Spread
Spread and minimum similarity determine the amount and quality of the search terms included in the results list. The Minimum similarity is an absolute limit (for example 80% - below this value no data records are considered). In contrast, the Spread parameter is a relative limit. It influences the search results relative to the similarity value of the best hit. The Spread is stated in percent. This value is deducted from the similarity value of the best hit. The result is the minimum similarity value which has to be met by data records to be included into the results list.
Example 1
FACT-Finder finds five records with the following similarities: 90%, 88%, 80%, 78%, 75%.
The Spread is 7%. 90% - 7% = 83%. Therefore, two records are included in the results set – those registering 90% and 88%.
The effect of the Spread is only applied in a second step, after the reduction of results through the Minimum similarity. Even if the Spread is set to allow a hit with a similarity value of 70%, if the Minimum similarity is set to 80%, the record will not be included.
Example 2
FACT-Finder finds five records with the following similarities: 90%, 88%, 80%, 78% and 75%.
The Minimum similarity is set to 80% and the Spread is set to 15%. 90% - 15% = 75%. Based on the Spread, a hit with 78% and also one with 75% would be included. However, only three records are included (those with similarity values of 90%, 88% and 80%), as the Spread is applied with reference to the Minimum similarity. Below, there are no more hits.
Impact
The Spread has no effect on the search performance. It is only applied after the search. However, it has a positive effect on the quality of the search result. A lower Spread ensures that, in the case of good hits (the first hit has a very high similarity), the hits with a “relatively” low similarity (80%) are not displayed. If, however, there are no good hits, the hits with a lower similarity are displayed.
How to change settings
See the Search Algorithm page.
Recommendation
We recommend a Spread of between 8% and 12%. The default setting is 12%. If you find that the last few pages of search results contain unsuitable products, you should reduce this value in order to therefore limit the similarity range.
When using Personalization and/or Semantic Enhancer, we recommend re-evaluating the Spread with regard to these modules.