Top Research Leaders in Data Mining, Data Science, and KDD

In honor of the upcoming KDD-2014, 20th KDD Conference on Knowledge Discovery and Data Mining, we looked at the research in this field, using a very good tool from Microsoft Academic Search.

Overall, it finds 37,377 publications and 269,627 citations. Both publications and citations started growing dramatically around 1996. The Microsoft Academic search data shows that the growth peaked in 2009, and declined dramatically to almost zero around 2011. Such change cannot be explained only by term "Data Mining" becoming less popular and being replaced by "Data Science" and "Big Data". Since the total number of conferences and publications in the field has only grown since 2011, it seems that Microsoft Academic search data for 2011 and later years in incomplete. See more detailed analysis at the bottom of this post.



Here is a list of 20 top researchers in the field, according to Microsoft Academic Search for Data Mining for all time. The authors are ranked using field rating, which is similar to H-index in that it calculates the number of publications by the author and the distribution of citations to the publications, but it only calculates publications and citations within a specific field to show the impact of the scholar within that field.

N	Author (Affiliation)	Total Publications	Data Mining Publications	Field Rating
1	Jiawei Han (UIUC)	655	379	71
2	Philip S. Yu (UIC)	788	393	52
3	Rakesh Agrawal (Microsoft)	359	113	48
4	Christos Faloutsos (CMU)	484	216	45
5	Hans-Peter Kriegel (U. Munich)	452	167	37
6	Eamonn J. Keogh (UCR)	196	118	36
7	George Karypis (U. Minnesota)	330	115	36
8	Heikki Mannila (Aalto U.)	283	130	35
9	Andrew Mccallum (U. Mass Amherst)	256	78	35
10	Jian Pei (SFU)	419	151	34
11	Padhraic Smyth (UCI)	271	73	34
12	Rajeev Motwani (Stanford)	271	55	34
13	Mohammed J. Zaki (RPI)	365	130	33
14	Vipin Kumar (U. Minnesota)	571	124	33
15	Hector Garcia-Molina (Stanford)	605	87	32
16	Charu C. Aggarwal (IBM)	231	144	31
17	Bing Liu (UIC)	533	101	30
18	Prabhakar Raghavan, Google	294	50	30
19	Dimitrios Gunopulos (National U. of Athens)	202	78	29
20	Johannes Gehrke (Cornell)	228	74	29

There is little correlation between field rating and total publications (R²=0.26). There is, as expected, good correlation between field rating and the number of Data Mining publications (R²=0.69)



We note that most of correlation is driven by top 2 researchers - Jiawei Han and Philip Yu who have over 300 publications each. If we exclude them, the correlation between field rating and Data Mining publications drops to R²=0.26.



Seven of these researchers have received KDD Innovation Award

• Jiawei Han, 2004
• Rakesh Agrawal, 2000
• Christos Faloutsos, 2010
• Heikki Mannila, 2003
• Padhraic Smyth, 2009
• Vipin Kumar, 2012

 
and 5 have received IEEE ICDM Research Contributions Award

• Jiawei Han, 2002
• Philip S. Yu, 2003
• Christos Faloutsos, 2006
• Hans-Peter Kriegel, 2013
• Heikki Mannila, 2009

 
Next I looked at the top researchers in the last 10 years. Jiawei Han, Philip Yu, and Christos Faloutsos are still in the top, but the rest of the list has many new names, showing showing rapid evolution of the field.

N-10	N-all	Change	Author (Affiliation)	Total Pubs	Data Mining Publications	Field Rating
1	2	up 1	Philip S. Yu (U. of Illinois Chicago)	788	229	30
2	1	dn 1	Jiawei Han (U. of Illinois Urbana-Champaign)	655	207	29
3	4	up 1	Christos Faloutsos (Carnegie Mellon U.)	484	129	24
4	10	up 6	Jian Pei (Simon Fraser U.)	419	102	21
5	30	up 25	Yufei Tao (Chinese U. of Hong Kong)	140	50	21
6	6	0	Eamonn J. Keogh (U. of California Riverside)	196	72	20
7	5	dn 2	Hans-Peter Kriegel (U. of Munich)	452	85	18
8	78	up 70	Qiang Yang (Hong Kong U. of Science and Technology )	412	79	18
9	16	up 7	Charu C. Aggarwal (IBM)	231	77	16
10	69	up 59	Lise Getoor (U. of Maryland)	210	65	16
11	87	up 76	Jianyong Wang (Tsinghua U.)	155	50	16
12	14	up 2	Vipin Kumar (U. of Minnesota)	571	46	16
13	109	up 96	Xifeng Yan (U. of California Santa Barbara)	160	39	16
14	55	up 41	Jeffrey Xu Yu (Chinese U. of Hong Kong)	428	97	15
15	23	up 8	Ke Wang (Simon Fraser U.)	546	57	15
16	46	up 30	Beng Chin Ooi (National U. of Singapore)	291	54	15
17	74	up 57	Dimitris Papadias (Hong Kong U. of Science & Technology)	200	42	15
18	7	dn 11	George Karypis (U. of Minnesota)	330	42	15
19	90	up 71	Wei-ying Ma (Microsoft)	335	40	15
20	19	dn 1	Dimitrios Gunopulos (National and Kapodistrian U. of Athens)	202	40	15

Researchers with the largest gains in field ranking are:

• up 96, Xifeng Yan UCSB
• up 76, Jianyong Wang, Tsinghua University
• up 71, Wei-ying Ma Microsoft
• up 70, Qiang Yang HKUST
• up 59, Lise Getoor U. of Maryland
• up 57, Dimitris Papadias, HKUST
• up 41, Jeffrey Xu Yu Chinese University of Hong Kong
• up 30, Beng Chin Ooi National University of Singapore

 
so we can see that researchers from China (including Hong Kong) are becoming the leaders in data mining field.

Microsoft Academic search shows a decline in papers on "Data Mining" starting around 2010 (chart above). This is especially clearly seen in the Microsoft Academic search for KDD conference on Knowledge Discovery and Data Mining, which is the first and most-cited conference for this field.



The chart shows that the number of publications in KDD conference related to Data Mining peaked at 247 in 2009 (probably including workshop papers), but then dropped to 125 publications in 2010, and only 3 in 2011, and 1 in 2012.

While some of the change could be due to other terms like "Data Science" or "Large Scale" replacing "Data Minig" in the paper titles and session topics, this cannot explain such drop and having only 1 "Data Mining" paper for KDD-2012. Since KDD conference main focus is on Knowledge Discovery and Data Mining, all of KDD papers should be considered as relevant to Data Mining.

We also note that KDD 2011 conference had 126 accepted papers and KDD-2012 had 133 accepted papers, so a lot of data is missing from Microsoft Academic search for Data Mining starting with at least 2011.

See also "The decline and fall of Microsoft Academic Search", which shows that after 2011 Microsoft Academic Search is missing a lot of papers. Thus analysis using Microsoft Academic Search option "Last 5 years" will be unrepresentative. However, the all-time list of researchers is still very good, as evident to all who follow the field.

Related:

• KDD-2014, Top Conference on Knowledge Discovery and Data Mining, New York
• Top 10 in Data Mining: keywords, authors, publications, organizations
• Top Conferences in Data Mining