The rationale behind a result that Rainbird has produced can be viewed by clicking the ‘i’ symbol next to the result. Clicking the ‘i’ symbol will take you to the evidence tree, which provides a visual representation of the rationale behind a result. The evidence tree will present you with a fact card detailing the conditions of the rule used to produce a result:
Figure 1: Evidence tree fact card
The rationale can be further interrogated by clicking the square symbol in the “Conditions” header to access the Salience View.
The Salience View is presented as a donut chart, and is a visual representation of the impact the weight of each condition in the rule had on the certainty of the fact the rule has produced.
Figure 2: The salience view. Two conditions with equal weighting make up the rule
In Figure 2, both conditions had equal weighting in producing the fact ‘Pete speaks English’. The overall certainty factor is only 75%, despite both conditions producing 100% certain facts, because the certainty factor on the rule has been set to 75%, so Rainbird can only ever be a maximum of 75% sure a fact it produces because of the “%Person speaks language %Language” rule will be true. Each of the conditions impacted, with 100% certainty, half of the maximum rule certainty.
If a user was less than 100% certain about any facts they provided, or if Rainbird was less than 100% certain about any facts it already knew or created, this would impact the overall certainty and be demonstrated in the salience view:
Figure 3: Changing the certainty of an answer affects the overall certainty
In Figure 3 above, a user was only 90% certain that Pete lives in England, which has affected the impact of the ‘lives in’ condition and the overall certainty of the ‘Pete speaks English’ fact accordingly.
For more information on how the certainty and impact are calculated or affected, check out the articles on Overall Certainty and/or Weight.
If it was determined that, the country a person lives in should have no (zero) impact on the rule, the condition would be set to have 0 weight. The zero-weighted condition would then be listed under the ‘zero salience conditions’ section of the salience view, along with any other conditions that have not impacted the rule.
In Figure 4 below, because the ‘lives in’ condition has zero weighting, and therefore zero influence on the rule, the ‘national language’ condition will have a 100% impact on the overall certainty factor:
Figure 4: No affect for Zero salience conditions
Let’s add a new optional condition to the ‘speaks language’ rule, ‘%Person born in %Country’, and make the question on the ‘born in’ relationship skippable. Each of the three conditions in the rule has the default weighting of 100%. Now, if we were to run a query on the ‘speaks language’ relatonship, and we skipped the ‘born in’ question, the salience view would grey out the skipped condition and reduce the overall certainty of the result:
Figure 5: The ‘born in’ condition is grey out as it has not been met
The skipped condition has had 0% impact on the overall rule. Despite having zero impact on this rule, as the condition has been assigned a weighting and has an overall impact on the certainty of the result, the ‘born in’ condition appears as part of the doughnut diagram and is not listed under the zero salience conditions section.
The result in Figure 5 has been returned with 50% certainty, out of a possible 75%. This is because only two of the three conditions, which were equally weighted, have been met.
The RBLang below will generate the example map built in the article. Click on ‘Export .rbird’ to download the knowledge map, or ‘copy RBLang’ and paste the code directly into Rainbird.
Query and Results
To see a detailed evidence tree, please run the query on the ‘receives final decision’ relationship.
