I felt the same about number 3.

  [x in tweet.split() for x in wordlist]

Seems better to me than

  map(lambda x: x in tweet.split(),wordlist)

Although the natural thing for me to do would be

  [x for x in wordlist if x in tweet]

Which both generates a list of words instead of booleans(['scala', 'sbt'] instead of [True, False, False, False, False]) and also matches the 'sbt' in tweet.

or just:

    set(tweet.split()) & set(wordlist)

Or with numpy:

    2 * numpy.arange(1, 11)

The usual notation for working with elements of vector spaces is nice when working with arrays of numbers.

It isn't.