Unsupervised Extractive Summarization for Product Description Using Coverage Maximization with Attribute Concept

摘要

This paper presents a system that generates unsupervised extractive summarization of product descriptions. The system utilizes a coverage maximization model with concepts introduced as a practical solution for extractive summarization. Product descriptions present more challenges to this task, including but not limited to, shorter paragraphs and infrequent concepts distributed across the paragraph. To overcome these challenges, we propose domain-specific concepts, namely product attributes, in addition to general linguistic concepts such as named entities and syntactic dependency. The result demonstrates that product attributes can help generate better summarization for product descriptions.