| vinayakb | 38b7ca4 | 2012-03-05 05:44:15 +0000 | [diff] [blame] | 1 | use dataverse fuzzy1; |
| 2 | |
| 3 | declare type DBLPType as open { |
| 4 | id: int32, |
| 5 | dblpid: string, |
| 6 | title: string, |
| 7 | authors: string, |
| 8 | misc: string |
| 9 | } |
| 10 | |
| 11 | declare nodegroup group1 on nc1, nc2; |
| 12 | |
| 13 | declare dataset DBLP(DBLPType) |
| 14 | partitioned by key id on group1; |
| 15 | |
| 16 | write output to nc1:'/tmp/dblp.adm'; |
| 17 | |
| 18 | set simthreshold '.5'; |
| 19 | |
| 20 | for $paperR in dataset('DBLP') |
| 21 | for $paperS in dataset('DBLP') |
| 22 | where $paperR.title ~= $paperS.title and $paperR.id < $paperS.id |
| 23 | return { 'R': { 'dblpid': $paperR.dblpid, 'title': $paperR.title }, |
| 24 | 'S': { 'dblpid': $paperS.dblpid, 'title': $paperS.title }} |