Hour 2 · Bibliometrics · Step 13 of 1587%

AI-assisted cluster interpretation

Hand it the cluster — then catch it being wrong.

~15 min

Drop the cluster's keywords into an AI chat tool. Ask for candidate labels and a first-draft narrative of what ties them together. The labels are suggestions, not answers. A model will name a cluster confidently and miss the point. You own the interpretation.

AI prompts (2)

Prompt

Cluster labeller

When: VOSviewer/Biblioshiny gave you a cluster — you want naming options.

You are helping me label a cluster from a keyword co-occurrence map.

Cluster keywords (with weight or frequency if available):
<PASTE — e.g. online learning (42), engagement (31), MOOC (18), flipped classroom (15), self-regulation (12)>

Field context: <e.g. higher education research, 2018–2024>
Other clusters in the same map (just names, for contrast): <PASTE>

Please:
1. Propose 3 candidate labels at different levels of abstraction (narrow / mid / broad).
2. For each label, list which keywords it covers well and which it leaves out.
3. Flag the single keyword most likely to belong to a different cluster.
4. Recommend one label, with a one-sentence justification.

Do not invent keywords that are not in the list.

Prompt

Cluster narrative drafter

When: You've picked a label and want a first-draft paragraph for the results section.

Draft a short paragraph (≈120 words) describing this cluster for the results section of a bibliometric paper.

Cluster label: <PASTE>
Keywords with weights: <PASTE>
Representative papers (title + year + 1-line finding): <PASTE 3–5>
Time trend (rising/stable/declining): <state if known>

Style: formal academic English, present tense, third person. Avoid hype words ("revolutionary", "cutting-edge"). End with one sentence on what the cluster suggests about the field's trajectory.

Check every cited paper actually belongs to the cluster, and that the trend statement matches your data. Models smooth over inconvenient facts.