Retrieval and embeddings · Lesson 8 of 16 · 8 min

Compare similarity scores

Similarity scores are ranking signals under a specific embedding and metric setup. They are not calibrated probabilities of truth or correctness.

Open the lab Course map

Mastery 0/3Run itExplain itApply it

learner@clairos:~ $

↑ history · Tab complete · Ctrl+C cancel · Esc, Tab to leaveExit 0

Your objectives

Hints are always available. Wrong turns never cost XP.

1
Read the case evidence
Read /opt/ai-lab/cases/aiops-retrieval-embeddings-008.json. Separate the recorded facts from your assumptions.
Hint & worked example
Similarity scores are ranking signals under a specific embedding and metric setup. They are not calibrated probabilities of truth or correctness.
The complete case JSON is read successfully.
```
cat /opt/ai-lab/cases/aiops-retrieval-embeddings-008.json
```
Common mistake to avoid: Convert cosine similarity directly into factual confidence.
2
Isolate the deciding signal
Select the truth_probability field from the case JSON. What does it establish—and what does it not establish?
Hint & worked example
Treat 0.86 as a similarity signal, not an 86 percent probability that the answer is true.
A successful JSON query isolates truth_probability.
```
jq .truth_probability /opt/ai-lab/cases/aiops-retrieval-embeddings-008.json
```
Common mistake to avoid: Using a different case or an unrelated signal as the justification.
3
Record your operational decision
Choose the justified code: treat-as-ranking-signal, change-without-evidence, or ignore-the-signal. Save that one code followed by a newline in /home/learner/lab/aiops-retrieval-embeddings-008-decision.txt.
Hint & worked example
Treat 0.86 as a similarity signal, not an 86 percent probability that the answer is true.
The decision artifact contains the correct code, not just a matching command in history.
```
printf '%s\n' 'treat-as-ranking-signal' > /home/learner/lab/aiops-retrieval-embeddings-008-decision.txt
```
Common mistake to avoid: Convert cosine similarity directly into factual confidence.

Understand the pattern

Similarity scores are ranking signals under a specific embedding and metric setup. They are not calibrated probabilities of truth or correctness.

Interpret the evidence

Treat 0.86 as a similarity signal, not an 86 percent probability that the answer is true.

A tempting mistake

Convert cosine similarity directly into factual confidence.

All names, measurements, prices, versions, and thresholds in this exercise are fictional training data. The evidence supports this bounded scenario, not a claim about a live provider.

Worked example and expected result

$ cat /opt/ai-lab/cases/aiops-retrieval-embeddings-008.json

{
  "fixture": "Fictional training data; not a live measurement.",
  "score": 0.86,
  "metric": "cosine",
  "truth_probability": null
}

Exit status: 0. The command completed successfully in the training fixture.

$ jq .truth_probability /opt/ai-lab/cases/aiops-retrieval-embeddings-008.json

null

Exit status: 0. The command completed successfully in the training fixture.

$ printf '%s\n' 'treat-as-ranking-signal' > /home/learner/lab/aiops-retrieval-embeddings-008-decision.txt

(no output; exit 0)

Exit status: 0. The command completed successfully in the training fixture.

Try a variation

Name the assumption that would have to change for “treat-as-ranking-signal” to stop being the justified action. Do not silently replace missing evidence with a guess.

Explain it

What does the evidence in “Compare similarity scores” justify?Make the change now and gather the missing evidence afterward.Treat 0.86 as a similarity signal, not an 86 percent probability that the answer is true.Convert cosine similarity directly into factual confidence.

Apply it elsewhere

A colleague concludes: “Convert cosine similarity directly into factual confidence.” What should you do?Reject that conclusion; return to the case evidence and its stated limits.Accept it because the training command completed.Treat this fixture as a live measurement.

Keep the momentum.

Complete the lab, then explain and apply the pattern.

Next: Spot a missing citation Try an incident drill

Anonymous lesson and mastery progress lasts for this session. Sign in to keep synced lesson XP between sessions; mastery stays device-local. No speed bonus, lost lives, or hint penalties.

Keep learning

Case methodology: evidence, limits, decision