rai-pyrel

PyRel

Summary

What: The PyRel v1 language surface — types, concepts, properties, relationships, data loading, expressions, derived-property business rules, and query construction.

When to use:

• Writing or reviewing any PyRel model code; looking up imports, types, or declaration patterns
• Translating a business rule to PyRel ("flag high-value customers", tiers, segments, scores) — see Rules Authoring
• Querying: select, filter, join, aggregate, rank, export — see Querying
• Debugging syntax errors, FDError, Unground Variables, or empty/unexpected results

When NOT to use: reasoner-specific tasks route to their reasoner skill — it owns the patterns and pitfalls:

• Ontology design decisions (concept vs property, identity, data mapping, enrichment) — see rai-ontology
• Optimization formulation (decision variables, constraints, objectives) — see rai-prescriptive-problem
• Graph analysis (centrality, community, reachability, paths) — see rai-graph-analysis
• GNN modeling and training — see rai-predictive-modeling, rai-predictive-training
• Reasoner routing and question discovery — see rai-discovery; connection/config — see rai-setup

Overview: Modeling declares the surface (concepts, properties, relationships); Definitions bake business logic into the model; Rules Authoring is the workflow for deriving new properties from natural-language rules; Querying reads it all back. Most logic belongs in definitions — queries should be simple reads over what definitions computed.

---

Quick Reference

from relationalai.semantics import (
    Model, Float, Integer, String, Date, DateTime, distinct,
)
from relationalai.semantics import Number            # always Number.size(p,s), never bare
from relationalai.semantics.std import aggregates as aggs
from relationalai.semantics.std.aggregates import rank, desc, asc, top, bottom
from relationalai.semantics.std import strings, math, numbers

model = Model("my_model")
Product = model.Concept("Product", identify_by={"id": Integer})
Product.cost = model.Property(f"{Product} has {Float:cost}")            # many-to-one
Product.supplier = model.Property(f"{Product} supplied by {Supplier:supplier}")  # functional FK
model.define(Product.new(model.data(df).to_schema()))                   # load data
result = model.where(Product.cost > 10).select(Product.id, Product.cost).to_df()
print(Product.cost > 10)      # readable repr — verify structure before querying
Product.cost.inspect()        # executes, prints DataFrame