| /* |
| * Description : Fuzzy joins two datasets, DBLP and CSX, based on ~= using Jaccard of their titles' 3-gram tokens. |
| * DBLP has a 3-gram index on title, and we expect the join to be transformed into an indexed nested-loop join. |
| * We expect the top-level equi join introduced because of surrogate optimization to be removed, since it is not necessary. |
| * Success : Yes |
| */ |
| |
| drop dataverse test if exists; |
| create dataverse test; |
| use dataverse test; |
| set import-private-functions 'true'; |
| |
| create type DBLPType as closed { |
| id: int32, |
| dblpid: string, |
| title: string, |
| authors: string, |
| misc: string |
| } |
| |
| create type CSXType as closed { |
| id: int32, |
| csxid: string, |
| title: string, |
| authors: string, |
| misc: string |
| } |
| |
| create dataset DBLP(DBLPType) primary key id; |
| |
| create dataset CSX(CSXType) primary key id; |
| |
| create index ngram_index on DBLP(title) type ngram(3); |
| |
| write output to nc1:"rttest/inverted-index-join-noeqjoin_ngram-fuzzyeq-jaccard.adm"; |
| |
| set simfunction 'jaccard'; |
| set simthreshold '0.5f'; |
| |
| for $a in dataset('DBLP') |
| for $b in dataset('CSX') |
| where gram-tokens($a.title, 3, false) ~= gram-tokens($b.title, 3, false) and $a.id < $b.id |
| return {"atitle": $a.title, "btitle": $b.title} |