yes, good point. What I want to reach is to calculate some "average" of a group of articles and find the most relevant new articles for this group.
that's why I took a look at mahout recommender. Does it make sense? Or I'd better investigate solr more like this? Best Regards Alexander Aristov On 24 July 2012 06:23, Lance Norskog <[email protected]> wrote: > A text search engine might be the right tool for this. Text search > engines do recommendations. Lucene uses TF/IDF (and other distance > algorithms). TF/IDF is basically cosine similarity. > http://lucene.apache.org/solr/ > > When you do a text search in newspaper articles for, say, "yoga", the > TF/IDF algorithm implements essentially this: every newspaper article > has a term vector. You create a newspaper article term vector that has > only one word, and find the nearest other term vectors by cosine > similarity. > > On Mon, Jul 23, 2012 at 7:03 AM, Alexander Aristov > <[email protected]> wrote: > > People > > > > i need your suggestion. I want to build a recommendation (item based) > > system but I need to use text files for data model. > > > > Is it possible to use texts for preferences and find similar items based > on > > terms? > > > > Best Regards > > Alexander Aristov > > > > -- > Lance Norskog > [email protected] >
