University of Twente Student Theses

Login
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.

Keyword-guided structured abstract generation for deep learning papers using ChatGPT-4o

Medina Gimenez, Celia (2025) Keyword-guided structured abstract generation for deep learning papers using ChatGPT-4o.

[img] PDF
685kB
Abstract:The exponential growth of deep learning research has made it difficult for publishers and readers to write and evaluate scientific papers in an efficient way. While automatic summarisation tools exist, they normally generate brief, unstructured outputs and lack transparency or content validation mechanisms. This project introduces a framework that generates IMRAD-structured abstracts using OpenAI’s GPT-4o, guided by extracted keywords and evaluated through both automatic and Large Language Model (LLM) based methods. The system incorporates a keyword validation loop that enforces the inclusion of the most important concepts and iteratively improves abstract quality. Evaluation is performed using semantic similarity metrics, natural language inference (NLI), and judgment from two independent LLMs (Gemini and Claude), each rating factual accuracy, clarity, completeness, and keyword relevance. Results show that the proposed framework improves factual consistency, coverage, and semantic alignment over a simple prompt baseline, though may introduce trade-offs in clarity. These findings demonstrate the value of structured prompting, and keyword feedback in scientific summarisation.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Business & IT BSc (56066)
Link to this item:https://purl.utwente.nl/essays/107569
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page