[Paper Review] Leveraging BERT for Extractive Text Summarization on Lectures
The paper presents Lecture Summarization Service, a Python RESTful system that uses BERT embeddings and KMeans clustering to extract key sentences from lecture transcripts for user-specified summary lengths.
In the last two decades, automatic extractive text summarization on lectures has demonstrated to be a useful tool for collecting key phrases and sentences that best represent the content. However, many current approaches utilize dated approaches, producing sub-par outputs or requiring several hours of manual tuning to produce meaningful results. Recently, new machine learning architectures have provided mechanisms for extractive summarization through the clustering of output embeddings from deep learning models. This paper reports on the project called Lecture Summarization Service, a python based RESTful service that utilizes the BERT model for text embeddings and KMeans clustering to identify sentences closes to the centroid for summary selection. The purpose of the service was to provide students a utility that could summarize lecture content, based on their desired number of sentences. On top of the summary work, the service also includes lecture and summary management, storing content on the cloud which can be used for collaboration. While the results of utilizing BERT for extractive summarization were promising, there were still areas where the model struggled, providing feature research opportunities for further improvement.
Motivation & Objective
- Motivate and enable automated extractive summarization of lecture content.
- Leverage BERT to produce meaningful sentence embeddings for clustering-based summarization.
- Provide a scalable, cloud-based service for lecture summarization and content management.
Proposed method
- Compute text embeddings with the BERT model to represent sentences.
- Apply KMeans clustering on sentence embeddings to form clusters.
- Select sentences closest to the cluster centroids to form the summary.
- Offer a Python-based RESTful service to generate summaries of a user-defined length.
- Include features for lecture and summary management with cloud storage for collaboration.
Experimental results
Research questions
- RQ1Can BERT-based sentence embeddings improve the quality of extractive summaries for lectures?
- RQ2Is clustering-based selection (sentences near centroids) effective for producing concise, representative summaries of lecture content?
- RQ3Can a cloud-backed RESTful service reasonably support user-defined summary lengths and collaborative management?
Key findings
- BERT embeddings + KMeans clustering can identify representative sentences for summaries.
- The service provides a practical workflow for generating summaries at a user-defined length.
- The approach shows promising results but also presents areas where the model struggles and opportunities for improvement.
- The system includes lecture and summary management with cloud storage to support collaboration.
Read this paper
Open the full text in the Nubint reader and ask about any passage as you read.
Read this paper on NubintThis review was generated by AI from the original paper. Please check the paper before citing it.