EDF2013: Big Data Tutorial: Marko Grobelnik
13 likes9,983 views
The document discusses the concept of 'big data', outlining its key characteristics, challenges, and the technologies and techniques required for effective processing and analysis. It emphasizes the importance of distributed computing and storage, as well as highlighting popular programming models like MapReduce and NoSQL databases. Additionally, it touches on the risks and complexities associated with data mining and the necessity for robust algorithms and frameworks to manage large datasets.
![Hops Nodes
1 10
2 78
3 396
4 8648
5 3299252
6 28395849
7 79059497
8 52995778
9 10321008
10 1955007
11 518410
12 149945
13 44616
14 13740
15 4476
16 1542
17 536
18 167
19 71
6 degrees of separation [Milgram ’60s] 20 29
Average distance between two random users is 6.622
21 16
10
90% of nodes can be reached in < 8 hops 23 3
24 2
“Planetary-Scale Views on a Large Instant-Messaging Network” Leskovec & Horvitz WWW2008 25 3](https://image.slidesharecdn.com/edf2013bigdatatutorialmarkogrobelnik-130418055545-phpapp02/85/EDF2013-Big-Data-Tutorial-Marko-Grobelnik-57-320.jpg)
EDF2013: Big Data Tutorial: Marko Grobelnik
- 1. Marko Grobelnik [email protected] Jozef Stefan Institute Dublin, April 9th 2013
- 2. Introduction ◦ What is Big data? ◦ Why Big-Data? ◦ When Big-Data is really a problem? Market Overview Tools Techniques Applications Literature
- 4. ‘Big-data’ is similar to ‘Small-data’, but bigger …but having data bigger it requires different approaches: ◦ techniques, tools, architectures …with an aim to solve new problems ◦ …or old problems in a better way.
- 5. From “Understanding Big Data” by IBM
- 7. Big-Data
- 9. Key enablers for the appearance and growth of “Big Data” are: ◦ Increase of storage capacities ◦ Increase of processing power ◦ Availability of data
- 20. Source: WikiBon report on “Big Data Vendor Revenue and Market Forecast 2012-2017”, 2013
- 24. Where processing is hosted? ◦ Distributed Servers / Cloud (e.g. Amazon EC2) Where data is stored? ◦ Distributed Storage (e.g. Amazon S3) What is the programming model? ◦ Distributed Processing (e.g. MapReduce) How data is stored & indexed? ◦ High-performance schema-free databases (e.g. MongoDB) What operations are performed on data? ◦ Analytic / Semantic Processing (e.g. R, OWLIM)
- 25. Computing and storage are typically hosted transparently on cloud infrastructures ◦ …providing scale, flexibility and high fail-safety Distributed Servers ◦ Amazon-EC2, Google App Engine, Elastic, Beanstalk, Heroku Distributed Storage ◦ Amazon-S3, Hadoop Distributed File System
- 26. Distributed processing of Big-Data requires non- standard programming models ◦ …beyond single machines or traditional parallel programming models (like MPI) ◦ …the aim is to simplify complex programming tasks The most popular programming model is MapReduce approach Implementations of MapReduce ◦ Hadoop (http://hadoop.apache.org/), Hive, Pig, Cascading, Cascalog, mrjob, Caffeine, S4, MapR, Acunu, Flume, Kafka, Azkaban, Oozie, Greenplum
- 27. The key idea of the MapReduce approach: ◦ A target problem needs to be parallelizable ◦ First, the problem gets split into a set of smaller problems (Map step) ◦ Next, smaller problems are solved in a parallel way ◦ Finally, a set of solutions to the smaller problems get synthesized into a solution of the original problem (Reduce step)
- 28. NoSQL class of databases have in common: ◦ To support large amounts of data ◦ Have mostly non-SQL interface ◦ Operate on distributed infrastructures (e.g. Hadoop) ◦ Are based on key-value pairs (no predefined schema) ◦ …are flexible and fast Implementations ◦ MongoDB, CouchDB, Cassandra, Redis, BigTable, Hbase, Hypertable, Voldemort, Riak, ZooKeeper…
- 30. …when the operations on data are complex: ◦ e.g. simple counting is not a complex problem ◦ Modeling and reasoning with data of different kinds can get extremely complex Good news about big-data: ◦ Often, because of vast amount of data, modeling techniques can get simpler (e.g. smart counting can replace complex model-based analytics)… ◦ …as long as we deal with the scale
- 31. Research areas (such as IR, KDD, ML, NLP, Usage SemWeb, …) are sub- cubes within the data Quality cube Context Streaming Scalability
- 32. A risk with “Big-Data mining” is that an analyst can “discover” patterns that are meaningless Statisticians call it Bonferroni’s principle: ◦ Roughly, if you look in more places for interesting patterns, than your amount of data will support almost anything, …and you are bound to find lots of nonsense Example taken from: Rajaraman, Ullman: Mining of Massive Datasets
- 33. Example: We want to find (unrelated) people who at least twice have stayed at the same hotel on the same day ◦ 109 people being tracked. ◦ 1000 days. ◦ Each person stays in a hotel 1% of the time (1 day out of 100) ◦ Hotels hold 100 people (so 105 hotels). ◦ If everyone behaves randomly (i.e., no terrorists) will the data mining detect anything suspicious? Expected number of “suspicious” pairs of people: ◦ 250,000 ◦ … too many combinations to check – we need to have some additional evidence to find “suspicious” pairs of people in some more efficient way Example taken from: Rajaraman, Ullman: Mining of Massive Datasets
- 34. Smart sampling of data ◦ …reducing the original data while not losing the statistical properties of data Finding similar items ◦ …efficient multidimensional indexing Incremental updating of the models ◦ (vs. building models from scratch) ◦ …crucial for streaming data Distributed linear algebra ◦ …dealing with large sparse matrices
- 35. On the top of the previous ops we perform usual data mining/machine learning/statistics operators: ◦ Supervised learning (classification, regression, …) ◦ Non-supervised learning (clustering, different types of decompositions, …) ◦ … …we are just more careful which algorithms we choose (typically linear or sub-linear versions)
- 36. An excellent overview of the algorithms covering the above issues is the book “Rajaraman, Leskovec, Ullman: Mining of Massive Datasets”
- 38. Good recommendations can make a big difference when keeping a user on a web site ◦ …the key is how rich the context model a system is using to select information for a user ◦ Bad recommendations <1% users, good ones >5% users click ◦ 200clicks/sec Contextual personalized recommendations generated in ~20ms
- 39. Domain Referring Domain Zip Code Sub-domain Referring URL State Page URL Outgoing URL Income URL sub-directories Age GeoIP Country Gender Page Meta Tags GeoIP State Country Page Title GeoIP City Job Title Page Content Job Industry Named Entities Absolute Date Day of the Week Has Query Day period Referrer Query Hour of the day User Agent
- 40. Trend Detection System User Stream of Log Files Stream of clicks profiles profiles (~100M page clicks per day) Sales Trends and updated segments Segments Segment Keywords NYT Stock Stock Market, mortgage, banking, Market investors, Wall Street, turmoil, New articles York Stock Exchange Campaign Health diabetes, heart disease, disease, heart, illness to sell segments $ Green Hybrid cars, energy, power, model, Energy carbonated, fuel, bulbs, Hybrid cars Hybrid cars, vehicles, model, engines, diesel Travel travel, wine, opening, tickets, hotel, sites, cars, search, restaurant Advertisers … …
- 41. 50Gb of uncompressed log files 50-100M clicks 4-6M unique users 7000 unique pages with more then 100 hits
- 42. Alarms Server Telecom Network Alarms Alarms Live feed of data Explorer (~25 000 ~10-100/sec devices) Server Alarms Explorer Server implements three real-time scenarios on the alarms stream: 1. Root-Cause-Analysis – finding which device is responsible for occasional “flood” of alarms 2. Short-Term Fault Prediction – predict which device will fail in next 15mins 3. Long-Term Anomaly Detection – detect unusual trends in the network …system is used in British Telecom Operator Big board display
- 43. The aim is to use analytic techniques to visualize documents in different ways: ◦ Topic view ◦ Social view ◦ Temporal view
- 44. Query Search Results Topic Map Selected group of news Selected story
- 46. US Elections US Budget Query Result set NATO-Russia Topic Trends Visualization Mid-East conflict Topics description
- 47. Dec 7th 1941
- 48. Apr 6th 1941
- 49. June 1944
- 50. Query Conceptual map Search Point Dynamic contextual ranking based on the search point
- 51. Observe social and communication phenomena at a planetary scale Largest social network analyzed till 2010 Research questions: How does communication change with user demographics (age, sex, language, country)? How does geography affect communication? What is the structure of the communication network? “Planetary-Scale Views on a Large Instant-Messaging Network” Leskovec & Horvitz WWW2008 51
- 52. We collected the data for June 2006 Log size: 150Gb/day (compressed) Total: 1 month of communication data: 4.5Tb of compressed data Activity over June 2006 (30 days) ◦ 245 million users logged in ◦ 180 million users engaged in conversations ◦ 17,5 million new accounts activated ◦ More than 30 billion conversations ◦ More than 255 billion exchanged messages “Planetary-Scale Views on a Large Instant-Messaging Network” Leskovec & Horvitz WWW2008 52
- 53. “Planetary-Scale Views on a Large Instant-Messaging Network” Leskovec & Horvitz WWW2008 53
- 54. “Planetary-Scale Views on a Large Instant-Messaging Network” Leskovec & Horvitz WWW2008 54
- 55. Count the number of users logging in from particular location on the earth “Planetary-Scale Views on a Large Instant-Messaging Network” Leskovec & Horvitz WWW2008 55
- 56. Logins from Europe “Planetary-Scale Views on a Large Instant-Messaging Network” Leskovec & Horvitz WWW2008 56
- 57. Hops Nodes 1 10 2 78 3 396 4 8648 5 3299252 6 28395849 7 79059497 8 52995778 9 10321008 10 1955007 11 518410 12 149945 13 44616 14 13740 15 4476 16 1542 17 536 18 167 19 71 6 degrees of separation [Milgram ’60s] 20 29 Average distance between two random users is 6.622 21 16 10 90% of nodes can be reached in < 8 hops 23 3 24 2 “Planetary-Scale Views on a Large Instant-Messaging Network” Leskovec & Horvitz WWW2008 25 3
- 59. Big-Data is everywhere, we are just not used to deal with it The “Big-Data” hype is very recent ◦ …growth seems to be going up ◦ …evident lack of experts to build Big-Data apps Can we do “Big-Data” without big investment? ◦ …yes – many open source tools, computing machinery is cheap (to buy or to rent) ◦ …the key is knowledge on how to deal with data ◦ …data is either free (e.g. Wikipedia) or to buy (e.g. twitter)