use dataverse fuzzy1;

declare type DBLPType as open {
 id: int32,
 dblpid: string,
 title: string,
 authors: string,
 misc: string
}

declare nodegroup group1 on rainbow-01, rainbow-02, rainbow-03,
rainbow-04, rainbow-05;

declare dataset DBLP(DBLPType) 
  partitioned by key id on group1;

write output to rainbow-01:"/home/hyracks/dblp-self-join.adm";

//
// -- - Stage 3 - --
//
for $ridpair in 
    //
    // -- - Stage 2 - --
    //
    for $paperR in dataset('DBLP')
    let $tokensR :=
        for $word in counthashed-word-tokens($paperR.title)
        for $token at $i in
            //
            // -- - Stage 1 - --
            //
            for $paper in dataset('DBLP')
            for $word in counthashed-word-tokens($paper.title)
            group by $item := $word with $paper
            order by count($paper)
            return $item
        where $word = $token
        order by $i
        return $i
    for $prefix_tokenR in subset-collection(
                                $tokensR, 
                                0,
                                prefix-len(
                                    len($tokensR), 'Jaccard', .8))

    for $paperS in dataset('DBLP')
    let $tokensS :=
        for $word in counthashed-word-tokens($paperS.title)
        for $token at $i in
            //
            // -- - Stage 1 - --
            //
            for $paper in dataset('DBLP')
            for $word in counthashed-word-tokens($paper.title)
            group by $item := $word with $paper
            order by count($paper)
            return $item
        where $word = $token
        order by $i
        return $i
    for $prefix_tokenS in subset-collection(
                                $tokensS, 
                                0,
                                prefix-len(
                                    len($tokensS), 'Jaccard', .8))

    where $prefix_tokenR = $prefix_tokenS

    let $sim := similarity(
                    len(counthashed-word-tokens($paperR.title)),
                    $tokensR,
                    len(counthashed-word-tokens($paperS.title)),
                    $tokensS,
                    $prefix_tokenR,
                    'Jaccard',
                    .8)
    where $sim >= .8 // and $paperR.id != $paperS.id
    group by $idR := $paperR.id, $idS := $paperS.id with $sim
    where $idR < $idS
    return {'idR': $idR, 'idS': $idS, 'sim': $sim[0]}

for $paperR in dataset('DBLP')
for $paperS in dataset('DBLP')
where $ridpair.idR = $paperR.id and $ridpair.idS = $paperS.id
return {'R': {'dblpid': $paperR.dblpid, 'title': $paperR.title}, 
        'S': {'dblpid': $paperS.dblpid, 'title': $paperS.title}, 
        'sim': $ridpair.sim}