Question

I am using RavenDB and allowing a user to enter an item into the database with which they can associate textual tags. The class for the item is

public class TagItem
    {
        public string Id { get; set; }
        public string UserId { get; set; }
        public List<string> Tags { get; set;}
        public DateTime DateCreated { get; set; }
    }
}

What i would like to do is get an instance count for each individual tag in a selection of items from the document query. For instance

Bacon : 12
Fudge : 8
Marshmallow : 6

I am currently doing this by iterating through the items and incrementing the count for each tag but it doesn’t seem to be a very elegant way of obtaining the results. I was wondering if anyone had a suggestion for a better way to achieve this? Thanks in advance.

— EDIT #1 —

Matt

I have tried to implement your suggestion of faceted search but I cant seem to figure out how to perform a selection using a subset of userIDs. I have created the index as the following

   public class Tag_Facets : AbstractIndexCreationTask<UserModel>
   {
    public Tag_Facets()
    {
        Map = locations => from user in users 
                           from tag in user.Tags
                           select new { Tag = tag.ToString().ToLower(), user.Id };
    }
   }

I have a List of UserIds that I would like to use to narrow this selection down before running the facet against it but none of the methods that I have tried want to compile. Can you provide me with an idea of what the Where clause would look like for this?

My FacetSetup is as follows

this.DocumentSession.Store(new FacetSetup { Id = "facets/TagFacets", Facets = new 
List<Facet> { new Facet { Name = "Tag" } } });

The query that i am trying to build is

var facetResults = this.DocumentSession.Query<LocationModel>("Tag/Facets")
                .Where(x => x.Id.Contains(new List<string> { "1", "2", "3" }))
                .ToFacets("facets/TagFacets");

which will not build because Contains doesn’t accept List

I was hoping that you might be able to point me in the right direction on how I might accomplish this.

Also, thank you for your work on Faceted Search. It seems to be what I’m looking for to perform a lot of the queries that I need to run in RavenDB. Cheers!

— EDIT #2 —

This is the SQL expression that I would use to achieve the results that I wanted.

SELECT t.Tag, COUNT(*) FROM UserTags t WHERE t.UserID in ('1', '2', '3', '4', '5') GROUP BY t.Tag

Answer 1

Another approach is to use faceted search and get the counts from that. It’s a little bit more flexible than Map/Reduce as it allows you to specify a query when you get the counts.

Given an index like this:

    from question in docs
    from tag in question.Tags
    select new 
    { 
        question.CreatedOn, 
        Tag = tag
    }

You then define the facet ranges like this (using the default behaviour, which will look for every unique term for you)

    var facetSetupDoc = new FacetSetup 
        { 
            Id = "facets/TagFacets", 
            Facets = new List<Facet> { new Facet {Name = "Tag"} } 
        }

Finally you query it like this:

    var facetResults = s.Query<Question>("QuestionTags")
                .Where(x => x.CreatedOn >= new DateTime(2008, 5, 20))
                .ToFacets("facets/TagFacets");

Take a look at this code sample, it shows how to implement your exact scenario.