Rearranging Relational Algebra Expressions
Comparison: Optimizing high-level vs. low-level language
We rely on the mathematical properties of relational algebra operators to
do this transformation:
| Before |
After |
project(
select(
join(Artist, Album),
name == 'The Detroit Cobras')
[title])
|
project(
join(
select(Artist, name == 'The Detroit Cobras'),
Album),
[title])
|
Compare to this low-level query, which also does a join and a select
From lecture 01:
tweets = {} // map from date to int
// Clear count for each date
for each date between '2016/03/01' and today:
tweets[date] = 0
// How many accounts started on 3/1/16? Record tweets by date for each.
accounts = 0
for each account in allAccounts:
if account.created == '2016/03/01':
accounts += 1
for each tweet in account.tweets:
tweets[date(tweet.timestamp)] += 1
// Print results, sorted by date
for each date between '2016/03/01' and today:
averageTweets = tweets[date] / accounts
print date, averageTweets
|