Skip to content

Relational classification filters overview

Relational classifications often follow relations that produce many related rows.

Filters are used to narrow those rows down so the classification evaluates only the rows that matter.

Instead of counting every related row, filters allow the system to:

  • keep rows that match a condition
  • remove rows that should not be considered
  • compare row sets
  • evaluate time relationships

Filters help express real business rules more precisely.

Types of filters

Relational classifications support three types of filters.

Classification filters

Classification filters reuse an existing classification to decide whether a row should remain in the result set.

Example:

Relation path

Booking → Room

Filter

Include classification: Active room

Only rooms marked as active are kept.

Set filters

Set filters compare two sets of rows produced during evaluation.

Example:

A customer may have bookings and reservations.

A set filter could check:

Bookings intersect Reservations

This allows the classification to reason about how two sets of rows relate to each other.

Temporal filters

Temporal filters compare time intervals.

Example:

A room booking may overlap with another booking.

A temporal filter could check:

Booking overlaps Room maintenance period

Temporal filters make it possible to evaluate scheduling conflicts and time-based logic.

When filters are evaluated

Filters are applied during evaluation, not after.

Evaluation happens step by step:

  1. Start with the row being evaluated
  2. Follow a relation to collect related rows
  3. Apply filters to remove unwanted rows
  4. Continue to the next relation step
  5. Count the remaining rows at the end

If filters remove all rows, the classification result becomes false.

Example

Tables

  • Customer
  • Booking
  • Room

Relations

Customer → Booking → Room

Filter

Include Active room

Evaluation

Customer Rooms found Rooms counted
Anna Room101, Room102 Room101
Erik Room103 Room103

The filter ensures that only active rooms are counted.

Why filters matter

Without filters, relational classifications could produce large result sets that do not reflect the intended rule.

Filters make relational logic:

  • more precise
  • easier to understand
  • easier to reuse

They ensure the classification counts only the rows that are relevant to the rule.

A simple way to think about filters

Filters answer the question:

Which of the related rows should actually be considered?

By applying filters during evaluation, the system can express complex rules while keeping the underlying relation paths simple.

Related concepts