Label-Free Topic-Focused Summarization Using Query Augmentation

TL;DR

Topic-focused summarization often requires large labeled datasets and extensive computation. This work introduces Augmented-Query Summarization (AQS), a label-free pipeline that combines paraphrase generation, question answering, hierarchical clustering, and generic abstractive summarization to produce topic-focused summaries from a query and its context. The authors analyze how query and context variations affect QA transferability and demonstrate a training-free method that adapts to new topics without topic-specific training. On real-world data (Debatepedia, QMSum, and ECF), AQS achieves competitive or superior summary quality with favorable efficiency, highlighting its potential for scalable, cost-effective personalized content extraction in data-rich settings.

Abstract

In today's data and information-rich world, summarization techniques are essential in harnessing vast text to extract key information and enhance decision-making and efficiency. In particular, topic-focused summarization is important due to its ability to tailor content to specific aspects of an extended text. However, this usually requires extensive labelled datasets and considerable computational power. This study introduces a novel method, Augmented-Query Summarization (AQS), for topic-focused summarization without the need for extensive labelled datasets, leveraging query augmentation and hierarchical clustering. This approach facilitates the transferability of machine learning models to the task of summarization, circumventing the need for topic-specific training. Through real-world tests, our method demonstrates the ability to generate relevant and accurate summaries, showing its potential as a cost-effective solution in data-rich environments. This innovation paves the way for broader application and accessibility in the field of topic-focused summarization technology, offering a scalable, efficient method for personalized content extraction.

Figures (2)

• Figure 1: The Augmented-Query Summarization (AQS) pipeline consists of four pretrained key components: a paraphrasing model, a question-answering model, hierarchical clustering, and an abstractive summarization model. AQS takes two text inputs: the query related to the topic and the typically longer context. It generates a single, topic-focused summary as its output. AQS is an adaptation approach, as all the key components can be derived from generic tasks, such as generic abstractive summarization.
• Figure 2: Illustration of the functionality of each part in the proposed method. Paraphrasing generates multiple queries from a single one. Using multiple queries may stabilise the QA model performance. On some inputs, a single query fails to yield a correct answer, while using multiple queries may yield, say 70%, correct answers. When the majority part of answers are correct, then the answer can be considered as correct. Note that a correct answer from QA might also contain irrelevant content, e.g., function words. Clustering significantly reduces redundancy in QA. This is likely because some incorrect answers are pretty long, and thus account for a large proportion of content. Therefore, removing these incorrect answers before summarization is needed. Besides, summarization helps reform a fluent sentence from the clustering-selected answers. Intuitively, the text redundancy in the summaries is even lower.

Theorems & Definitions (3)

• Example 1
• Example 2
• Example 3