Web KDD Future Directions Web KDD Future Directions

Web. KDD – Future Directions

Web. KDD – Future Directions n Research directions n Forum directions

Future Research Directions n n n Integrating data from multiple sources Capturing semantic user actions Role of s/w agents in etailing Sophisticated personalization, using analytical methods Marginal value of clickstream data n n Mining from warehouses vs. flat files Preserving privacy in data mining CRM – acquisition, retention, attrition reduction Automatic detection/building of user communities

Future Research Directions n n Recommender systems for mobile devices (mcommerce) Data preparation for mining Public repository to compare data mining methods Evaluating the true ROI of recommender systems n n n Using web sites for live user experiments Impact of web mining on privacy Path analysis Combining multiple mining techniques Conveying results/impact of mining to managers

Future Research Directions n Inferring psychological variables affecting ebehavior from clickstream analysis n Impact of content & price (and changes in them) on buyer behavior

Web Mining – Some Facts n Tremendous interest in Web. KDD 2000 n n n Easy to form top notch program committee Very high quality submissions 88 attendees + 26 on waiting list Lots of work in research & practice Growing number of publications n n SIGKDD Explorations, Vol. 1. 2, Vol. 2. 1 KDD SIGMOD, ICDE, VLDB Various journals

Web. KDD Future n n n Continue at a workshop level or expand? Possible themes for next year? Any other ideas?

Announcements n n Final versions of Web. KDD 1999 papers now available as a book from Springer Verlag – please visit Springer booth Authors of Web. KDD 2000 invited to submit an enhanced version of their paper for consideration for a special issue of the KDD journal on Web Mining - instructions will be posted on the Web. KDD 2000 web page Special issue of Informs journal – flyers near the door SIAM data mining conference at Chicago in April, 2001 will have two relevant mini-workshops n Data mining for e-commerce n Web mining – flyers near the door