Group project

This page presents the topics and summaries of each group project from the course. Each group explores a unique research question or application related to NLP!

Project timeline

1. Research proposal: Teams define their topic, research question, and proposed methods (1st R: 10/19; 2nd R: 11/7)
2. Background research: Literature review, data collection, and preparation of analytical framework (11/6 - 11/18)
3. Final presentation: In-class presentation of findings, methodology, and implications (11/20 - 12/4)
4. Final paper: Full written report summarizing the study, analysis, and conclusions (12/11)

Notes: Background researh/Final presentation slides will be updated during each phrase

Overview of group projects

🔴 Group 1: Adapting visual LLMs for gameplay in Pokémon FireRed

• Members: Leo, Issac, Erica
• Research question: How can a visual large language model (VLLM) be adapted to interact with and successfully play Pokemon Fire Red?
• Keywords: reinforce learning, machine learning, Pokémon, decision making, VLLM
• Background research presentation slides
• Final presentation slides

🔴 Group 2: Generating math learning materials with LLMs

• Members: Conrad, Noah
• Research question: How effectively can large language models (LLMs) generate accurate and course-aligned learning materials for mathematics education?
• Keywords: education, learning materials, study assistance
• Background research presentation slides
• Final presentation slides

🔴 Group 3: Hallucinations in LLMs

• Members: Ashton, Fariha
• Research question: How do current benchmarks differ in evaluating hallucinations in LLM-based reading comprehension, and what gaps or inconsistencies affect their interpretation?
• Keywords: hallucinations, reading comprehension, benchmark
• Background research presentation slides
• Final presentation slides

🔴 Group 4: Location detection from unstructured chat messages

• Members: Natalie, Olivia
• Research question: How accurately can natural language processing models identify location entities in uncleaned or informal text such as chat messages?
• Keywords: NER, location extraction, noisy text
• Background research presentation slides
• Final presentation slides

🔴 Group 5: Predicting age from social media language

• Members: Angel, Eliana, Max
• Research question: What linguistic features can be leveraged to predict a writer’s age or age range in an age prediction NLP task
• Keywords: age prediction, text classification, social media
• Background research presentation slides
• Final presentation slides

🔴 Group 6: Leveraging linguistic structure for low-resource language modeling

• Members: Alex, Christopher
• Research question: How does incorporating morphosyntactic information (e.g., from CoNLL-U formatted data) influence model accuracy when training NLP systems for low-resource languages?
• Keywords: low-resource languages, morphosyntax, POS tagging
• Background research presentation slides
• Final presentation slides

🔴 Group 7: Identifying idioms in English text

• Members: Dan, Jacob
• Research question: Given a dataset containing idiomatic expressions in context, how can a large language model or other NLP system accurately identify, classify, and label idioms within text?
• Keywords: idiom, information extraction, classification, span identification
• Background research presentation slides
• Final presentation slides

🔴 Group 8: Applying NLP-based modeling techniques for musical feature recognition

• Members: Mildness, Shaun
• Research question: How effectively can NLP–based models identify and classify key musical characteristics such as pitch, duration, and mode from audio or symbolic input?
• Keywords: music information retrieval, Audio Processing, Feature Extraction
• Background research presentation slides
• Final presentation slides

🔴 Group 9: Analyzing data science discussions on Stack Overflow

• Members: Atharva
• Research question: Can attention-based models outperform feature-based methods in detecting low-quality posts by identifying quality-relevant text segments?
• Keywords: text quality detection, low-quality posts, quality-signaling segments
• Background research presentation slides
• Final presentation slides

Presentation schedule

Presentation guidelines

Download PDF  |  Open in new tab