Field Emphasis

Unless configured otherwise, FACT-Finder treats all fields in the search database as being equal. This is naturally not the case, however. An image URL is not relevant for the search results. On the other hand, the product name field is highly relevant.

FACT-Finder gives you the option of setting the searchability and emphasis of each field compared to other searched fields. The emphasis of fields which are not searched can be influenced as it does not exist. This emphasis is specified in percent. Strictly speaking, FACT-Finder does not expect “emphasis” (meaning the field “values” 90%) but rather a “reduction” in relevance (field values 10% less than one without reduction). This process is termed field reduction. It is not possible to increase the relevance.

If nothing is entered for field emphasis reduction all fields will be weighted the same.

The following table contains suggested values for some types of fields. These values will need to be modified in view of your data in order to achieve optimum results:

Field type

Reduction

Product name

0%

Brief description

15%

Detailed description

30%

Manufacturer

0%

Category - Level 1

5%

Category - Level 2

2%

Category - Last level

0%

Impact

Terms that are located in a field whose relevance has been reduced have less impact on the record similarity value.

How to change settings

See: Basic Settings → Field Prioritization.

Recommendation

Fields that are very important in the search process and whose content customers are likely to search precisely (for example, article names, manufacturer names), should not be reduced in value. Descriptive fields, on the other hand, should be significantly reduced in value, depending on the amount of additional information these contain and if they contain critical information (e.g. “contrary to London, Madrid is…” would be returned as a result if searching for “London” as well as searching for “Madrid”). If your category system is hierarchical, it is also advisable that you undertake a gradation of the value reduction of these fields. The final category level is more relevant to the search process.

If you work with combined category names (pillows & blankets), you should not search this category level or depreciate it if possible. This way, not every search for “pillow” will also produce all blankets because they also include the search term.

Rule of thumb: a 10% reduction of value of a field only results in approx. 2% loss of relevancy.

Page Contents