Cludo Search is not a black box - the search relevance is based on a unique blend of machine learning and human customization.
How to determine Relevancy
Relevancy for search can be measured using the Mean Reciprocal Rank (MRR). Mean Reciprocal Rank is a statistical measure that takes a list of possible search page rankings and ordered by the position of the relevancy page ranking and click-through rates. For example, if someone searches a term, clicks on the first-page result, that would be a perfect MRR score of 1. The reciprocal rank is calculated using:
Let's show you an example of how this is calculated for search. For example:
|Search Query||Page Rankings||Clicked on Ranking||Rank in Rankings||Reciprocal Rank|
|dog||doggy, doghouse, dogs||dogs||3||1/3|
|monkey||monkey, monkey bars, monkey pets||monkey bars||2||1/2|
|cat||cat, catholic, category||cat||1||1|
The standard for what is considered a good MRR rate is 0.5. This means on average your search visitors are clicking on the 2nd result or higher.
How to customize your search algorithm
With Cludo, there are multiple ways you can customize your search algorithm, from determining what page results show up for specific queries, to emphasizing pages that are part of a specific section of the site or using machine learning to dynamically adjust your algorithm based on what's most relevant at that time.
Going from tools that allow the most control to the standard Cludo algorithm, here are the tools that you can use to customize your search algorithm.
1. Page rankings
Custom page rankings overrule all settings, controllable on a per-query basis.
2. Machine Learning (intelligent rankings)
Automated intelligent rankings. Cludo learns and refines relevance overtime on a per-query basis. Essentially: the more clicks on a result, the higher the ranking, meaning that your users help improve your search result relevance over time.
3. Engines (boosting)
Automated rankings based on rules the administrator sets from Engine settings. Boosting lets you define sections of your site that should have a higher/lower priority amongst your search results.
4. Search Application
The search application, no rules applied, is based on the Okapi BM25 algorithm (BM=Best Matching).