Research map / Large language models / AI and society
Word2Vec4Kids: Interactive Challenges to Introduce Middle School Students to Word Embeddings
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Presents Word2Vec4Kids, a native macOS interactive app teaching middle schoolers word embeddings via Word2Vec exploration games, and reports engagement and post-survey results from 38 students.
Problem
Middle school students lack accessible, engaging tools for learning AI/NLP representation concepts such as word embeddings, which matter for AI literacy.
Approach
A native macOS educational app built on a slimmed 200-dimensional Word2Vec vocabulary (70k words, offensive words removed), with a custom Swift nearest-neighbor engine validated against gensim, a PCA-based 2D visualization (including an emotion-biased projection), and game modes (Neighbor Navigator, Word Quest, Word Alchemy) for hands-on exploration.
Key results
- 44.5% of 38 middle school students achieved a perfect 4/4 score on the multiple-choice post-survey concept questions.
- 87% of students correctly placed 'cow' in the bottom-right quadrant of the emotion-biased PCA visualization.
- 75% of students attempted at least one Word Quest challenge, the most-played game mode.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use Word2Vec4Kids as an interactive introduction to word embeddings | Middle school students · Post-survey concept questions | perfect score rate: 44.5% | Helped |
| Use Word2Vec4Kids to teach word-placement concepts | Middle school students · Post-survey word-placement question | accuracy: 87% | Helped |
| Use Word2Vec4Kids to teach benefits of vector embeddings | Middle school students · Post-survey vector-embedding benefits question | correct response rate: 42% | Mixed |
| Use the interactive Word Quest game mode | Middle school students · Word Quest | students attempting at least one challenge: 75% | Helped |
| Use Word2Vec4Kids across grade levels | Middle school students · Post-survey concept questions | score | No clear effect |
| Increase time spent using the application | Word2Vec4Kids · Post-survey concept questions | score correlation with application usage | No clear effect |
Surprising findings
- No significant difference in post-survey performance across grades 6-8, and no strong correlation between app usage and scores.
- Off-the-shelf Apple MLWordEmbedding gave questionable nearest neighbors, so the authors wrote a custom Swift package to match gensim output.
Limitations
- Small, self-selected sample of 38 students from one STEM-focused charter school, with a majority planning STEM majors.
- Limited session time (5-10 minutes) prevented full exploration; open-ended responses had high non-response rates (55%).
- Pre-survey did not include items directly tied to the app, so no pre/post learning gain was measured.
Experimental setup
Datasets: Word2Vec pretrained vectors (slimmed 70,000-word, 200-dim set) CMU offensive/profane word list
Metrics: post-survey multiple-choice score distribution challenge completion engagement observations
Baselines: gensim Bandyopadhyay et al. online word embedding demos
Builds on: Word2Vec gensim NLP4All
Closest papers by meaning
- From Consumers to Critical Users: Prompty, an AI Literacy Tool for High School Students (AAAI 2024 · similarity 0.87)
- Supporting Upper Elementary Students in Learning AI Concepts with Story-Driven Game-Based Learning (AAAI 2024 · similarity 0.86)
- An XAI Social Media Platform for Teaching K-12 Students AI-Driven Profiling, Clustering, and Engagement-Based Recommending (AAAI 2025 · similarity 0.85)
- Unplugged K-12 AI Learning: Exploring Representation and Reasoning with a Facial Recognition Game (AAAI 2024 · similarity 0.85)
- Shaping AI Interest in Rural Middle Schools with Unplugged Learning: Gender Differences and Teacher Insights (AAAI 2025 · similarity 0.85)
- An Effectiveness Study of Teacher-Led AI Literacy Curriculum in K-12 Classrooms (AAAI 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.