Data From Field Example - Biasing Down PDFs in Search Results

Arbitrary document types can be biased down in search results with a combination of Data from Field and Rank Bias settings. For example, the percentage rank of PDFs could be decreased by 2% (to push them lower in relevance-ranked query results).

To accomplish this, first create an Integer Parametric field - either at profile creation or via ToolsParametric Field Tools - or repurpose an existing numeric field. In this example we will call this field docTypeBias.

Then define a Data from Field rule (under All Walk Settings) as follows:

  • REX Search: >>=application/pdf==

  • Replace: -2

  • From Field: MIME Type

  • From Meta Field: (Empty)

  • To Field: docTypeBias

This rule looks for "application/pdf" as the MIME type for each document. If it matches, we set docTypeBias to -2. Also, for improved performance, add docTypeBias to the Compound Index Fields setting.

After doing a New rewalk to fully populate this field, go to Search Settings and set Parametric Search OptionsdocTypeBias to Sortable and Rank Bias (here) to docTypeBias. This will add docTypeBias to the percentage rank, and thus bias PDFs down by 2%. To bias PDFs up instead, simply set the Replace value in the Data from Field rule to a positive number.


Copyright © Thunderstone Software     Last updated: Apr 18 2024
Copyright © 2024 Thunderstone Software LLC. All rights reserved.