====================
TASK 1 XQuery Update
====================

- Error messages important!

--------------------
1.1
--------------------
for $idattr in doc("addressbook")//address/@id
return (
   delete node $idattr,
   insert node element myID {string($idattr)} as first into $idattr/..
)

--------------------
1.2
--------------------
declare updating function local:insert-twitter($pid as xs:string, $tw as xs:string) {
   for $a in doc("addressbook")//address[@id = $pid]
   return 
     if(empty($a/twitternames)) 
		then insert node element twitternames { element twitter {$tw}} into $a
     else 
       if ($a/twitternames/twitter/text() = $tw)
       	then ()
       	else insert node <twitter> { $tw } </twitter> as last into $a/twitternames
};
local:insert-twitter("address3", "jo2")

--------------------
1.3
--------------------
let $new := 
	for $e in doc("addressbook")//address[@state = "sync"]
  return 
   copy $je := $e
   modify delete node $je/street 
                union $je/code 
                union $je/city 
                union $je/twitternames
   return ( element org { $e }, element cpy { $je })
return put(document { $new }, "/synced.xml")


=======================
TASK 2 Precision/Recall
=======================

-----------------------
2.1
-----------------------
Precision = |{relevant documents} intersection {retrieved documents}| / |{retrieved documents}|
Recall = |{relevant documents} intersection {retrieved documents}| / |{relevant documents}|

For the given example:
Precision = 8/(8+2) = 8/10 = 0.8
Recall = 8/(8+5) = 8/13 = 0.61

-----------------------
2.2
-----------------------
Precision 1.0, recall approaches 0.0
 falseNeg=unlimited, truePos=1, falsePos=0: pure but incomplete result
 
Recall 1.0, precision approaches 0.0 
 falseNeg=0, truePos=1, falsePos=unlimited: result resembles complete db content
 
 
================================
TASK 3 Ranking Documents: TF/IDF
================================

Term Frequency (Tf) = f(t,d)
Normalized Term Frequency (NTf) = f(t,d)/max{f(w,d):w->d}
Inverse Document Frequency (IDf)= log(|N|/{1+|{d->D:t->d}|})

--------------------------------
3.1
--------------------------------
For 'concert':
f(concert,d1) = 0 => NTf = 0
f(concert,d2) = 1 => NTf = 1/2 (max{f(w,d2):w->d2} = 2 for 'soo' or 'is')
f(concert,d3) = 0 => NTf = 0

For 'berlin':
f(berlin,d1) = 2 => NTf = 2/2 = 1 (max{f(w,d1):w->d1} = 2 for 'berlin')
f(berlin,d2) = 0 => NTf = 0
f(berlin,d3) = 1 => NTf = 1/2 (max{f(w,d3):w->d3} = 2 for 'of')

For 'live':
f(live,d1) = 0 => NTf = 0
f(live,d2) = 1 => NTf = 1/2 (max{f(w,d2):w->d2} = 2 for 'berlin')
f(live,d3) = 1 => NTf = 1/2 (max{f(w,d3):w->d3} = 2 for 'of')


--------------------------------
3.2
--------------------------------
IDf(concert,D) = log(3/1+1) = log(3/2)
IDf(berlin,D) = log(3/1+2) = 0
IDf(live,D) = log(3/1+2) = 0

--------------------------------
3.3
--------------------------------
score(Q,d1) = Tf(concert,d1).IDf(concert,D) + Tf(berlin,d1).IDf(berlin,D) + Tf(live,d1).IDf(live,D)
= 0.log(3/2) + 2.0 + 0.0
= 0

score(Q,d2) = Tf(concert,d2).IDf(concert,D) + Tf(berlin,d2).IDf(berlin,D) + Tf(live,d2).IDf(live,D)
= 1.log(3/2) + 0.0 + 1.0
= log(3/2)

score(Q,d3) = Tf(concert,d3).IDf(concert,D) + Tf(berlin,d3).IDf(berlin,D) + Tf(live,d3).IDf(live,D)
= 0.log(3/2) + 1.0 + 1.0
= 0