1 Reply Latest reply on Oct 29, 2012 7:28 AM by 969865

    Db indexing - mapping a multivalued comma separated column

      I am stuck up on a question of how to index a comma separated column value from database table into a dimension/property in Endeca.

      The same problem encountered in XML crawling was easily solved by adding additional tags in the xml record for the respective property as in below record for the property Category.

                <PROP NAME="Id">
                <PROP NAME="Name">
                     <PVAL>Toy Story</PVAL>
                <PROP NAME="ReleaseYear">
                <PROP NAME="Rating">
                <PROP NAME="Category">
                <PROP NAME="Category">
                <PROP NAME="Category">

      same record in data base looks like
      Id     Name          ReleaseYear     Rating     Category
      1     Toy Story     1995          3.0          Children,Animation,Comedy

      i am clueless as of how to map csv (Children, Animation, Comedy) value individually to a dimension/property in Endeca while indexing the same.

      Could someone please give me pointers to approach this problem.

      Thanks in anticipation:)


      Edited by: NitinHD on Oct 28, 2012 10:47 PM
        • 1. Re: Db indexing - mapping a multivalued comma separated column
          Following threads have were useful in solving the problem that i was facing:

          Thread: How to use comma separated values as separate dimensions
          Thread: Examples of manipulators

          The solution was to create a data manipulator for splitting the multivalued column and assigning it to a property/dimension.
          Following manipulator was created taking help from the above threads and documentation of Developer studio(Data Foundry Expression Reference and Developer Studio Help):

          <EXPRNODE NAME="OLD_NAME" VALUE="movieCategory"/>
          <EXPRNODE NAME="NEW_NAME" VALUE="Genre"/>
          <EXPRNODE NAME="SPLIT" VALUE=","/>