Data mining top-level meeting

Some people work very original, there are some very new things every year. Some people have a lot of articles, but mainly follow others ' work. There are many paper machine in the database field. In some places, the whole group is a big paper machine.

Personal feeling database researchers tend to think of data mining as a sub-domain of a database, and thus have lower rating for data mining meetings. For other backgrounds, however, data mining is an emerging area of relative independence, and therefore rating higher for its meetings.

Sigmod:97, the highest meeting of the database, covers a wide range of applications (since theoretical articles have pods). Didn't see?, admire like surging river water. This meeting is not only Double-blind review, but also has rebuttal procedure, it is unique and unique.

VLDB:95, a European database conference, has a history of more than 30 years. The host land is basically in accordance with the rule of one year Europe, the other continent rotation. It is the only meeting that is close to sigmod and is generally considered to be equally respected with SIGMOD. Its PC is a bit diversified, and it may take a bit of geographical balance when taking articles. As a result, submissions to the United States may even feel more difficult than sigmod. More articles from outside the United States can be seen at this meeting.

Pods:95 points. Is "the best meeting of database theory and also a good theoretical meeting". Always co-located with sigmod every year. The people who feel the algorithm background are dominant (you can count the number of pods articles from Motwani Group), and also some of the AI backgrounds (after all, Sigart is also one of the sponsors). Its influence is far less than sigmod, however, the quality of the article is relatively neat, variance less than sigmod (and any other database meeting). One bull said: "PODS never had a really bad paper," It's a place to be proud of.

Kdd::full paper 95 points, Poster/short paper 90 minutes. The highest meeting of data mining. Due to the lack of historical accumulation and the area of small circle, do not use to deny KDD now than Sigmod is still inferior. I think we can do this analogy: Kdd:sigmod=crypto:stoc. Looking back at the history of cryptography, the real most bull articles are usually sent in Stoc/focs rather than Crypto/eurocrypt, which is similar to today's data mining! However, if you look at today's cryptography articles, there are already top-level cryptology (I cannot write a name) and no longer contribute to stoc/focs. I think the same thing will happen in the near future in data mining, let's wait and see. The quality of KDD has been high in the past few years. The quality of its full paper is higher than that of SIGMOD/VLDB data mining paper. The reason is that there are few data mining sigmod/vldb reviewers, and the standard of peer review is not necessarily well mastered. In recent years, several SIGMOD/VLDB data mining paper have follow some KDD paper. And in KDD, it's hard to get a full paper. Fudan took an article last year, it is very valuable. This year they also took a SIGMOD demo, which shows that the work is really solid. I heard that in many places, if you can have a SIGMOD/VLDB/KDD, you can graduate, can have two articles can find a good job. "The revolution has not been successful, comrades still need to work!" ”

icde:92 points. A good database meeting is also a hodgepodge. The benefits are broad coverage, strong tolerance, the disadvantage is the uneven level of the article.

edbt:88 points, good database meeting, admission rate is very low however, the historical accumulation is insufficient, the influence is obviously less than ICDE.

ICDT:88, Pods's European version, second conference on database theory. Like Sigmod/vldb, ICDE and EDBT are comparable in quality and influence. Others, such as CIKM,ICDM,SDM,SSDBM,PKDD, are worse than the meetings above.

Cikm:85 points.

Sdm:full Paper 90 points, Poster/short paper 85 minutes. Siam Data Mining Conference, and ICDM and listed as the second in the field of data mining, there is a significant difference than KDD. It seems that the statistical background of the people more, but also a part of the machine learning background of people, compared diversified.

Icdm:full Paper 90 points, Poster/short paper 85 minutes. IEEE Data Mining Conference, with SDM and listed as the second in the field of data mining, there is a significant difference than KDD.

PKDD:83 (because the number of poster/short paper is very small, so no distinction). It seems to be the European version of KDD, but it's a big gap with KDD.

CIDR: The emerging conference in the database field, which only paper the Vision category, emphasizes the idea of innovation and does not require a very solid, complete results. It was launched in 2002 by Michael Tonebraker, Jim Gray, and David DeWitt. In the 2002 Sigmod keynote speech, Michael Stonebra Ker said that now SIGMOD paper have an awesome map and formula to be included, often not conducive to truly groundbreaking work, he wants everyone to do sea The job of change. CIDR is a two-year meeting, 93 and 95 each opened, the submission and collection of relatively few, but basically the main researcher and the field of the leader are there, more focused some. Although it has only been open for two sessions, it has become a very important database meeting. Of course, the meeting still needs time to test.

Data mining top-level meeting