Automatically extracting user reviews from forum sites

User reviews in forum sites are the important information source for many popular applications (e.g., monitoring and analysis of public opinion), which are usually represented in form of structured records. To the best of our knowledge, little existing work reported in the literature has systemically investigated the problem of extracting user reviews from forum sites. Besides the variety of web page templates, user-generated reviews raise two new challenges. First, the inconsistency of review contents in terms of both the document object model (DOM) tree and visual appearance impair the similarity between review records; second, the review content in a review record corresponds to complicated subtrees rather than single nodes in the DOM tree. To tackle these challenges, we present WeRE — a system that performs automatic user review extraction by employing sophisticated techniques. The review records are extracted from web pages based on the proposed level-weighted tree similarity algorithm first, and then the review contents in records are extracted exactly by measuring the node consistency. Our experimental results based on 20 forum sites indicate that WeRE can achieve high extraction accuracy.     

Six years of systematic literature reviews in software engineering: An updated tertiary study

Context

Since the introduction of evidence-based software engineering in 2004, systematic literature review (SLR) has been increasingly used as a method for conducting secondary studies in software engineering. Two tertiary studies, published in 2009 and 2010, identified and analysed 54 SLRs published in journals and conferences in the period between 1st January 2004 and 30th June 2008.

Objective

In this article, our goal was to extend and update the two previous tertiary studies to cover the period between 1st July 2008 and 31st December 2009. We analysed the quality, coverage of software engineering topics, and potential impact of published SLRs for education and practice.

Method

We performed automatic and manual searches for SLRs published in journals and conference proceedings, analysed the relevant studies, and compared and integrated our findings with the two previous tertiary studies.

Results

We found 67 new SLRs addressing 24 software engineering topics. Among these studies, 15 were considered relevant to the undergraduate educational curriculum, and 40 appeared of possible interest to practitioners. We found that the number of SLRs in software engineering is increasing, the overall quality of the studies is improving, and the number of researchers and research organisations worldwide that are conducting SLRs is also increasing and spreading.

Conclusion

Our findings suggest that the software engineering research community is starting to adopt SLRs consistently as a research method. However, the majority of the SLRs did not evaluate the quality of primary studies and fail to provide guidelines for practitioners, thus decreasing their potential impact on software engineering practice.     

基于LDA模型的餐馆评论排序

在餐馆评论中,存在评论文本未明确指出评价等级及评论文本不一致等问题。为此,提出一种基于LDA模型的餐馆评论排序方法。利用LDA模型对评论文本进行主题抽取,过滤掉不相关评论,基于过滤后的用户评论和用户给出的评论等级计算餐馆评论若干方面的得分,在该得分的基础上,利用逻辑回归进行训练,得到餐馆评论排序模型。实验结果表明,该方法的排序效果较优。     

融合评论文本和图像语义特征的兴趣点推荐

用户-兴趣点签到数据的高度稀疏性让传统的推荐算法的推荐效果大打折扣。基于此,提出评论文本和图像语义信息融合的兴趣点推荐新算法。该算法同时考虑用户评论对评分数据的可解释性和图像语义信息对兴趣点外观的描述性,充分利用评论文本和图像数据辅助用户偏好特征和兴趣点属性特征的学习。使用神经网络抽取与用户和兴趣点相关的评论文本和图像语义特征,分别建模用户-文本语义特征关系、兴趣点-图像语义特征关系,将两种关系与用户-兴趣点评分矩阵进行融合,基于概率矩阵分解构建统一的推荐模型。在Yelp数据集上实验表明,该算法有效地缓解了签到数据稀疏性带来的推荐准确性问题,在MAE和RMSE两项指标上均优于主流方法。     

Why provide an online review? An extended theory of planned behavior and the role of Big-Five personality traits

Online review, an important form of reputation systems, has been studied intensively because of its powerful impact on online retailers, intermediaries, and customers. However, to date, very little attention has been paid to factors that influence an individual’s intention to provide an online review. An extended theory of planned behavior and Big-Five personality framework are used in this study. We empirically examine our model by using a cross-sectional survey study, collecting data from a sample of 171 online shoppers. Results show that attitude, perceived pressure, neuroticism, and conscientiousness are significant predictors of an individual’s intention to provide an online review. Findings may help online retailers and/or intermediaries increase the number of online reviews provided, which will lead to more accurate rating information about transactions, products, or services and may serve as a stepping-stone to continuous improvements. Implications, limitations, and future research directions are discussed.     

基于情感极性与SMOTE过采样的虚假评论识别方法*

为了有效识别商品虚假评论,提出一种基于情感极性与SMOTE过采样的虚假评论识别方法。首先,根据在线虚假评论的特点,构建一个多维虚假评论特征模型。其次,在情感极性算法中增加了情感极性均值和情感极性标准差等统计指标来全面刻画虚假评论。最后,针对虚假评论中的类不平衡问题,使用SMOTE算法优化随机森林分类模型,从而提高虚假评论识别效果。基于大众点评网的真实评论数据进行了多组实验,实验结果表明该方法在正负样本不平衡的虚假评论数据集中具有更高的准确率、召回率及F值。综合考虑情感极性和正负样本不平衡等因素可帮助电商平台有效过滤虚假评论,为消费者提供更加真实可靠的评论数据。     

Expert Recommendations Based on Opinion Mining of User‐Generated Product Reviews

In this article, we introduce the idea of expert recommendations whose objective is to relate review comments with users’ tasks or expectations. We propose to use fine‐grained information such as opinions and suggestions extracted using natural language processing techniques from user reviews about products, to improve a recommendation system. While typical recommender systems compare a user profile with some reference characteristics to rate unseen items, they rarely make use of the content of reviews that users have provided on a given product. In this article, we present the application of an opinion extraction system to extract opinions and suggestions from the content of the reviews, the use of the results to compare other products with the reviewed one, and eventually the recommendation of better products to the user. The recommendations are given a confidence weight by using a trust social network.     

Enhancing a Newspaper Collection Using the World Wide Web

Abstract

Serial cancellations have reduced the print newspaper collection at the University of Toledo (UT) Library by about one third since 1990. In the meantime, the number of newspapers publishing on the World Wide Web has skyrocketed, prompting UT to investigate accessing newspapers online. The immediate goal of this project is to provide library users with access to online versions of cancelled print newspapers, and eventually to enhance the collection even further with additional electronic newspapers. This article examines the trends in online newspaper publishing, researches the availability of Web sites in relation to the UT newspaper collection, critically reviews the sites found, and discusses the issues surrounding access to an online collection.     

A transformer-based approach for improving app review response generation

Mobile apps are becoming an integral part of people's daily life by providing various functionalities, such as messaging and gaming. App developers try their best to ensure user experience during app development and maintenance to improve the rating of their apps on app platforms and attract more user downloads. Previous studies indicated that responding to users' reviews tends to change their attitude towards the apps positively. Users who have been replied are more likely to update the given ratings. However, reading and responding to every user review is not an easy task for developers since it is common for popular apps to receive tons of reviews every day. Thus, automation tools for review replying are needed. To address the need above, the paper introduces a Transformer-based approach, named TRRGen, to automatically generate responses to given user reviews. TRRGen extracts apps' categories, rating, and review text as the input features. By adapting a Transformer-based model, TRRGen can generate appropriate replies for new reviews. Comprehensive experiments and analysis on the real-world datasets indicate that the proposed approach can generate high-quality replies for users' reviews and significantly outperform current state-of-art approaches on the task. The manual validation results on the generated replies further demonstrate the effectiveness of the proposed approach.     

基于TRF-IM模型的个性化酒店评论摘要生成

针对传统的酒店评论摘要生成模型在生成摘要过程中存在对评论的上下文理解不够充分、并行能力不足和长距离文本依赖缺陷的问题,提出了一种基于TRF-IM(improved mask for transformer)模型的个性化酒店评论摘要生成方法。该方法利用Transformer译码器结构对评论摘要任务进行建模,通过改进其结构中的掩码方式,使得源评论内容都能够更好地学习到上下文语义信息;同时引入了用户类型的个性化词特征信息,使其生成高质量且满足用户需求的个性化酒店评论摘要。实验结果表明,该模型相比传统模型在ROUGE指标上取得了更高的分数,生成了高质量的个性化酒店评论摘要。