Vadis Consulting

Toggle banner

KDD 2013

Vadis is proud to announce its high ranking at the 2013 KDD Cup, the global Knowledge, Discovery and Data Mining Competition organized by ACM SIGKDD, the leading professional organization of data miners. The 2013 challenge was a paper-author identification problem. Vadis finished at 33rd position out of 554 teams coming from all over the globe.

SIGKDD, the Special Interest Group on Knowledge Discovery and Data Mining organizes each year an interdisciplinary leading research conference. This conference brings together researchers and practitioners from all aspects of data mining, knowledge discovery, predictive analytics and techniques for the analysis of Big Data and large-scale data analytics. SIGKDD sponsors the KDD Cup competition every year in conjunction with the annual conference and it is aimed at members of the industry and academia. This year’s challenge was based on Microsoft Academic Search platform data.

Microsoft Academic Search is a public search engine developed by Microsoft Research Asia. This open platform enables to explore all academic literature, papers and articles of numerous authors published in journals and/or presented at conferences, related to various fields of study, including computer science. The database covers more than 50 million publications and over 19 million authors, with weekly updates.
Microsoft Academic Search indexes not only millions of educational publications; it also enhances the main relationships between and among subjects, content and authors. The main innovation of this engine is based on the detection of named entities. One of the main challenges of providing this service is caused by the problem of author-name ambiguity. On one hand, there are many authors who publish under several variations of their own name. On the other hand, different authors might share a similar or even the same name. This year’s challenge was to predict when Microsoft’s detection tool was wrong.

Predictive modeling covers a set of techniques and algorithms allowing the prediction of a targeted behavior based on past observations. Vadis with its proprietary predictive modeling software RANK proved once again its effectiveness since a team of only 2 experts (Dr. Jordan Goblet and Dr. Doru Tanasa) worked on this challenge during less than a week to still finish at the 33rd position out of 554 teams. It is indeed not the first time Vadis ends in the top ranks of data mining competitions (KDD 2007  - 6th, ECML PKDD 2008 - 2nd, KDD 2009 - 16th, KDD2010 Cup - 11th…)

Our key clients

BNP Paribas Fortis
VP Bank
European Comission
Bank of china
Wolters Kluwers

Our key partners

Peking University
Brussels invest & export
Know your country
Sales Consult
World Compliance
Bureau van dijk
NET Computer
BECI Member
Also pay a visit to our sister site