nextQL Guide
nextQL is RecomNext's purpose-built query language for filtering, boosting, and segmenting recommendation candidates at request time.
What is nextQL?
nextQL lets product owners define per-scenario filter and booster expressions that the recommendation engine evaluates at request time against the candidate item set. It also supports segmentation expressions that compute dynamic segment membership for each item.
Filter Syntax
Filters return only items whose attributes satisfy the condition. Attributes are referenced by quoted name.
| Element | Syntax | Example |
|---|---|---|
| Attribute | Quoted string | 'price', 'category' |
| Comparison | attribute op value | 'price' < 100 |
| Operators | = != < <= > >= in not_in | 'status' != 'discontinued' |
| Array | [val1, val2, ...] | 'category' in ['a', 'b'] |
| And | expr and expr | 'price' < 100 and 'stock' > 0 |
| Or | expr or expr | 'onSale' = true or 'price' < 20 |
| Grouping | (expr) | ('a' = 1 or 'b' = 2) and 'c' > 0 |
Examples
// Electronics under $150 'category' = 'electronics' and 'price' < 150 // Highly rated books 'category' = 'books' and 'rating' > 4.5 // Items in specific categories 'category' in ['clothing', 'sports', 'electronics'] // In stock and not discontinued 'stock' > 0 and 'status' != 'discontinued'
Booster Syntax
Boosters multiply the score of matching items. Each booster is a single expression on its own line.
boost by <factor> where <filter_expression> // factor: any positive number (1.0 = no change) // condition: same syntax as filter expressions
Examples
boost by 2.0 where 'featured' = true boost by 1.5 where 'category' = 'electronics' boost by 1.3 where 'price' < 30 boost by 0.5 where 'stock' < 5
Segmentation Expressions
Used in auto-nextQL segmentations. These expressions evaluate against each item and return a set of segment names the item belongs to.
| Element | Syntax | Description |
|---|---|---|
| Attribute (set) | 'attributeName' | Returns the attribute's values as a set |
| Set literal | {"val1", "val2"} | Creates an explicit set |
| Empty set | {} | Empty set (no segments) |
| Set union | expr + expr | Combines two sets |
| Conditional | if <cond> then <expr> else <expr> | Conditional branching based on attribute |
| Map / Lambda | map(lambda 'x': <body>, <collection>) | Transform each element of a collection |
| String convert | string('attr') | Convert attribute value to string |
| Date format | string('attr', '%Y-%m') | Format date with tokens: %Y, %m, %d |
| Grouping | (expr) | Parentheses for precedence |
Examples
// Group by genres + conditional tag
'genres' + (if 'critically_acclaimed'
then {"Critically Acclaimed"} else {})
// Price-tier segmentation
if 'price' < 20 then {"Budget"}
else (if 'price' > 100 then {"Premium"}
else {"Mid-Range"})
// Transform dates into month buckets
map(lambda 'x': {string('created', '%Y-%m')},
'category')
// Combine category with brand
'category' + 'brand'Attribute Resolution
Attributes are resolved from the item's top-level fields first (e.g., category), then from a nested attributes object. Array-valued attributes are automatically unwound into individual set members in segmentation expressions.