How to retrieve unique values in typeahead
Hi experts, Please help me out on this. I have a collection of about 30K documents which pertain to pop artists (eg. Madonna, Michael Jackson). These artist names are indexed in the field named artist_t which has the following properties in dynamic field declaration: dynamicField name=*_t type=text indexed=true stored=true/ Most of the documents will have MJ as their artist. I am using EdgeNGram filter factory to get a typeahead implementation. i.e. when I type in m I would get madonna, michael jackson, miley cyrus etc as results. The problem that I have now is that all these terms are repeated. When I search for m, instead of madonna, michael jackson I am getting MJ repeated many times in the initiall 10 docs that solr brings by default. I need to make all these artists unique i.e if I search m, I should get individual results just once? How should I change the schema file and is there a query tweaking required? Any help would be dearly appreciated. Thanks and regards, Manas
Re: How to retrieve unique values in typeahead
I have a collection of about 30K documents which pertain to pop artists (eg. Madonna, Michael Jackson). These artist names are indexed in the field named artist_t which has the following properties in dynamic field declaration: dynamicField name=*_t type=text indexed=true stored=true/ Most of the documents will have MJ as their artist. I am using EdgeNGram filter factory to get a typeahead implementation. i.e. when I type in m I would get madonna, michael jackson, miley cyrus etc as results. The problem that I have now is that all these terms are repeated. When I search for m, instead of madonna, michael jackson I am getting MJ repeated many times in the initiall 10 docs that solr brings by default. I need to make all these artists unique i.e if I search m, I should get individual results just once? How should I change the schema file and is there a query tweaking required? http://wiki.apache.org/solr/TermsComponent (can be used for Auto-Suggest) can eliminate repeated terms. With this solution you don't need EdgeNGram anymore. If you want to suggest more than one term, you can add ShingleFilterFactory to your index analyzer chain.
