Qiudan Li

Combining empirical experimentation and modeling techniques: A design research approach for personalized mobile advertising applications

We propose a design research approach combining behaviour and engineering techniques to better support user modeling in personalized mobile advertising applications. User modeling is a practical means of enabling personalization; one important method to construct user models is that of Bayesian networks. To identify the Bayesian network structure variables and the prior probabilities, empirical experimentation is adopted and context, content, and user preferences are the salient variables. User data collected from the survey are used to set the prior probabilities for the Bayesian network. Experimental evaluation of the system shows it is effective in improving user attitude toward mobile advertising.

Which photo groups should I choose? A comparative study of recommendation algorithms in Flickr

Over the last few years, the social media site Flickr has gained massive popularity. Besides traditional operations on photo sharing, Flickr also offers millions of groups for users to join in order to share photos with others, and the number of groups still increases daily. Choosing among so many options is challenging for users. As such, helping users easily find their desirable groups has become increasingly important. In this paper, we provide a systematic experimental evaluation of several collaborative filtering algorithms to recommend groups for Flickr users. In particular, we design and compare seven Flickr group recommendation models: three memory-based models and four model-based models. Our results suggest that model-based approaches are beneficial compared with memory-based approaches in terms of top-k recommendation metric. Models with tags perform well for sparse data, whereas models without tags are more suitable for dense data. Furthermore, incorporating tags in the recommendation algorithms leads to an improvement of precision on the top 2% performance.Over the last few years, the social media site Flickr has gained massive popularity. Besides traditional operations on photo sharing, Flickr also offers millions of groups for users to join in order to share photos with others, and the number of groups still increases daily. Choosing among so many options is challenging for users. As such, helping users easily find their desirable groups has become increasingly important. In this paper, we provide a systematic experimental evaluation of several collaborative filtering algorithms to recommend groups for Flickr users. In particular, we design and compare seven Flickr group recommendation models: three memory-based models and four model-based models. Our results suggest that model-based approaches are beneficial compared with memory-based approaches in terms of top- k recommendation metric. Models with tags perform well for sparse data, whereas models without tags are more suitable for dense data. Furthermore, incorporating tags in the recommendation algorithms leads to an improvement of precision on the top 2% performance.

Boosting the Performance of Web Spam Detection with Ensemble Under-Sampling Classification

Anti-spam has become one of the top challenges for the Web search. In this paper, we explore the Web spam detection as a binary classification problem. Based on the fact that reputable pages are more easy to be obtained than spam ones on the Web, an ensemble under-sampling classification strategy is adopted, which exploits the information involved in the large number of reputable Websites to full advantage. The strategy is based on the predicted spamicity of every sub-classifiers, in which both content-based and link-based features are taken into account. The experiments on standard WEBSPAM-UK2006 benchmark showed that the ensemble strategy can improve the web spam detection performance effectively.

Improving personalized services in mobile commerce by a novel multicriteria rating approach

With the rapid growth of wireless technologies and mobile devices, there is a great demand for personalized services in m-commerce. Collaborative filtering (CF) is one of successful techniques to produce personalized recommendations for users. This paper proposes a novel approach to improve CF algorithms, where the contextual information of a user and the multicriteria ratings of an item are considered besides the typical information on users and items. The multilinear singular value decomposition (MSVD) technique is utilized to explore both explicit relations and implicit relations among user, item and criterion. We implement the approach in an existing m-commerce platform, and encouraging experimental results demonstrate its effectiveness.

User community discovery from multi-relational networks

Online social network services (SNS) have been experiencing rapid growth in recent years. SNS enable users to identify other users with common interests, exchange their opinions, and establish forums for communication, and so on. Discovering densely connected user communities from social networks has become one of the major challenges, to help understand the structural properties of SNS and improve user-oriented services such as identification of influential users and automated recommendations. Previous work on community discovery has treated user friendship networks and user-generated contents separately. We hypothesize that these two types of information can be fruitfully integrated and propose a unified framework for user community discovery in online social networks. This framework combines the author-topic (AT) model with user friendship network analysis. We empirically show that this approach is capable of discovering interesting user communities using two real-world datasets. Highlights? This paper presents a unified framework for user community detection in social network services. ? The framework integrates the user friendship networks and user-generated contents. ? It can help in understanding the structural properties of online social network services. ? Empirical evaluations on real-world datasets show the efficacy of the method.

Flickr group recommendation based on tensor decomposition

Over the last few years, Flickr has gained massive popularity and groups in Flickr are one of the main ways for photo diffusion. However, the huge volume of groups brings troubles for users to decide which group to choose. In this paper, we propose a tensor decomposition-based group recommendation model to suggest groups to users which can help tackle this problem. The proposed model measures the latent associations between users and groups by considering both semantic tags and social relations. Experimental results show the usefulness of the proposed model.

Mining Fine Grained Opinions by Using Probabilistic Models and Domain Knowledge

The explosive growth of the user-generated content on the Web has offered a rich data source for mining opinions. However, the large number of diverse review sources challenges the individual users and organizations on how to use the opinion information effectively. Therefore, automated opinion mining and summarization techniques have become increasingly important. Different from previous approaches that have mostly treated product feature and opinion extraction as two independent tasks, we merge them together in a unified process by using probabilistic models. Specifically, we treat the problem of product feature and opinion extraction as a sequence labeling task and adopt Conditional Random Fields models to accomplish it. As part of our work, we develop a computational approach to construct domain specific sentiment lexicon by combining semi-structured reviews with general sentiment lexicon, which helps to identify the sentiment orientations of opinions. Experimental results on two real world datasets show that the proposed method is effective.

Link based small sample learning for web spam detection

Robust statistical learning based web spam detection system often requires large amounts of labeled training data. However, labeled samples are more difficult, expensive and time consuming to obtain than unlabeled ones. This paper proposed link based semi-supervised learning algorithms to boost the performance of a classifier, which integrates the traditional Self-training with the topological dependency based link learning. The experiments with a few labeled samples on standard WEBSPAM-UK2006 benchmark showed that the algorithms are effective.

An Examination of Electronic Cigarette Content on Social Media: Analysis of E-Cigarette Flavor Content on Reddit

In recent years, the emerging electronic cigarette (e-cigarette) marketplace has shown great development prospects all over the world. Reddit, one of the most popular forums in the world, has a very large user group and thus great influence. This study aims to gain a systematic understanding of e-cigarette flavors based on data collected from Reddit. Flavor popularity, mixing, characteristics, trends, and brands are analyzed. Fruit flavors were mentioned the most (n = 15,720) among all the posts and were among the most popular flavors (n = 2902) used in mixed blends. Strawberry and vanilla flavors were the most popular for e-juice mixing. The number of posts discussing e-cigarette flavors has increased sharply since 2014. Mt. Baker Vapor and Hangen were the most popular brands discussed among users. Information posted on Reddit about e-cigarette flavors reflected consumers’ interest in a variety of flavors. Our findings suggest that Reddit could be used for data mining and analysis of e-cigarette-related content. Understanding how e-cigarette consumers’ view and utilize flavors within their vaping experience and how producers and marketers use social media to promote flavors and sell products could provide valuable information for regulatory decision-makers.

A Bayesian Network-Based Framework for Personalization in Mobile Commerce Applications

Providing personalized services for mobile commerce (m-commerce) can improve user satisfaction and merchant profits, which are important to the success of m-commerce. This paper proposes a Bayesian network (BN)-based framework for personalization in m-commerce applications. The framework helps to identify the target mobile users and to deliver relevant information to them at the right time and in the right way. Under the framework, a personalization model is generated using a new method and the model is implemented in an m-commerce application for the food industry. The new method is based on function dependencies of a relational database and rough set operations. The framework can be applied to other industries such as movies, CDs, books, hotel booking, flight booking, and all manner of shopping settings.