Minutes from the meeting of 6/4 Attending: Serge,Vasilis,Jason,Yue, Michael,Janet * Michael talked about his previous work on materialized object views and maintenance. * We discussed what to do next, especially, what new ideas can we have to improve our existing view mechanism (and maybe to produce a paper). We agree that materialized view maintenance is interesting, and for the next meeting, we will try to: -- Come up with one or two basic view maintenance algorithm, or at least some good examples (see Jason's write up: ~mchughj/lore96/papers/views/mat2.ps) -- Read related work: - Chaw's work on change representation (DOEM paper) - Jason's examples - Yue and Hector's paper: ~zhuge/public_html/papers/gsviews.ps - Dan Suciu's paper on maintaining views using UnSQL -- Think about the following questions: - Is it possible to decide whether a view is affected (and how) by a base database update at compile time, using DataGuide or similar data structures? - How to decide whether a view is affected (and how) by a base update at run time, using some kind of access structure, such as inverse indices? - What are the appropriate (and good) data structures to use for view maintenance?