change similar queue length

eda/node2vec/src/main/scala/com/gmei/Main.scala

@@ -149,7 +149,7 @@ object Main {
       val matrix = new IndexedRowMatrix(rows)
       val lsh = new Lsh(
         minCosineSimilarity = 0.5,
-        dimensions = 128,
+        dimensions = 20,
         numNeighbours = 200,
         numPermutations = 10,
         partitions = 200,
@@ -173,12 +173,12 @@ object Main {
 
 
       // group by neighbours to get a list of similar words and then take top k
-      val result = remapSecond.groupBy(_._1).map {
+      val result = remapSecond.filter(_._1.startsWith("diary")).groupBy(_._1).map {
         case (word1, similarWords) =>
-          // sort by score desc. and take top 10 entries
-          val similar = similarWords.toSeq.sortBy(-1 * _._3).filter(_._2.startsWith("diary")).take(50).map(_._2).mkString(",")
+          // sort by score desc. and take top 20 entries
+          val similar = Try(similarWords.toSeq.sortBy(-1 * _._3).filter(_._2.startsWith("diary")).take(20).map(_._2).mkString(",")).getOrElse(null)
           (word1,s"$similar")
-      }
+      }.filter(_._2.split(",").length > 9)
       result.take(20).foreach(println)
 
       val similar_result = result.toDF("cid","similarity_cid")