Spark streamingを使用したtwitter解析によるレコメンドサービス例

46
1 GMO イイイイイイイイイイイ イイイイイイイイイイ イイイ2015.8.1 Spark Streaming イイイイイ Twitter イイイイイ イイイイイイイイイイ

Transcript of Spark streamingを使用したtwitter解析によるレコメンドサービス例

  1. 1. 1 GMO 2015.8.1 Spark Streaming Twitter
  2. 2. 1. 2. 3. 4.Spark StreamingTwitter Mikasa- 5.Spark 6.Spark Streaming 7. 8. 2
  3. 3. 1. 3 Java2EE Perl/Ruby Hadoop Github: https://github.com/AKB428 Twitter: @n428dev Qiita: http://qiita.com/AKB428 Hatena: http://akb428.hatenablog.com/ : http://akibalab.info/ Junichi Noda Larry Wall
  4. 4. 2. GMO 132 20152 Spark Streaming and GraphX (45) Spark Streaming 4
  5. 5. 2. Twitter Spark Streaming SparkScalaScala Spark 5
  6. 6. 3. 6
  7. 7. 7
  8. 8. 4. Spark Streaming Twitter Mikasa- https://github.com/AKB428/mikasa_ol https://github.com/AKB428/mikasa_rs 8
  9. 9. 9
  10. 10. http://youtu.be/o6xq_RdNv6s 10
  11. 11. 11
  12. 12. 12
  13. 13. Twitter Twitter Streaming API 400 Spark Streaming 13
  14. 14. 14 kuromoji (CSV)
  15. 15. 15
  16. 16. 16
  17. 17. Spark Streaming Mikasa560 60*2=120) 17
  18. 18. https://github.com/AKB428/mikasa_ol/blob/maste r/src/main/scala/mikasa.scala https://github.com/AKB428/mikasa_ol/blob/maste r/dictionary/anime_2015_3Q.txt 18
  19. 19. 19
  20. 20. 5 (F1)20153 TV http://tv-anime.xyz/5/ (F1) http://tv-anime.biz/5/ (F2) http://lovelive-net.com/5/ (F2) http://akiba-net.com/5/ (F2)TV http://telev.net/5/ F1= F2= 20
  21. 21. 21 60 (F1)20153 TV http://tv-anime.xyz/60/ (F1) http://tv-anime.biz/60/ (F2) http://lovelive-net.com/60/ (F2) http://akiba-net.com/60/ (F2)TV http://telev.net/60/ F1= F2=
  22. 22. 22
  23. 23. https://akiba-souken.com/article/24303/ 23
  24. 24. 2 (2015/) 24
  25. 25. 3 2015722 722 25
  26. 26. Twitter https://twitter.com/Akihabaara_itso https://twitter.com/AuctionMadoka 26
  27. 27. 27
  28. 28. 28
  29. 29. 5! (F1)20152 TV http://tv-anime.xyz/5/ (F1) http://tv-anime.biz/5/ (F2) http://lovelive-net.com/5/ (F2) http://akiba-net.com/5/ (F2)TV http://telev.net/5/ F1= F2= 29
  30. 30. 30 60 (F1)20152 TV http://tv-anime.xyz/60/ (F1) http://tv-anime.biz/60/ (F2) http://lovelive-net.com/60/ (F2) http://akiba-net.com/60/ (F2)TV http://telev.net/60/ F1= F2=
  31. 31. Mikasa Twitter Twitter Filter + 31
  32. 32. Twitter Filter + 32
  33. 33. MikasaConoHa() .com http://www.onamae.com/ https://www.conoha.jp/ 33
  34. 34. 4. Spark Streaming TwitterUtilJavaTwitter twitter4j TwitterTwitterAPIAPI IP Twitter 34
  35. 35. 5. Spark HadoopMapReduce(DAG) HadoopHadoop 2015611 1.4.0 https://spark.apache.org/releases/spark-release-1-4-0.html 1.4.0SparkR(RSpark APIScala, Java, Python 35 SparkR
  36. 36. 5. Spark SparkDAG 36
  37. 37. 5. Spark 2015615 IBM3500Spark http://japan.zdnet.com/article/35065964/ 2015 61517 Spark summit 2015 () NTT 2Hadoop CDHHDPSpark Spark1.4UI http://qiita.com/AKB428/items/71900eecdf2350e1e45f 37
  38. 38. 6. Spark Streaming Spark 13 IP 38
  39. 39. 6. Spark Streaming ssc.socketTextStream(" localhost", 7777) IP 39
  40. 40. Mikasa https://gist.github.com/AKB428/c30bc6a979e05fa3a022 TwitterAPIAmazonAPI1 AmazonAPI TwitterAPI10 Mikasa OL https://github.com/AKB428/mikasa_ol Mikasa RS https://github.com/AKB428/mikasa_rs 40
  41. 41. 7. -1 Scala Scala PlaySBT 41
  42. 42. 7. -2 Learning Spark Spark 42
  43. 43. 8. TwitterSpark StreamingSQLDB SQL 43
  44. 44. GMO Hadoop http://recruit.gmo.jp/engineer/jisedai/ 44
  45. 45. 45
  46. 46. 46