QUERY TUNING WITH JARO_WINKLER_SIMILARITY

Hello all,

I'm using utl_match.jaro_winkler_similarity in my query. And it is performing very badly. I'm thinking of creating a function based index on jaro_winkler_similarity. Can someone please share thoughts on this.

SELECT *

FROM

  (SELECT

    /*+ FIRST_ROWS */

    a.*,

    ROWNUM rnum

  FROM

    (SELECT PFI                      AS a1,

      ADDRESS_CLASS                  AS a2,

      BUILDING_NAME                  AS a3,

      COMPLEX_NAME                   AS a4,

      EZI_ADDRESS                    AS a5,

      FLOOR_NO_1                     AS a6,

      FLOOR_NO_2                     AS a7,

      FLOOR_PREFIX_1                 AS a8,

      FLOOR_PREFIX_2                 AS a9,

      FLOOR_SUFFIX_1                 AS a10,

      FLOOR_SUFFIX_2                 AS a11,

      FLOOR_TYPE                     AS a12,

      HOUSE_NUMBER_1                 AS a13,

      HOUSE_NUMBER_2                 AS a14,