Thunderstone Search Appliance Manual

Parametric Field Tools

  Parametric Tools lets you work with Parametric Fields of this profile.

Modify Parametric Fields

Modify Parametric Fields allows you add, remove, or rename Parametric Fields in your profile after the profile has been created.

These changes require migrating the data to a new structure, and can be performed immediately or as part of the next crawl. Parametric Field changes that have not yet been migrated can be viewed here, and can be canceled if desired.

List/Edit Enums  

The List/Edit Enums page allows examination and editing of Parametric Search Enum fields. (See here for more on defining Enum fields.)

The List option shows every distinct value of a particular Enum field in the profile, in a table with several columns. The Source Text column is the actual crawled text value. The Display Text column is the text that will be displayed and used in searches; this defaults to the same as the Source Text, but can be edited (via download) if different text is desired. Enum Int is the internal integer value currently used to identify this Enum value in the database. New Enum Int is the new value for Enum Int that will be used at the next new crawl; if empty, no renumbering will occur.

This renumbering is occasionally needed due to the ad-hoc nature of Enum values' assignment. Instead of requiring a fixed, pre-set list of text values to be defined prior to a crawl - and thus being unable to accept new values encountered during a crawl - Enum fields can accumulate new text values dynamically during the crawl. The system will attempt to assign Enum Int integers such that their numerical order matches the alphabetical Source Text order; however, since the data is unlikely to be crawled in Source Text order, a complete alphabetical ordering may not always be possible. At the end of a crawl, this out-of-order situation is checked for, and if detected, a new set of New Enum Int values is generated, to be assigned at the next new crawl. (The renumbering is delayed until new crawl because the entire database must be renumbered, which is time consuming and cannot be done atomically.) This situation is similar to the reason a library's Dewey-Decimal book numbers may "bunch up", with a lot of subject terms very close together in a small decimal fraction span: many new subjects may have been added in between an existing pair of subject numbers. In the case of Enum fields however, instead of creating ever-lengthening fractional decimals, the entire set of integers may be renumbered, with proper spacing for new inserts.

The Download option will download the same data shown with the List option, but in Excel CSV format. This allows editing of the Display Text column if desired, i.e. to show a different or "clean" string to users for a particular value. The ordering of Enum values (for Group By Field- ordering, and relational operators search) may also be changed, by editing the New Enum Int column so that its ascending numeric order is the desired value order. (The column may be left empty or 0 if no change in the integer is desired.) Note: The Source Text and Enum Int columns must not be changed, so that edited values (for Display Text or New Enum Int) can be properly identified.

The Upload option takes an Excel CSV file - previously downloaded via Download - and applies any edits that were made. Note that new New Enum Int values will only be applied at the next new walk, as the entire database must be renumbered as well.