Description

When used correctly, LLMs have been a blessing for UFO Data Science, and when given the right data and prompts. See prior work posted to this sub by VerbaGPT (https://old.reddit.com/r/UFOs/search?q=author%3APristine_Ad3669&restr…), deckerRTM (https://old.reddit.com/r/UFOs/comments/1qw4kzm/finding_patterns_in_15…), and Pristine\_Ad3669 (https://old.reddit.com/r/UFOs/comments/1rc0jdq/tier_1_ufo_color_perce…).

Last week, I switched from ChatGPT to Claude Code and the difference has been night and day (and not just because Claude generated the graphs in dark mode!)

I used up-to-date data from just MUFON and NUFORC, which I can't share here because it's proprietary. But don't worry, I did open-source a project with 331K reports up to 2023 (https://github.com/thomhastings/updb-analysis). It uses the old UPDB.app PostgreSQL database, available on the Wayback Machine (https://web.archive.org/web/20230824194810/https://updb.app/phenomeno…), with 12 sources including MUFON and NUFORC.

Here's the initial prompt I used for my analysis:

Use Natural Language Processing and sentiment analysis to generate insights about what emotions people feel when they see a UFO, by analyzing the CSV datasets in this directory.

Below is the generated output.

UFO Witness Emotion Analysis — NLP & Sentiment Report

Dataset: MUFON + NUFORC | Reports analysed: 288,037 | Generated: 2026-02-22

Dashboard

UFO Emotion Dashboard

Overall Sentiment Distribution

|Sentiment|Reports|Share|
|:-|:-|:-|
|Positive|180,459|62.7%|
|Negative|81,758|28.4%|
|Neutral|25,820|9.0%|

Mean VADER compound score: 0.2693  *(range: −1 = most negative, +1 = most positive)*

Emotion Category Prevalence

*% of reports containing at least one emotion-matched word*

|Emotion|Reports|Prevalence|
|:-|:-|:-|
|Curiosity|116,749|40.5%|
|Calm / Peaceful|71,279|24.7%|
|Confusion / Disbelief|64,579|22.4%|
|Unease / Paranoia|55,824|19.4%|
|Shock / Surprise|51,266|17.8%|
|Awe / Wonder|31,511|10.9%|
|Fear / Terror|25,306|8.8%|
|Excitement / Elation|10,934|3.8%|

Emotion Word Share

*Share of all matched emotion tokens*

|Emotion|Tokens|Share|
|:-|:-|:-|
|Curiosity|207,935|32.0%|
|Calm / Peaceful|108,504|16.7%|
|Confusion / Disbelief|96,619|14.9%|
|Unease / Paranoia|74,616|11.5%|
|Shock / Surprise|70,541|10.9%|
|Awe / Wonder|41,232|6.3%|
|Fear / Terror|37,971|5.8%|
|Excitement / Elation|12,662|1.9%|

Mean VADER Compound Score by Source

|Source|Mean Compound|
|:-|:-|
|MUFON|0.2744|
|NUFORC|0.2649|

Emotion ↔ Sentiment Correlation

*Pearson r of each emotion count vs. VADER compound score*

|Emotion|r|Direction|
|:-|:-|:-|
|Fear / Terror|−0.068|↓ negative|
|Confusion / Disbelief|\+0.011|↑ positive|
|Unease / Paranoia|\+0.056|↑ positive|
|Shock / Surprise|\+0.057|↑ positive|
|Calm / Peaceful|\+0.084|↑ positive|
|Curiosity|\+0.107|↑ positive|
|Excitement / Elation|\+0.109|↑ positive|
|Awe / Wonder|\+0.139|↑ positive|

Dominant Emotion per Report

|Emotion|Reports|Share|
|:-|:-|:-|
|Curiosity|78,052|40.3%|
|Confusion / Disbelief|38,551|19.9%|
|Calm / Peaceful|27,454|14.2%|
|Awe / Wonder|15,816|8.2%|
|Shock / Surprise|15,330|7.9%|
|Fear / Terror|13,891|7.2%|
|Excitement / Elation|2,951|1.5%|
|Unease / Paranoia|1,457|0.8%|

Word Cloud

UFO Witness Vocabulary Word Cloud

Sentiment Score Distributions by Emotion

VADER Sentiment KDE by Emotion

Emotion Radar — Overall vs MUFON vs NUFORC

Emotion Prevalence Radar Chart

Key Insights

1. Curiosity & Awe Dominate — "Curiosity" appears in 40.5% of all reports, closely followed by "Calm / Peaceful" at 24.7%. Witnessing a UFO primarily triggers a drive to observe and make sense of what is seen.
2. Mixed Sentiment — Wonder Edges Out Fear — Positive: 62.7% | Neutral: 9.0% | Negative: 28.4%. Despite the strangeness of sightings, more witnesses describe the experience in positive or neutral terms than in fearful or negative ones.
3. Dominant Single Emotion — When one emotion dominates a report, it is most commonly "Curiosity", followed by "Confusion / Disbelief". This suggests the first reaction is usually intellectual (trying to identify the object) before emotional.
4. Fear Correlates Most Negatively with Sentiment — Reports heavy in fear-language score lowest on VADER, confirming that fear-laden accounts use strongly negative vocabulary (terrified, horror, panic).
5. Awe and Calm Correlate Positively with Sentiment — Reports describing beauty, silence, or transcendence use more positive language, suggesting a subset of witnesses find the experience spiritually uplifting.
6. MUFON vs NUFORC Tone — MUFON reports (mean 0.2744) tend to be slightly more positive than NUFORC reports (0.2649), likely reflecting differences in reporting style (investigator-mediated vs. self-submitted).