SIGMOD 07

2007 ACM SIGMOD / PODS (Special Interest Group On Management of Data / Principles Of Database Systems) Conferences and a pre-conference Workshop were held on June 10-14 at Beijing International Convention Center.

General Background:

The joint conferences of SIGMOD/PODS are considered the top forums in the areas of database/data management research. Publishing a paper in them, even occasionally, would be an admiration by colleagues in the area, while a series of publications in them will establish a researcher / group a star / center of excellence. The annual conferences series started in 1975, almost all have been held in the US with only three exceptional: '96 in Montreal, '04 Paris and this year Beijing. Many Chinese delegates reckoned that holding this year's event in Beijing is regarded by the Chinese research community a big flatter. It was therefore a big gathering also for the Chinese research community.

There are two Chinese regular papers entered this year's SIGMOD (probably in all its history) by Peking Univ and Harbin Institute of Technology, respectively, and a few short abstracts. However, Chinese-backgrounded (not a precise description for the case of Singapore) researchers outside China forms a very strong community, with quite a number of papers in. This year's Program-Committee Chair is Prof Beng Chin Ooi of Singapore National Univ. Hong Kong and Singapore are also very strong with several COEs and naturally as destinations for China-educated students to do their Ph.D. and post-doc researches. Therefore the past week was also a reunion for many Chinese friends/acquaintances from all over the world.

Technical Background:

As I am new to this area, my view on the technical background is mainly from my chatting with the participants an (I hope my colleague Jidong Chen will be able to help me to beefup this part when he returns from another database conference in Yellow Mountains).

1) Technical/topical shift. A few years back, the industrial presence in SIGMOD was dominated by Oracle/IBM. In the recent years, search engine companies start to gain momentum. Google/Yahoo/Baidu/Sohu have strong presence (eg, two papers in Industrial Session 1 were from Baidu and Yahoo).

2) Dynamically/mobilly/sensor-network/web-crawl/RFID related/generated data and their management become hot topics. In database sense or use of such data, they typically have a high dimension of attributes. Therefore how to efficiently process, store, retrieve, index, calculate, …, these data need new ideas.

3) Because the sort of data in (2) are often related to personal information, security/privacy/anonymity become new applied research topics in the database/data management area. In past two years submissions to SIGMOD, these topics form a majority, and this years 7 security topic papers were accepted to make security the top topic in terms of number of papers accepted / per topic.

4) Calling for a probabilistic model is in the horizon. Database and related topics were historically deterministic. Nowadays, it becomes necessary to query for, eg, probability for laptop77 appeared in room2 between 10am-2pm, or likelihood for the junction of Road X and Street Y in the northward bound to be congested in the next 20 minutes. Unlike in the usual case (in my training) that a probability computation model may ease computation, in database, it makes things harder (from my understanding of the first keynote speech of PODS).

5) As a security guy, I sat in the two sessions of security topics. In all talks, the adversary model remains passive. It seems the area uses an even more restricted attacking model: an adversary can only gain access to queries which are legal from the database interface, ie, data any user will be allowed to query. What the things for a typical research topic are, eg, to calculate whether some information will not become available to the adversary after how many (hopefully infeasible) number of queries. The passive/restricted adversary model might be reasonable for this area, but would be considered too confined (ie too strong an assumption) for the mainstream security area.

Miscellaneous:

1) Three pre-conference workshops started on Sunday, including a PhD? Student Workshop. Most participants of the workshops were Chinese students. Therefore Sunday was the best day for me to try face recognition for the Chinese students.

2) Wednesday Lunchtime Business Meeting. This is a "town hall meeting" open to all participants in the lunchtime. Seven undergraduate students worldwide were awarded SIGMOD Undergraduate Prize. Mr Wang ZhongYuan of Ren Min Univ was among the winners.

3) Having been talking to Chinese students a lot, spreading the news of ERC and explaining our instantiation of Web 2.0 into Research 2.0.

4) Google, M/soft Research, IBM Research and HP Labs were advertising job openings. IBM even held a "Student Reception" on Tuesday evening.

E-Proceedings:

I have a USB storage of e-proceedings from the registration package. I haven't seen any restriction against copying. So interested please feel free to ask for a copy.

Thanks, Wenbo