A Template Approach for Summarizing Restaurant Reviews

Abstract

In the era of rapid development of social networks, user reviews of restaurant review websites have grown rapidly. In order to allow users to quickly grasp the key points of review information on review sites, this paper provides an abstractive multi-text summary method that can automatically generate template-based review summaries based on predefined topics and sentiments. In particular, for each predefined topic and each type of sentiment (positive or negative), this study uses the TextRank algorithm to find the most representative sentences to form a summary. This method allows users to quickly grasp the positive and negative opinions of each important aspect of the restaurant. The previous research on generating abstracts from reviews either did not generate abstracts based on topics, or based on topics generated by random models. However, the latter method cannot guarantee that the topics generated by the random model are really the topics that the user needs. But for a restaurant review, some topics are very important and indispensable. In order to ensure that abstracts can be generated for these essential topics, our method predefines the topics that must be generated, and then generates abstracts for these topics. In the evaluation, this study compared the template method with the Refresh and Gensim systems based on criteria such informativeness, clarity, usefulness and likes. The results show that the method proposed in this paper is superior to the other two summary methods.

Improved Graph-Based Arabic Hotel Review Summarization Using Polarity Classification

“…To evaluate the summaries, we adopted the BLEU and ROUGE metrics. Although these metrics are widely used for summarization tasks, evaluating summaries needs some kind of human judgment, as done in [3,34]. Moreover, when we analyzed how these algorithms work, we found that they have some limitations, as discussed in Section 4.5.…”

“…Review datasets should also include human summaries, preferably by more than one native Arabic speaker. Nevertheless, other qualitative methods should be considered for evaluating summaries, such as those suggested in [3,35]. Different approaches, including deep learning, could be investigated and compared with graph-based methods.…”

“…Additionally, more focus is needed on multi-document summarization, as few studies have been conducted in this area, especially for the Arabic language. Additionally, review summarization could benefit from adding an aspect-based categorization [3]. Finally, combining text summarization with more text preprocessing, such as correcting spelling and eliminating duplicates, would improve the performance and summary readability.…”

“…ATS aims to produce a smooth and short summary that includes important information from the original text [2]. Several studies addressed ATS for English documents [3,4]; however, fewer studies are available for the Arabic language due to its rich morphological structure, the range of dialects, and the scarcity of data and tools [5]. Arabic is the fifth most spoken language globally, with more than 400 million speakers worldwide [6].…”

“…Reviews usually include positive, negative, and neutral sentiments [3]; we hypothesize that involving sentiment in the summarization process could enhance the performance of the ATS system and improve the readability of the generated summary. This study proposes an approach for summarizing Arabic hotel reviews based on review polarity.…”