Alberto Lerner
Contact: last '.' first sent to the gmail domain
I have finished my Ph.D. at the Ecole Nationale Superieure de Telecommunications (ENST/Paris) in 2003. Thanks professors Dennis Shasha, Eric Simon and Talel Abdessalem! I have since worked in the industry and in research labs designing and building large data management systems. I'll be also keeping my work home page updated.
During my thesis work, Prof. Shasha and I considered the opportunities of manipulating data that is naturally ordered. Think about it: in Biology, sequences abound; in Finances, times series; and so in other domains. We wanted to be able to manipulate large amounts of such data with a declarative query language. Existing query languages, most notably SQL, lack the notion of order -- or had it added as an after-thought.
I don't mean order as in 'ORDER BY,' which is a late sort step. I mean expressing computations such as 'for each row, subtract the column "price" from the previous row's "price" value'. The keyword here being 'previous'. How to define 'previous' if data is not ordered? (If you like a challenge, try that query. Look at its query plan. Is the optimizer being smart? Could it?)
We developed a query language, its algebra and their query engine that are capable of improving both on expressiveness and performance of previous attempts. For more information, please take a look at the AQuery project page.