In July this year, The 2018 SIGIR Workshop On eCommerce was held in Ann Arbor, Michigan, USA, and Rakuten Institute of Technology participated in order to present a report on the SIGIR eCom'18 Rakuten Data Challenge that our Boston office organized in the months leading up to the event itself.
The SIGIR Workshop on eCommerce is a platform for publication and discussion focused on Information Retrieval and NLP research, specifically for their applications in e-commerce. The workshop brings together many participants from academia and industry to discuss the current challenges and solutions in the area today.
The Rakuten Data Challenge focused on the problem of large-scale taxonomy classification, a very fundamental issue in the e-commerce field. In order to catalogue products well and enable them to be searchable and available for things like personalized recommendations, the categorization needs to be very accurate and easily managed. Manual approaches are not scalable, as they often end up with very deep tree structures, and research in how the issue can be solved has been limited due to a lack of publicly available commercial data. To lift this problem, and help the field along, Rakuten Institute of Technology Boston organized the challenge and provided Rakuten data to the participants. We are happy that it attracted many participants that brought many interesting results to the table, and we look forward to organizing similar challenges in the future.
We also provide data like this and more in our Rakuten Data Release program, where researchers can sign up to receive real-world data from our various services to use in their research. If you have something going on and this type of data could help you, please do not hesitate to reach out to us.
For more information on the event and Rakuten Data Release, see the links below.