Slashdot Mirror

← Back to Stories (view on slashdot.org)

MapReduce For the Masses With Common Crawl Data

Posted by timothy on from the gotta-be-here-some-place dept.

New submitter happyscientist writes "This is a nice 'Hello World' for using Hadoop MapReduce on Common Crawl data. I was interested when Common Crawl announced themselves a few weeks ago, but I was hesitant to dive in. This is a good video/example that makes it clear how easy it is to start playing with the crawl data."

29 comments

  1. Thanks for posting this.. by kvvbassboy · · Score: 1

    This will be my first (and hopefully not last) headfirst dive into MapReduce.

    1. Re:Thanks for posting this.. by InsightIn140Bytes · · Score: 5, Informative

      Then you probably want to use it with some local data so you don't rack up huge bill. One Hadoop job on the whole dataset costs at least like $200, and that's for simple stuff.

    2. Re:Thanks for posting this.. by kvvbassboy · · Score: 1

      Warning heeded, but I saw this on a blog post at commoncrawl.org.

      This bucket is marked with Amazon Requester-Pays flag, which means all access to the bucket contents requires an an http request that is signed with your Amazon Customer Id. The bucket contents are accessible to everyone, but the Requester-Pays restriction ensures that if you access the contents of the bucket from outside the EC2 network, you are responsible for the resulting access charges. You don’t pay any access charges if you access the bucket from EC2, for example via a map-reduce job, but you still have to sign your access request. Details of the Requeser-Pays API can be found here: http://docs.amazonwebservices.com/AmazonS3/latest/dev/index.html?RequesterPaysBuckets.html

      If I understood that right, at least getting started with the tutorial will not result in me coughing up $200. Correct me if I am mistaken.

    3. Re:Thanks for posting this.. by InsightIn140Bytes · · Score: 2

      You don't need to pay for accessing it, but you still need to pay for the processing power, storage and RAM in your EC2. Of course you can start by only accessing specific day like in the video so you don't need so much processing power for it, and hence pay less. But then you also won't be able to process 99.9% of the crawl data.

    4. Re:Thanks for posting this.. by symbolset · · Score: 1

      MapReduce is an implementation of an algorithm first presented in a 1970's issue of the ACM. I would commend to startups membership and ownership of the patent-expired content composed therein. There's a lot of untapped potential in there yet - and much dross. If we will stand on the shoulders of giants though it's good to know where the giants were and what they did. Brin was a good scholar here, and Page gave something new. It was the fusion of old ideas and new that made Google. If you want to be the new Google, the ACM journals are a good start. Just remember to add something new too.

      --
      Help stamp out iliturcy.
    5. Re:Thanks for posting this.. by Anonymous Coward · · Score: 0

      Definitely not $200. I did the tutorial and it cost less than a dollar.

    6. Re:Thanks for posting this.. by Anonymous Coward · · Score: 0

      I'm not sure MapReduce is an algorithm so much as a class of algorithms, or design pattern. The algorithm implemented depends on the Map and Reduce operators used. MapReduce falls out very easily from consideration of which parts of an algorithm can be parallelized. Even embarrassingly parallel algorithms like a raytracer eventually have to combine their results to form a single image. This is all MapReduce is really expressing. I'm sure you can find it before 1970.

  2. Re:Hot Damn! Now I Can Find All The Pr0n Of My Mis by wmbetts · · Score: 1

    So you're hoping to find child porn, ghost fetish stuff, or both?

    --
    "Ubuntu" -- an African word, meaning "Slackware is too hard for me". - stolen from Dan C alt.os.linux.slackware
  3. No, Really I Am Absolutely Serious by MichaelCrawford · · Score: 0

    The problem I've got is that searches with Google and the like turn up a lot of junk that I'm not looking for, with the file search engines like FilesTube simply ignoring the numeric years specified in my search queries.

    What I want to do is find PDF files of specific issues (Month and Year combinations) of certain magazine titles. But when I try these searches, the results contain a lot of years that I had not specified, with the year I did specify not falling anywhere in the resulting pages.

    There are all kinds of ways I could use a regular expression to turn up the download sites for every magazine I want to find, but to the best of my knowledge none of the currently available search engines can take regular expressions. For example a copy of an old magazine that is available for download will typically be labeled with the file size, so I could include "[0-9]* *[Mm][Bb]" in my query. That would distinguish magazines available for download from those that are merely discussed online.

    Just last night I sat up all night long looking for old magazines, only to turn up two from the era of my interest. There is no end to the pr0n that is available online, but I don't find most of it at all interesting anymore. What I do find interesting is my goal of completely recovering the magazine collection that my ex demanded that I pitch before she would agree to come visit me for the first time.

    --
    Request your free CD of my piano music.
    1. Re:No, Really I Am Absolutely Serious by larry+bagina · · Score: 1

      "index of" and "parent directory" are good terms for finding virtual directory listings.

      --
      Do you even lift?

      These aren't the 'roids you're looking for.

  4. How Does One Profile a MapReduce Job? by MichaelCrawford · · Score: 2

    I think any total newbie that tried to process all the crawl data would soon find that his first attempt would not terminate until after The Heat Death of the Universe.

    Surely there must be some doc on how to make such jobs runs faster, use less memory as well as less storage?

    --
    Request your free CD of my piano music.
  5. You know you're old when by Anonymous Coward · · Score: 0, Interesting

    more than 50% of any given sentence sounds like gibberish. And yet you know someone somewhere is as excited as you were when you got your first floppy drive...

  6. Heh. Now That's Really Funny. Thanks for the Tip! by MichaelCrawford · · Score: 1

    WGet is chugging away even when I speak. I'm gonna have to cough up for more storage.

    Here is an SEO tips for y'all. I didn't discover it, but I stumbled across it just now:

    placing the terms "index of", "parent directory", "name", "last modified", "size" and "description" on your web pages is a real good way to attract visitors.

    I wasn't able to turn up any actual Apache directory listings for Penthouse Pet of the Year Corinne Alphen. They were all your typical pr0n site that not only weren't presenting directory listings, but none of the sites I looked at had any photos of her, scantily clad or otherwise.

    Directory listings for well-known models though, turned right up.

    --
    Request your free CD of my piano music.
  7. Regarding crawling by gajop · · Score: 3, Interesting

    Hmm, similar article so I'll ask a question of personal nature.

    I've recently created a crawler to collect certain information from a website, that would help me gather data sets for a small machine learning project.
    While I've followed robots.txt and nofollow links, site's TOU was against it. After confirming with the admin, I was told that it's not allowed to gather information, as the site owns it (as it's written in the TOU).

    The data however is publicly available, so you actually wouldn't have to agree to a TOU to collect the data, and as it's some data I wanted, I still concluded I should get a small sample (less than 1% of the total data, around 200MB) at least, to see if something's even possible to be done with it.

    What are your thoughts /.? Should I have abandoned the attempt, have I done right or even should I disregard their plead and simply get as much as I please (during a long period of time, as to not hammer on it's bandwidth)?

    1. Re:Regarding crawling by SmurfButcher+Bob · · Score: 1

      Felony.

      --

      help me i've cloned myself and can't remember which one I am

    2. Re:Regarding crawling by AllyGreen · · Score: 1

      If its publicly available, surely you can get the data elsewhere?

  8. Re:Happy Data Mining from the Golden Girls!!11! by Anonymous Coward · · Score: 0

    I'm pretty sure the word "Traveled" was not in the original song.

  9. Re:Happy Data Mining from the Golden Girls!!11! by geminidomino · · Score: 1

    Not sure if trolling (if not, well played), but it is.

    Citation

  10. Re:Heh. Now That's Really Funny. Thanks for the Ti by Anonymous Coward · · Score: 0

    add -.htm? and -.php to your searches

  11. I have no idea by Roachie · · Score: 1

    what this is.

    --
    This sig is not paradoxical or ironic.
  12. Actually, entry Level EC2 is free for 1 year by tlambert · · Score: 1

    Actually, entry Level EC2 is free for 1 year, and has been since Nov. 2010.

    You don't need to pay for accessing it, but you still need to pay for the processing power, storage and RAM in your EC2

    See here:
    http://www.infoworld.com/d/cloud-computing/amazon-web-services-offers-ec2-access-no-charge-531

    -- Terry