Student Profiles
Once the platform has interpreted a child’s assessment evidence into skill statuses and area statuses, it takes one more step: composing all of those statuses into a single learning profile for the child. The profile is not a score. It is a structured interpretation of where the child is in their reading development and what kind of support is most likely to help.
This page explains how the profile is built, what it contains, how summaries are written for different audiences, and what role the teacher plays in bringing the profile into effect.
From statuses to a profile
The interpretation pipeline produces skill statuses for each of the 79 individual skills and area statuses for each of the reading areas. The profile module takes all of those statuses for a given assessment window and composes them into one coherent picture.
The goal of this composition step is to identify the most meaningful instructional pattern for the child right now. Two children with different patterns of weak skills may receive different profiles even if their overall distributions of “meeting” versus “below” statuses look similar, because the location of weaknesses in the skill map matters, not just the count.
The layers of a profile
A profile has five layers, each adding more specificity.
The main picture (primary profile)
The primary profile is the central characterization of the child’s reading pattern. It is selected from a closed catalog of roughly 12 patterns. Examples of what these patterns capture include:
- Decoding is the primary bottleneck; foundational word-reading skills need strengthening
- Comprehension gaps appear even when decoding is adequate
- Fluency is limiting comprehension despite accurate word reading
- A broad difficulty spanning decoding, fluency, and comprehension together
- A pattern that is generally on track but with specific skill alerts worth watching
There is also a guard state for cases where there is not yet enough evidence to form any reliable picture. When this guard state applies, no intervention is recommended and no support plan is assigned. The teacher sees a clear signal that more assessment data is needed.
The access driver
For profiles where comprehension is a central concern, the platform identifies what is driving the comprehension gap. Is the child’s access to text limited by difficulty with decoding individual words, by accuracy errors that interrupt meaning-making, or by fluency limitations that prevent holding meaning across a passage? This distinction matters because the type of support that helps a decoding-limited child is different from the support that helps a fluency-limited child, even when both show comprehension gaps.
The comprehension subtype
When comprehension itself is the focus, the platform further identifies whether the gap is primarily in literal comprehension (understanding what the text directly states), inferential comprehension (drawing conclusions the text implies), or a mix of both. These subtypes guide the specific instructional activities within a support plan.
Modifier profiles (up to five)
The primary profile captures the main pattern. Modifier profiles layer on secondary patterns that co-occur with it. Up to five modifiers can be active at the same time. The five available modifiers are:
| Modifier | What it signals |
|---|---|
| Vocabulary, syntax, and morphology gap | A co-occurring weakness in word knowledge, sentence structure, or morphological patterns |
| Arabic orthographic accuracy monitoring | A pattern in Arabic-specific accuracy features that warrants gentle monitoring |
| Writing and spelling support | A note that writing and spelling support should accompany the primary intervention |
| Listening and language support | A note that listening comprehension and oral language support are relevant |
| On track with a skill alert | The child’s overall reading is progressing well, but one or more specific skill alerts are open |
Modifier profiles do not replace the primary profile; they refine it. An intervention plan is built around the primary profile with the modifiers informing supplementary components.
Fine-grained signals (sub-flags, up to eight)
The most specific layer of the profile is a set of up to eight sub-flags. These are the most granular signals from the skill map, surfaced to help educators and specialists design the details of an intervention. Sub-flags include things like a madd length-mark accuracy pattern, a short-vowel marking accuracy note, a vocabulary alert, or a writing and spelling signal.
Sub-flags are internal to the educator-facing view. They are never shown to children. They inform intervention design analytics and help specialists who are reviewing a child’s plan understand the specific skill pattern in detail.
A note on Arabic-accuracy features
When a child shows a pattern on Arabic-specific orthographic accuracy features (shadda, tanween, madd, short vowels), the platform routes this to the “Arabic orthographic accuracy monitoring” modifier rather than to the primary profile or a gateway-level concern. This is a deliberate design choice: these features require monitoring and teacher attention, but they should not, on their own, drive a major instructional reorganization. The modifier appears in the teacher’s view but is not visible to the child or shown to parents.
Profile confidence
The platform records a confidence level alongside the primary profile: high, medium, or low. Confidence reflects how much comparable, consistent evidence was available to support the profile selection.
A high-confidence profile means the skill statuses across the map told a clear, consistent story, and the profile captures it well. A low-confidence profile means the evidence supported a conclusion but with more uncertainty than usual; the teacher should review it carefully and consider whether gathering more data before acting would be worthwhile.
Anti-overlap rules
The profile catalog includes four rules that prevent incompatible primary profiles from being selected together. These rules exist because some combinations of patterns, if both were labeled simultaneously, would produce a plan that is internally contradictory. The rules ensure that when two plausible patterns are in tension, the one with the stronger evidence base takes priority and the other is recorded as a secondary note rather than a co-equal primary.
The profile starts pending: the teacher decides
Every profile the platform computes starts in a “pending teacher review” state. It does not become active until the teacher explicitly confirms it.
This is a core principle. The platform interprets the evidence and proposes a profile. The teacher reviews the proposal in light of what they know about the child from daily classroom interaction, from conversations with the family, and from their own observations. If the profile matches what the teacher sees, they confirm it. If the teacher believes something important is missing or that more data is needed, they can mark the profile accordingly.
A profile confirmed by the teacher becomes active and drives the intervention recommendation. A profile the teacher flags for more data is recorded as withdrawn; no active profile exists for that child until a new one is confirmed.
This teacher-confirmation requirement is not a formality. It is the mechanism that keeps the teacher in the decision-making role and prevents the platform from automatically imposing support plans without professional review.
What happens when evidence updates
When new assessment data arrives for a child and a new profile is computed, the new profile is placed in pending. The teacher reviews it again. The previously active profile moves to “superseded” in the archive; it remains accessible for historical reference but is no longer the operating profile.
This cycle repeats with each new assessment window, so the profile stays current with the child’s development and the teacher’s ongoing judgment.
Plain-language summaries for four audiences
Each active profile produces a written summary in Arabic tailored for a specific audience. Four audience types are supported:
Teacher summary: Provides the full educational context. It describes what the child’s pattern suggests about their reading development, what the profile means in instructional terms, and what kinds of activities and support approaches are indicated. It uses professional educational language that assumes teacher expertise.
Parent summary: Written entirely in growth-focused language. It describes what the child is learning and making progress on, what they are working toward, and how families can support reading at home. It contains no numeric scores, no internal platform labels, and no language that clinical contexts might use to describe reading difficulties. The goal is to give parents an honest and encouraging picture of their child’s development without the complexity that would belong in an educator-only conversation.
School leader summary: Provides the key pattern for the child in a form suited to a principal or administrator who is reviewing multiple children’s records for a school-level view of reading development.
Student task framing: A short, task-level framing used to introduce activities to the child. This is not a description of the child’s profile; it is a task prompt. No profile description is ever shown to the child.
How summaries are written
All summary text comes from a library of pre-written templates. There are 15 profile templates, each rendered for four grade levels, producing 60 base variants. The templates are written by specialists, reviewed for language quality, and governed by the platform’s language safety rules.
No AI system generates summary text. The templates are selected and populated by the platform’s rules engine based on the confirmed profile. Every summary passes through a language safety filter before it is written to the database. The parent-audience version applies a stricter version of that filter: it blocks numeric scores, internal code substrings, any vocabulary that might be associated with clinical labeling, and any language that is not growth-focused.
This approach ensures that what families receive is consistent, carefully crafted, and free of language that could be misinterpreted or cause unnecessary concern.
Determinism and audit trail
Profile resolution is fully deterministic. Given the same skill status snapshots and the same catalog version, the platform always produces the same profile. This means any past profile can be re-examined: an administrator can run a dry-run replay of the resolution against the original inputs and verify that the output matches. No rows are written during a replay; it is purely a verification step.
Every profile assignment permanently records which version of the profile catalog, which version of the narrative template library, and which version of the language safety filter was used. This makes the entire profile computation auditable over time.