SVM is fine but can be very expensive (and complex) for training especially for text-like applications. Regularized logistic regression can be just about as good for document classification and is much easier to implement. I suspect that random forests would work very well as well.
As a GSOC project, SVM would be a good thing to implement for mahout. So would all of the other algorithms. On 3/29/08 10:47 AM, "Marko Novakovic" <[EMAIL PROTECTED]> wrote: > I collabotate with one proffesor form my faculty, > whose phd thesis was about machine learning in SE-s. > He uses combination of Naive Bayes and SVM. I didn't > understand his solution enough. > But I think that SVM is very useful and deployable > algorithm for SE-s. > Do you think that I should change anything in my > application. > > Greetings > > --- Ted Dunning <[EMAIL PROTECTED]> wrote: > >> >> SVM is not the only solution to these problems. For >> many search engine >> applications, it isn't even likely to be the best. >> Regularized logistic >> regression is a strong candidate as are random >> forests and boosted trees. >> >> Beware of any author who claims that their algorithm >> for machine learning >> that claims to be better than all others. The >> algorithm may well have some >> virtues, but it is unlikely to be universal. It is >> more likely that the >> author who claims this simply has a limited view of >> the range of things that >> might need to be done. >> >> >> On 3/29/08 10:23 AM, "Marko Novakovic" >> <[EMAIL PROTECTED]> wrote: >> >>> The implementation of SVM algorithm at Hadoop >> platform >>> >>> Abstract: >>> >>> I have been researching in Search Engines >>> functionalities, like ranking, presenting relevant >>> page to users, etc. >>> I noted that the most usable solution for search >>> engines is Support Vector Machine. >>> The best solution for determination relevant page >>> ranking for user based search result is SVM. >>> Reference to this problem is article: >>> T. Joachims, F. Radlinski: "Search Engines that >>> Laerning from Implicit Feedback," IEEE Computer, >>> August 2007, pp 38 >>> According to SVM is very complex algorithm, which >> has >>> a lot of operations, >>> I decided to implement SVM algorithm at Hadoop >>> platform. >>> >>> Dear Apache, >>> >>> My Idea: >>> >>> I have idea to implement model and solution for >>> retrieving relevant ranking Web pages driven by >> user's >>> past behavior. >>> According to SE-s have a lot of crawled Web pages, >>> this operation must be realized distributed if we >> want >>> to obtain results in real time and have fresh >> learned >>> database. >>> So we should paralelize all algorithms, which are >> used >>> for processing Web pages. >>> So I decided to implement the most used and >> exploited >>> algorithm in machine learning, deployed in >> operating >>> Web pages. >>> I also, choose SVM algorithm because it is very >>> complex algorithm for implementation >>> and I like temptations and I am not affraid of >> hard >>> tasks. >>> I tend to achieve most a big degree of >> performances >>> through paralelization. >>> I will exploit working on this project for writing >> new >>> article about deployment of clustering at SE-a. >>> I have prepared to this project reading articles: >>> [1] C. Burges, "A Tutorial on Suppot Vector >> Machines >>> for Pattern Recognition," Kluwer Academin >> Publishers, >>> Boston >>> [2] R.E Fan, P.H Chen, C.J. Lin, "Working Set >>> Selection Using Second Order Information for >> Training >>> Support Vector Machines," Journal of Machine >> Learning >>> Research 6 (2005), pp 18891918 >>> I also have read Hadoop documentation and examined >>> your implementations of algoritm kMeans at this >>> platform. >>> >>> Methodoligies of Development: >>> >>> - Test Driven Development >>> - Deployment ANT an JUnit >>> - SDK: Eclipse >>> - SVN System for Versioning >>> - Javadoc >>> >>> About Me: >>> >>> My resume you can see at link >>> http://atisha34.googlepages.com/. >>> I also participate in some academic projects at my >>> college: >>> - Working at topic based Search Engine, called >> Grain, >>> which is in construction at my faculty. >>> - Tutorial about SE-s, mentored by professor >> Veljko >>> Milutinovic: "The New Avenues in Search Engines" >>> presentation: >>> http://atisha34.googlepages.com/Searchengines.ppt >>> abstract: >>> >> > http://atisha34.googlepages.com/TheNewAvenuesinWebSearch.docx >>> I should publish article driven by this >> presentation >>> at IPSI Magazine. >>> - Other projects in which I participate aren't >> related >>> to machine learning and search engines. >>> >>> My Interests: >>> - Search Engines >>> - Software Engineering and Test Driven Development >>> - Machine Learning >>> - Database Modeling and OO Design >>> - ERP and Business Processes >>> >>> Sincerely Yours, >>> Marko Novakovic >>> >>> --- Karl Wettin <[EMAIL PROTECTED]> wrote: >>> >>>> Marko Novakovic skrev: >>>> >>>> Hi Marko, >>>> >>>>> I apply for SVM algorithm at Hadoop platform. >>>>> I hope that I will be accepted by Google and >>>> Appache, >>>>> I am serious in intention to do this jos as >> great. >>>> >>>> great news! Feel free to post your proposal here >>>> too. >>>> >>>> >>>> karl >>>> >>> >>> >>> >>> >>> >> > ______________________________________________________________________________ >>> ______ >>> Looking for last minute shopping deals? >>> Find them fast with Yahoo! Search. >>> >> > http://tools.search.yahoo.com/newsearch/category.php?category=shopping >>> >> >> > > > > > ______________________________________________________________________________ > ______ > Special deal for Yahoo! users & friends - No Cost. Get a month of Blockbuster > Total Access now > http://tc.deals.yahoo.com/tc/blockbuster/text3.com >
