Jev article depth classification

Classify an excerpt by its technical depth

Paste an excerpt to distinguish a concept overview, ordered procedure and implementation mechanism. The body must support the level; jargon and a title are not enough.

Label the explanation in an article row.Live API
Try an example

Original synthetic input. Edit the task or evidence to make a new API request.

Source: Workflow inspiration on X

Source and example scope

The inputs, rubric and preview are original. This window demonstrates the stated decision only; it does not reproduce the source application or its measurements.

Classify excerpt depth.

Decision rules and labels
Concept overviewProcedural detailImplementation internalsNeed more text
A real response from the Jev API

Edit any input to explore a different decision. No request is sent until you run it.

Your next decision

Not run yet

Choose an example or write your own input, then run it to see the ranked choices.

Results are not stored by this site. Your submitted text is processed by the API provider.

What the example checks

Six caching articles with different depths

Jev article depth classification compares six rows about caching: a purpose explanation, a setup procedure, an LRU mechanism, mixed instructions, a title without body text and unexplained jargon. Each decision stays attached to its row ID.

One subject can serve different reading needs

A table of articles can contain beginner explanations, operational tutorials and descriptions of implementation mechanisms on the same subject.

Read the example notes

All six rows here concern caching, so their shared topic does not distinguish the kinds of explanation. The depth field describes the kind of technical explanation the supplied excerpt contains. This original demo sends one row at a time to Jev and inspects one of four permitted labels. It does not fetch an article body from its URL.

The source post describes PostgreSQL article filtering

Avi Chawla's post on X describes pg-jev filtering, classifying and ranking article rows in PostgreSQL, including technical depth.

Read the example notes

The extension's repository documents Choice and ordered Score functions. Our editable window uses the existing Choice API to classify excerpt depth; it does not run the PostgreSQL extension, reproduce the recording or calculate a Score value. Author-reported timings, cache behavior and Hacker News data are outside this test.

Using the result in your application

The labels separate a concept explanation, reproducible procedural steps and a concrete internal mechanism.

Read the example notes

A row with both setup steps and an internal explanation receives the deepest level supported by its supplied text. A headline alone cannot establish depth, and naming hash maps or complexity symbols is not the same as explaining them. This rubric is an authored editorial convention, not a universal measure of article quality or correctness.

How it works

Try the example in four steps

Use the sample to understand the rule, then test your own input for article depth.

  1. Illustration of adding the text or evidence for a decision

    Add the information to check

    Paste an excerpt to distinguish a concept overview, ordered procedure and implementation mechanism.

  2. Illustration of defining a question and its allowed choices

    Confirm the question and labels

    Open the decision rules and keep the intended distinction between Concept overview, Procedural detail, Implementation internals, Need more text. Edit the choices only when testing your own rule.

  3. Illustration of sending one input and comparing the returned choices

    Run the decision

    Use a sample first, then try your own input. Compare the selected label with the other returned probabilities.

  4. Illustration of reviewing a result and an uncertain item before using it

    Review the next step

    Review Need more text before using a result. The label is a decision about the supplied information, not an executed action.

Our API test

What the six original inputs returned.

What the six original inputs returned.
Original test inputFixture expectationObserved labelLeading probability
Label a concept-only rowConcept overviewConcept overview100%
Label explicit setup stepsProcedural detailProcedural detail100%
Label an implementation explanationImplementation internalsImplementation internals100%
Use the deepest supported levelImplementation internalsImplementation internals100%
Preserve a row without body textNeed more textNeed more text100%
Do not infer depth from jargonConcept overviewConcept overview99%
  • 6 of 6 original fixture expectations matched. Contrary outputs are preserved in the table and downloads; these checks are not held-out accuracy or production suitability.
  • Measured adapter round trips: 453-1188 ms. Estimated input cost: $0.00002457-$0.00002583 per request. These observations are not a latency guarantee or invoice.

6 actual Jev API calls on October 3, 2026. Original synthetic inputs check this rubric and handoff; they do not establish held-out accuracy. Probabilities are model outputs.Verified

Understand the results

Label a concept-only row

The first row describes the purpose and staleness problem of caching without specifying a procedure or data structure.

Read the example notes

Concept overview is the expectation recorded before the request. Jev selected Concept overview with probability 1.00. A concept overview can be appropriate for its audience; depth does not measure quality. The judgment covers only this supplied excerpt, so a different section of the complete article could contain deeper material that the model never sees.

Label explicit setup steps

The second row lists controls, a concrete lifetime value and a way to check whether the configuration worked.

Read the example notes

Procedural detail is expected because those ordered steps make the excerpt an operational tutorial. We did not open a real settings screen or verify that any product has a Cache tab. The field describes the text's structure. A working procedure and a technically correct procedure still need independent validation against the actual product.

Label an implementation explanation

The third row explains how a hash map and linked list interact during a hit and an eviction.

Read the example notes

Implementation internals is the authored expectation because the excerpt describes the mechanism rather than merely naming its parts. Its O(1) claim belongs to the synthetic fixture; this page does not benchmark code or prove complexity. The model's label also does not detect every omitted invariant, memory leak or concurrency bug in a full implementation.

Use the deepest supported level

The mixed row begins with a setup instruction, then describes concrete changes to the map and list.

Read the example notes

The rule explicitly selects the deepest supported level, so Implementation internals is expected even though the first sentence resembles a tutorial. Without that tie-breaking convention, two readers might choose different labels for the same passage. A production enrichment job should version its rubric and keep it with the saved request before comparing labels across time.

Preserve a row without body text

The fifth record has an enthusiastic title but no article body excerpt.

Read the example notes

Need more text is expected; promotional metadata does not supply an internal explanation. This is distinct from assigning a low depth score to a complete article. It marks missing evidence for this row. A database workflow could collect the permitted body text in a later step and make a fresh request. It must not overwrite this observation as though the earlier request saw that text.

Do not infer depth from jargon

The final row uses technical terms and a complexity symbol without explaining an operation or dependency.

Read the example notes

Jev selected Concept overview with probability 0.99, matching the expectation under this rubric. A term count would not distinguish it from the mechanistic LRU passage. The saved response retains the remaining 0.01 probability for Need more text. This checks the supplied jargon-only fixture; it does not establish reliable grading across all technical writing.

Source and example scope

Fixture expectations are authored before the calls. Any contrary model choice remains visible beside its expectation and complete distribution; no repeated judgment is used to obtain a preferred label.

Using the result in your application

Download the exact sample requests and saved observations below.

Read the example notes

Expected labels are original fixture checks and stay outside the request sent to Jev. The observation file preserves the complete probabilities and the hash of each exact request. Editing a task, rubric or candidate requires a fresh response; a matching sample name does not make an older answer current.

Using the result in your application

The JavaScript runner makes one independent SDK request for a chosen sample ID.

Read the example notes

It reads TYPESAFE_API_KEY privately from the environment and has no automatic retry. The Python consumer operates offline: it checks the rubric, labels, unique sample identities, request hashes and valid probability distributions before selecting a fixed local preview. A tied leading result remains pending for review.

Source and example scope

The page reports measured adapter round trips and estimated input cost for the recorded calls.

Read the example notes

Those values describe these requests, not a provider latency guarantee or a billing statement. Six deliberately authored fixtures can check the stated behavior, but they do not establish accuracy on an unseen dataset. Keep disagreements and missing-input examples when testing an integration.

Runnable offline example

Replay one request and inspect the saved preview.

View the code example
Replay commands
npm install @typesafe-ai/sdk@0.6.0
# Set TYPESAFE_API_KEY privately in your shell.
node run-choice-sample.mjs samples.json cache-overview
python preview_choice_workflow_v4.py samples.json saved-api-results.json
Scope of this demo

Where does this decision fit?

Useful for a bounded excerpt-depth field

  • Distinguishing concepts, operational steps and concrete mechanisms on the same subject.
  • Versioning an explicit deepest-supported rule for mixed passages.
  • Keeping missing body text separate from a complete concept overview.

Needs additional implementation or validation

  • Ranking complete articles from unread links or promotional titles.
  • Proving technical correctness, complexity or production accuracy.
  • Calling the Score primitive, running pg-jev or writing to PostgreSQL.
Example FAQ

Questions about this original decision workflow.

Depth labels describe the supplied excerpt. They do not certify article quality or execute a database update.

Is this the pg-jev PostgreSQL extension?

No. The source describes pg-jev, but this page makes a separate Choice request for one synthetic article row. No SQL statement runs and no database result is updated.

Does this return a numeric Score?

No. It returns probabilities for four offered Choice labels and selects the highest one. Ordered depth labels are an editorial convention here, not a probability-weighted Score or a ranking of article rows.

Can the label describe an entire linked article?

Only if the relevant article text is actually supplied. This demo receives excerpts and does not open their URLs. Its output is limited to the text in the exact request.

Why keep a row ID with the decision?

A later enrichment job needs to attach the observation to the exact input row. The offline consumer preserves that ID in a pending packet after validating the request hash; it does not perform a database write.

Try a task from your own workflow

Change the supplied task or evidence here, then inspect the live response. Playground supports another explicit decision.