DATA ANALYSIS · NLP · TOPIC MODELING

YouTube AI
Education Analysis

Mapping Indonesian public discussion, topics, and personas around AI in education.

An NLP analysis of 4,281 YouTube comments from five videos, combining text preprocessing, feature extraction, topic modeling, and clustering to surface meaningful discussion patterns.

ROLEData Analyst · NLP Researcher

RESPONSIBILITIESData Pipeline · Topic Modeling · Persona Clustering

TECH STACKPython · BERTopic · LDA · KMeans · NLTK

DATASET & CORPUS4,281 Comments · 5 YouTube Videos

COMMENT SEMANTIC SPACE4,281 signals mapped

manusiateknologianakgenerasiprofindonesiabelajarkerjaAI

● Human roles● AI optimism● Speaker response

01 / PROJECT OVERVIEW

Listening to a public conversation about AI and education.

This project turns unstructured Indonesian YouTube comments into a readable map of attitudes, themes, and audience personas around AI in education.

Topic modeling and clustering distinguish discussions about human roles, optimism, and responses to speakers.

4,281COMMENTS ANALYZED

5YOUTUBE VIDEOS

3TOPICS / PERSONAS

02 / DISCUSSION PERSONAS

Three recurring conversation patterns.

ETIKA & PERAN MANUSIA

37.18%manusia · kerja · buat

Distinct discussion pattern identified through comment clustering and topic interpretation.

OPTIMIS TERHADAP AI

38.65%gen · aku · anak

Distinct discussion pattern identified through comment clustering and topic interpretation.

KOMENTAR TERHADAP NARASUMBER

24.17%indonesia · prof · stella

Distinct discussion pattern identified through comment clustering and topic interpretation.

03 / ANALYSIS ARTIFACTS

From language signals to cluster structure.

Word cloud of Indonesian YouTube comments after preprocessing — WORD FREQUENCY / POST-PROCESSING

CLUSTER DISTRIBUTION · K=10

Cluster structure reveals a large volume of general video responses alongside more specific discussion themes.

04 / METHODOLOGY

A reproducible discussion-analysis pipeline.

01COMMENT COLLECTION

02PREPROCESSING

03FEATURE EXTRACTION

04TOPIC MODELING

05CLUSTERING

06PERSONA IDENTIFICATION

07VISUALIZATION

05 / KEY INSIGHTS

What the discussion reveals.

Human roles stay central

Many comments frame AI through human judgement, responsibility, teaching, and work.

Optimism is practical

Positive discussion often treats AI as a useful aid for learning and productivity.

Context shapes discussion

Responses to speakers form a distinct pattern alongside the AI topic itself.

PythonPandasNumPyNLTKTF-IDFSentence TransformersLDABERTopicUMAPHDBSCANKMeansDBSCANMatplotlibSeabornWordCloudpyLDAvis

Explore the YouTube Analysis Repository

View the comment analysis, clustering pipeline, persona identification, and visualization workflow.

GITHUB REPOSITORY