Will electronic commerce be the killer app for data mining?  There are good arguments that it may.  E-commerce sites collect massive amounts of data on customer purchases, browsing patterns, usage times, and preferences.  Sites also can collect information on competitors' offerings and prices.  They can adjust their assortments, prices, and promotions quickly and dynamically, based on changing trends and personalization rules.  Because e-businesses implement close-loop computerized solutions, many of the traditional barriers to the effective application of data mining are significantly lower, such as access to data, data transformations, process automation, and timeliness of discoveries.

As our understanding of data mining has improved, the core technologies are being deployed with specific goals in mind, and often as components of larger systems.  We are moving along the technology adoption lifecycle, crossing Moore's "chasm" from the early adopters to the early majority.  With this shift, solutions showing clear return on investment (ROI) now become critical.

This special issue of the journal Data Mining and Knowledge Discovery is dedicated to data mining, knowledge discovery, and e-business.

We solicit high-quality, original papers describing applications of data mining and knowledge discovery to electronic commerce and e-business, as well as applied and fundamental research addressing data mining issues particular to these areas.  In all cases, the papers should describe clearly the contributions to the field, how the paper supports these contributions, and the relationships to existing work.  For applications papers, contributions should include a clear description of the problem, evidence of significant ROI or important new capabilities (as much as possible), and lessons learned with potential generalizations.

All areas of electronic commerce are relevant.  Particular problems of interest include, but are not limited to: personalization (both model discovery and deployment), mass customization, increasing market basket value (e.g., cross-selling), improving customer satisfaction and loyalty, improving search facilities, recommender systems (e.g., collaborative filtering), improving navigation, improving marketing, improving advertising (e.g., ad matching and profiling), increasing frequency of visits and conversion rates, reducing costs, business-to-consumer and business-to-business transactions, competitive intelligence, shopping agents, and the transfer of mined knowledge to conventional stores and conventional distribution channels (e.g., direct channels, self-service channels, indirect channels).
 
 
 
 
 

Also relevant are general technical issues when applied to e-commerce.  These include, but are not limited to: integration with larger e-commerce systems and data warehouses, incorporating performance feedback (e.g., campaign management) to improve models, data transformations (e.g., creation of customer signatures and profiles), multi-level data (e.g., hierarchical data), text mining, clickstream mining (e.g., web log analysis and abstractions), integration with syndicated data, incorporating prior business knowledge, post-processing operations (e.g., visualization and workflow integration), privacy issues, and emerging standards (e.g., APIs).

Each paper should describe the following (when relevant):

• The e-commerce application and the need for data mining.  Can the formulation be abstracted?  How significant is the problem?
• Who are the users of the techniques?  Of the learned knowledge?
• Who "paid" for the work?
• How are success and ROI measured?
• What was the size of the data?  Was it limited (e.g., was sampling used)?
• How were the raw data transformed into formats suitable for existing algorithms?  What processes were required?  Why were certain transformations done?  Were others tried?  Examples: data cleaning, missing values, data rollup/aggregrations, hierarchical abstraction, customer signature generation, denormalization, and feature construction.
• Why were specific algorithms chosen, and which others were tried?
• What role did background knowledge play and how has it affected the process?
• What were the post-processing operations?  How were the results explained to users (visualizations, dimensionality reduction, explanations, what-if scenarios, sensitivity analyses)?
• How did the results affect the target users?  Is mining done repeatedly, or was this a one-shot task?

Authors are encouraged to submit high quality, original work that neither has appeared in, nor is under consideration by, other journals. Submissions should be in 12pt font, 1.5 line-spacing, and should not exceed 28 pages.  Shorter submissions, including technical notes also are solicited.

 Electronic submissions are required; postscript or Acrobat PDF will be accepted.

Authors are encourage to look at the review form prior to submission.

 Please follow the instructions here to submit papers. You do not have to submit hardcopies to Kluwer.

• Submission deadline: 16 Nov 1999
• Acceptance notification: 22 Feb 2000

Please check http://robotics.stanford.edu/~ronnyk/ecommerce-dm/ for more details and review criteria.  Authors are encouraged consider the criteria when crafting their submissions. Specific questions and clarifications should be sent to ecommerce-dm@cs.stanford.edu