dataframe.scala

def topHashtags(tweets: DataFrame, n: Int): DataFrame =
  tweets
    .select(explode(split($"text", "\\s+"))) // split it into words
    .select(lower($"col") as "word")         // normalize hashtags
    .filter("word like '#%'")                // filter hashtag words
    .groupBy($"word")                        // group by each hashtag
    .agg(count("*") as "count")              // aggregate the count
    .orderBy($"count" desc)                  // order
    .limit(n)                                // limit to top results