cloud computing.

9. There is a trend that DB will be integrated more tightly with
operating system and low-level network protocol. As the emergence of
multi-core, GPU and cloud computing, DB guys and researchers from
other areas about systems realized that there is plenty of room for
them to build the next generation DB system.

10. As for our paper, a lot of people are interested in it. During the
presentation, the room was full of audiences while I was busy with
explaining our work all the time during the poster session.

p.s. 对于large scale graph search正如beyondlei说的，RDF data和relational data都可以建模成为graph,之后就是一个graph上mining或search的问题。当然这里最困难的是图算法一般都比较复杂，而我们所要面对的图又比较大，因此就有很大研究关注大规模图上的高效搜索和挖掘算法。