My recommendation would be to simplify the alert down to it's most basic/minimum required conditions. For example, based on the other conditions the vendor type will always be "Windows" so there's no need to specifically state it as part of the rule. Also, if the status is "Unreachable" you will receive alerts for this condition even when the node status is "Down" or some other configured upstream dependency is down. That's probably not the desired effect at all so I would recommend removing that anyway. As for "unknowns", while extremely infrequent these are general polling errors that can occur periodically. Certainly not something that's worthy of notification unless it's been sustained for longer than two polling cycles.
Try configuring the alert in its most basic form as written below and see if this helps resolve the issue.
Trigger Alert when All of the Following Apply
Component Name is equal to SNMP Service
Component Status is equal to Down