Business Meeting Organizer A MultiAgent Meeting Scheduler using
Business Meeting Organizer A Multi-Agent Meeting Scheduler using Mobile Context Kathleen Yang Kathleen. Yang@sv. cmu. edu Neha Pattan Neha. Pattan@sv. cmu. edu Alejandro Rivera Alejandro. Rivera@sv. cmu. edu Martin Griss Martin. Griss@sv. cmu. edu C a r n e g i e M e l l 1
Agenda • Problem • Vision • Related Work • Solution • Architecture • Implementation • Experimental results • Conclusion • Q&A C a r n e g i e M e l l 2
The problem… Organizing meetings with multiple people C a r n e g i e M e l l • When to meet? • Where to meet? • Who must attend the meeting? • When to reschedule? 3
Vision Business Meeting Organizer • Schedules meetings • Co-ordinates common time slots and venues • Tracks context of attendees • Reschedules meetings C a r n e g i e M e l l 4
Related Work • Lin J. H. , J. Wu and S. L. Lai “Ameetzer” • Agent based meeting modeling to automate the task of scheduling meetings • Chen, H. “An intelligent broker architecture for context aware systems” • Broker architecture to maintain a shared model of the context • Haynes T. , S. Sen, N. Arora and R. Nadela. : “An automated meeting scheduling system that utilizes user preferences” • Adopting user preferences in meeting scheduling systems • Griss, M. , Letsinger, R. , Cowan D. , Vanhilst, M. and R. Kessler: “Cool. Agent: Intelligent Digital Assistants for Mobile” • Adaptation of multi agent based meeting schedulers to changing environment • Crawford, E. and M. Veloso: “Opportunities for learning in multi-agent meeting scheduling” C a r n e g i e M e l l 5
Goals • Mobile client that interacts with end user • Each user owns a software agent that acts like the user’s Personal Secretary • Agents communicate and negotiate meetings on behalf of their owners • Agent is responsible for tracking owner and notifying other secretaries • Agent has access to owner’s personal data, calendar and contacts • Agent executes as per owner’s preferences C a r n e g i e M e l l 6
Business Meeting Organizer • Multi agent based solution • Architecture • Implementation C a r n e g i e M e l l 7
Solution Agent Negotiation User B A Agent B Agent A Proposal Negotiation C a r n e g i e M e l l Confirmation 8
Architecture (1/2) HTTP SMS Agent Behaviors HTTP Server Behavior 1 Behavior 2 SMS Sender Services Calendar C a r n e g i e M e l l Contacts Preferences … Internet 9
Architecture (2/2) C a r n e g i e M e l l 10
Implementation (1/2) • Multi-agent environment • Privacy maintained through strict rules • Based on user preferences • Integration of location C a r n e g i e M e l l – Computes the time to travel between meetings – Agrees to a meeting only if the time to travel is less than available time between meetings: both before and after proposed meeting – Sends a reminder on time to user to start travelling to next meeting 11
Implementation (2/2) • Details about meeting reminder – Agent queries for user’s location every 5 minutes – Client application on user’s cell phone sends an update to the agent, consisting of GPS co-ordinates – Agent tracks most recent location of user – Agent computes time to travel to next meeting – Agent sends a reminder 10 minutes before the time when user should start travelling to next meeting C a r n e g i e M e l l 12
Experimental Results • Time Consumption • Calendar Privacy • Context Awareness C a r n e g i e M e l l 13
Time Consumption Comparison Four Participants Tools Meeting #1 Meeting #2 Meeting #3 Meeting #4 Meeting #5 C a r n e g i e M e l l BMO Google calendar Email Telephone Face-to-face Negotiation Request time (minutes) pending time (minutes) 1 2 0 0 7 11 5 50 0 0 14
Calendar Privacy Comparison Four Participants Meeting Meeting C a r n e g i e M e l l #1 #2 #3 #4 #5 Tools Calendar visible to others BMO Google calendar Email Telephone Face-to-face No Yes No No No 15
Context Awareness Comparison Four Participants C a r n e g i e M e l l Tools Location Calendar Meeting #1 BMO Continuous Meeting #2 Google calendar No Continuous Meeting #3 Meeting #4 Email Telephone One shot Meeting #5 Face-to-face One shot 16
Conclusion • Time spent using BMO is a fraction of that using manual techniques like email and phone calls • Manual techniques do not scale well for meetings with several attendees, BMO does! • BMO scores over other techniques in terms of – Efficiency – Privacy – Context Awareness C a r n e g i e M e l l 17
Q&A C a r n e g i e M e l l 18
- Slides: 18