@@ -484,7 +484,7 @@ object FactorGenerationStage extends DatagenStage with Logging {
484
484
.join(undirectedKnows(personKnowsPerson).as(" knows" ), $" Person1.id" === $" knows.Person1Id" , " leftouter" )
485
485
.join(personNumFriends.as(" personNumFriends" ), $" personNumFriends.Person1Id" === $" knows.Person2Id" , " leftouter" )
486
486
487
- frequency(knows2, value = $" frequency" , by = Seq ($" Person1.id " ), agg = sum)
487
+ frequency(knows2, value = $" frequency" , by = Seq ($" knows.Person1Id " ), agg = sum)
488
488
},
489
489
// posts
490
490
" personNumPosts" -> Factor (PersonType , PostType ) { case Seq (person, post) =>
@@ -534,27 +534,20 @@ object FactorGenerationStage extends DatagenStage with Logging {
534
534
},
535
535
// tags
536
536
" personNumTags" -> Factor (PersonHasInterestTagType ) { case Seq (interest) =>
537
- frequency(
538
- interest,
539
- value = $" TagId" ,
540
- by = Seq ($" personId" )
541
- )
537
+ frequency(interest, value = $" TagId" , by = Seq ($" personId" ))
542
538
},
543
539
" personNumFriendTags" -> Factor (PersonHasInterestTagType , PersonKnowsPersonType ) { case Seq (interest, personKnowsPerson) =>
544
- frequency(
545
- undirectedKnows(personKnowsPerson).as(" knows" )
546
- .join(interest, $" personId" === $" knows.Person2Id" , " leftouter" ),
547
- value = $" TagId" ,
548
- by = Seq ($" knows.Person1Id" )
549
- )
540
+ val personNumTags = frequency(interest, value = $" TagId" , by = Seq ($" personId" ), agg = count)
541
+ .select($" personId" .as(" Person1Id" ), $" frequency" )
542
+
543
+ val friendTags = personNumTags.as(" personNumTags" )
544
+ .join(undirectedKnows(personKnowsPerson).as(" knows" ), $" personNumTags.Person1Id" === $" knows.Person2Id" , " leftouter" )
545
+
546
+ frequency(friendTags, value = $" frequency" , by = Seq ($" knows.Person1Id" ), agg = sum)
550
547
},
551
548
// forums
552
549
" personNumForums" -> Factor (ForumHasMemberType ) { case Seq (hasMember) =>
553
- frequency(
554
- hasMember,
555
- value = $" ForumId" ,
556
- by = Seq ($" PersonId" )
557
- )
550
+ frequency(hasMember, value = $" ForumId" , by = Seq ($" PersonId" ))
558
551
},
559
552
" personNumFriendForums" -> Factor (ForumHasMemberType , PersonKnowsPersonType ) { case Seq (hasMember, personKnowsPerson) =>
560
553
frequency(
@@ -575,11 +568,7 @@ object FactorGenerationStage extends DatagenStage with Logging {
575
568
},
576
569
// companies
577
570
" personNumCompanies" -> Factor (PersonWorkAtCompanyType ) { case Seq (workAt) =>
578
- frequency(
579
- workAt,
580
- value = $" CompanyId" ,
581
- by = Seq ($" PersonId" )
582
- )
571
+ frequency(workAt, value = $" CompanyId" , by = Seq ($" PersonId" ))
583
572
},
584
573
" personNumFriendCompanies" -> Factor (PersonWorkAtCompanyType , PersonKnowsPersonType ) { case Seq (workAt, personKnowsPerson) =>
585
574
frequency(
0 commit comments