The explosive popularity of microblogging services produce a large volume of microblogging messages. It presents great difficulties for a user to quickly gauge his/her followees' opinions when the user interface is overwhelmed by a large number of messages. Useful information is buried in disorganized, incomplete, and unstructured text messages. We propose to organize the large amount of messages into clusters with meaningful cluster labels, thus provide an overview of the content to fulfill users' information needs. Clustering and labeling of microblogging messages are challenging because that the length of the messages are much shorter than conventional text documents. They usually cannot provide sufficient term co-occurrence information for capturing their semantic associations. As a result, traditional text representation models tend to yield unsatisfactory performance. In this paper, we present a text representation framework by harnessing the power of semantic knowledge bases, i.e., Wikipedia and Wordnet. The originally uncorrelated texts are connected with the semantic representation, thus it enhances the performance of short text clustering and labeling. The experimental results on Twitter and Facebook datasets demonstrate the superior performance of our framework in handling noisy and short microblogging messages.

People use various social media sites for different purposes. The information on each site is often partial. When sources of complementary information are integrated, a better profile of a user can be built. This profile can help improve online services such as advertising across sites. To integrate these sources of information, it is necessary to identify individuals across social media sites. This paper aims to address the cross-media user identification problem. We provide evidence on the existence of a mapping among identities of individuals across social media sites, study the feasibility of finding this mapping, and illustrate and develop means for finding this mapping. Our studies show that effective approaches that exploit information redundancies due to users’ unique behavioral patterns can be utilized to find such a mapping. This study paves the way for analysis and mining across social networking sites, and facilitates the creation of novel online services across sites. In particular, recommending friends and advertising across networks, analyzing information diffusion across sites, and studying specific user behavior such as user migration across sites in social media are one of the many areas that can benefit from the results of this study.

Recommender system has become an important component in modern eCommerce. Recent research on recommender systems has been mainly concentrating on improving the relevance or profitability of individual recommended items. But in reality, users are usually exposed to a set of items and they may buy multiple items in one single order. Thus, the relevance or profitability of one item may actually depend on the other items in the set. In other words, the set of recommendations is a bundle with items interacting with each other. In this paper, we introduce a novel problem called the Bundle Recommendation Problem (BRP). By solving the BRP, we are able to find the optimal bundle of items to recommend with respect to preferred business objective. However, BRP is a large-scale NP-hard problem. We then show that it may be sufficient to solve a significantly smaller version of BRP depending on properties of input data. This allows us to solve BRP in real-world applications with millions of users and items. Both offline and online experimental results on a Walmart.com demonstrate the incremental value of solving BRP across multiple baseline models.

Microblogging, like Twitter and Sina Weibo, has become a popular platform of human expressions, through which users can easily produce content on breaking news, public events, or products. The massive amount of microblogging data is a useful and timely source that carries mass sentiment and opinions on various topics. Existing sentiment analysis approaches often assume that texts are independent and identically distributed (i.i.d.), usually focusing on building a sophisticated feature space to handle noisy and short texts, without taking advantage of the fact that the microblogs are networked data. Inspired by the social sciences findings that sentiment consistency and emotional contagion are observed in social networks, we investigate whether social relations can help sentiment analysis by proposing a Sociological Approach to handling Noisy and short Texts (SANT) for sentiment classification. In particular, we present a mathematical optimization formulation that incorporates the sentiment consistency and emotional contagion theories into the supervised learning process; and utilize sparse learning to tackle noisy texts in microblogging. An empirical study of two real-world Twitter datasets shows the superior performance of our framework in handling noisy and short tweets.

In this paper, we consider the problem of learning from multiple related tasks for improved generalization performance by extracting their shared structures. The alternating structure optimization (ASO) algorithm, which couples all tasks using a shared feature representation, has been successfully applied in various multitask learning problems. However, ASO is nonconvex and the alternating algorithm only finds a local solution. We first present an improved ASO formulation (iASO) for multitask learning based on a new regularizer. We then convert iASO, a nonconvex formulation, into a relaxed convex one (rASO). Interestingly, our theoretical analysis reveals that rASO finds a globally optimal solution to its nonconvex counterpart iASO under certain conditions. rASO can be equivalently reformulated as a semidefinite program (SDP), which is, however, not scalable to large datasets. We propose to employ the block coordinate descent (BCD) method and the accelerated projected gradient (APG) algorithm separately to find the globally optimal solution to rASO; we also develop efficient algorithms for solving the key subproblems involved in BCD and APG. The experiments on the Yahoo webpages datasets and the Drosophila gene expression pattern images datasets demonstrate the effectiveness and efficiency of the proposed algorithms and confirm our theoretical analysis.

A multimode network consists of heterogeneous types of actors with various interactions occurring between them. Identifying communities in a multimode network can help understand the structural properties of the network, address the data shortage and unbalanced problems, and assist tasks like targeted marketing and finding influential actors within or between groups. In general, a network and its group structure often evolve unevenly. In a dynamic multimode network, both group membership and interactions can evolve, posing a challenging problem of identifying these evolving communities. In this work, we try to address this problem by employing the temporal information to analyze a multimode network. A temporally regularized framework and its convergence property are carefully studied. We show that the algorithm can be interpreted as an iterative latent semantic analysis process, which allows for extensions to handle networks with actor attributes and within-mode interactions. Experiments on both synthetic data and real-world networks demonstrate the efficacy of our approach and suggest its generality in capturing evolving groups in networks with heterogeneous entities and complex relationships.

This study of collective behavior is to understand how individuals behave in a social networking environment. Oceans of data generated by social media like Facebook, Twitter, Flickr, and YouTube present opportunities and challenges to study collective behavior on a large scale. In this work, we aim to learn to predict collective behavior in social media. In particular, given information about some individuals, how can we infer the behavior of unobserved individuals in the same network? A social-dimension-based approach has been shown effective in addressing the heterogeneity of connections presented in social media. However, the networks in social media are normally of colossal size, involving hundreds of thousands of actors. The scale of these networks entails scalable learning of models for collective behavior prediction. To address the scalability issue, we propose an edge-centric clustering scheme to extract sparse social dimensions. With sparse social dimensions, the proposed approach can efficiently handle networks of millions of actors while demonstrating a comparable prediction performance to other nonscalable methods.

The volume of microblogging messages is increasing exponentially with the popularity of microblogging services. With a large number of messages appearing in user interfaces, it hinders user accessibility to useful information buried in disorganized, incomplete, and unstructured text messages. In order to enhance user accessibility, we propose to aggregate related microblogging messages into clusters and automatically assign them semantically meaningful labels. However, a distinctive feature of microblogging messages is that they are much shorter than conventional text documents. These messages provide inadequate term co occurrence information for capturing semantic associations. To address this problem, we propose a novel framework for organizing unstructured microblogging messages by transforming them to a semantically structured representation. The proposed framework first captures informative tree fragments by analyzing a parse tree of the message, and then exploits external knowledge bases (Wikipedia and WordNet) to enhance their semantic information. Empirical evaluation on a Twitter dataset shows that our framework significantly outperforms existing state-of-the-art methods.

Behavioral targeting (BT) is a widely used technique for online advertising. It leverages information collected on an individual's web-browsing behavior, such as page views, search queries and ad clicks, to select the ads most relevant to user to display. With the proliferation of social networks, it is possible to relate the behavior of individuals and their social connections. Although the similarity among connected individuals are well established (i.e., homophily), it is still not clear whether and how we can leverage the activities of one's friends for behavioral targeting; whether forecasts derived from such social information are more accurate than standard behavioral targeting models. In this paper, we strive to answer these questions by evaluating the predictive power of social data across 60 consumer domains on a large online network of over 180 million users in a period of two and a half months. To our best knowledge, this is the most comprehensive study of social data in the context of behavioral targeting on such an unprecedented scale. Our analysis offers interesting insights into the value of social data for developing the next generation of targeting services.

The prolific use of participatory Web and social networking sites is reshaping the ways in which people interact with one another. It has become a vital part of human social life in both the developed and developing world. People sharing certain similarities or affiliates tend to form communities within social media. At the same time, they participate in various online activities: content sharing, tagging, posting status updates, etc. These diverse activities leave behind traces of their social life, providing clues to understand changing social structures. A large body of existing work focuses on extracting cohesive groups based on network topology. But little attention is paid to understanding the changing social structures. In order to help explain the formation of a group, we explore different group-profiling strategies to construct descriptions of a group. This research can assist network navigation, visualization, and analysis, as well as monitoring and tracking the ebbs and tides of different groups in evolving networks. By exploiting information collected from real-world social media sites, extensive experiments are conducted to evaluate group-profiling results. The pros and cons of different group-profiling strategies are analyzed with concrete examples. We also show some potential applications based on group profiling. Interesting findings with discussions are reported.

The increasing popularity of social media is shortening the distance between people. Social activities, e.g., tagging in Flickr, book marking in Delicious, twittering in Twitter, etc. are reshaping people's social life and redefining their social roles. People with shared interests tend to form their groups in social media, and users within the same community likely exhibit similar social behavior (e.g., going for the same movies, having similar political viewpoints), which in turn reinforces the community structure. The multiple interactions in social activities entail that the community structures are often overlapping, i.e., one person is involved in several communities. We propose a novel co-clustering framework, which takes advantage of networking information between users and tags in social media, to discover these overlapping communities. In our method, users are connected via tags and tags are connected to users. This explicit representation of users and tags is useful for understanding group evolution by looking at who is interested in what. The efficacy of our method is supported by empirical evaluation in both synthetic and online social networking data.

Multi-label problems arise in various domains such as multi-topic document categorization, protein function prediction, and automatic image annotation. One natural way to deal with such problems is to construct a binary classifier for each label, resulting in a set of independent binary classification problems. Since multiple labels share the same input space, and the semantics conveyed by different labels are usually correlated, it is essential to exploit the correlation information contained in different labels. In this paper, we consider a general framework for extracting shared structures in multi-label classification. In this framework, a common subspace is assumed to be shared among multiple labels. We show that the optimal solution to the proposed formulation can be obtained by solving a generalized eigenvalue problem, though the problem is nonconvex. For high-dimensional problems, direct computation of the solution is expensive, and we develop an efficient algorithm for this case. One appealing feature of the proposed framework is that it includes several well-known algorithms as special cases, thus elucidating their intrinsic relationships. We further show that the proposed framework can be extended to the kernel-induced feature space. We have conducted extensive experiments on multi-topic web page categorization and automatic gene expression pattern image annotation tasks, and results demonstrate the effectiveness of the proposed formulation in comparison with several representative algorithms.

The study of collective behavior is to understand how individuals behave in a social network environment. Oceans of data generated by social media like Facebook, Twitter, Flickr and YouTube present opportunities and challenges to studying collective behavior in a large scale. In this work, we aim to learn to predict collective behavior in social media. In particular, given information about some individuals, how can we infer the behavior of unobserved individuals in the same network? A social-dimension based approach is adopted to address the heterogeneity of connections presented in social media. However, the networks in social media are normally of colossal size, involving hundreds of thousands or even millions of actors. The scale of networks entails scalable learning of models for collective behavior prediction. To address the scalability issue, we propose an edge-centric clustering scheme to extract sparse social dimensions. With sparse social dimensions, the socialdimension based approach can efficiently handle networks of millions of actors while demonstrating comparable prediction performance as other non-scalable methods.

With the pervasive availability of Web 2.0 and social networking sites, people can interact with each other easily through various social media. For instance, popular sites like Del.icio.us, Flickr, and YouTube allow users to comment shared content (bookmark, photos, videos), and users can tag their own favorite content. Users can also connect to each other, and subscribe to or become a fan or a follower of others. These diverse individual activities result in a multi-dimensional network among actors, forming cross-dimension group structures with group members sharing certain similarities. It is challenging to effectively integrate the network information of multiple dimensions in order to discover cross-dimension group structures. In this work, we propose a two-phase strategy to identify the hidden structures shared across dimensions in multi-dimensional networks. We extract structural features from each dimension of the network via modularity analysis, and then integrate them all to find out a robust community structure among actors. Experiments on synthetic and real-world data validate the superiority of our strategy, enabling the analysis of collective behavior underneath diverse individual activities in a large scale.

Multi-task learning (MTL) aims to improve generalization performance by learning multiple related tasks simultaneously. In this paper, we consider the problem of learning shared structures from multiple related tasks. We present an improved formulation (iASO) for multi-task learning based on the non-convex alternating structure optimization (ASO) algorithm, in which all tasks are related by a shared feature representation. We convert iASO, a non-convex formulation, into a relaxed convex one, which is, however, not scalable to large data sets due to its complex constraints. We propose an alternating optimization (cASO) algorithm which solves the convex relaxation efficiently, and further show that cASO converges to a global optimum. In addition, we present a theoretical condition, under which cASO can find a globally optimal solution to iASO. Experiments on several benchmark data sets confirm our theoretical analysis. 1.

Social media such as blogs, Facebook, Flickr, etc., presents data in a network format rather than classical IID distribution. To address the interdependency among data instances, relational learning has been proposed, and collective inference based on network connectivity is adopted for prediction. However, connections in social media are often multi-dimensional. An actor can connect to another actor for different reasons, e.g., alumni, colleagues, living in the same city, sharing similar interests, etc. Collective inference normally does not differentiate these connections. In this work, we propose to extract latent social dimensions based on network information, and then utilize them as features for discriminative learning. These social dimensions describe diverse affiliations of actors hidden in the network, and the discriminative learning can automatically determine which affiliations are better aligned with the class labels. Such a scheme is preferred when multiple diverse relations are associated with the same network. We conduct extensive experiments on social media data (one from a real-world blog site and the other from a popular content sharing site). Our model outperforms representative relational learning methods based on collective inference, especially when few labeled data are available. The sensitivity of this model and its connection to existing methods are also examined.

Multi-label problems arise in various domains such as multi-topic document categorization and protein function prediction. One natural way to deal with such problems is to construct a binary classifier for each label, resulting in a set of independent binary classification problems. Since the multiple labels share the same input space, and the semantics conveyed by different labels are usually correlated, it is essential to exploit the correlation information contained in different labels. In this paper, we consider a general framework for extracting shared structures in multi-label classification. In this framework, a common subspace is assumed to be shared among multiple labels. We show that the optimal solution to the proposed formulation can be obtained by solving a generalized eigenvalue problem, though the problem is non-convex. For high-dimensional problems, direct computation of the solution is expensive, and we develop an efficient algorithm for this case. One appealing feature of the proposed framework is that it includes several well-known algorithms as special cases, thus elucidating their intrinsic relationships. We have conducted extensive experiments on eleven multi-topic web page categorization tasks, and results demonstrate the effectiveness of the proposed formulation in comparison with several representative algorithms.

A multi-mode network typically consists of multiple heterogeneous social actors among which various types of interactions could occur. Identifying communities in a multi-mode network can help understand the structural properties of the network, address the data shortage and unbalanced problems, and assist tasks like targeted marketing and finding influential actors within or between groups. In general, a network and the membership of groups often evolve gradually. In a dynamic multi-mode network, both actor membership and interactions can evolve, which poses a challenging problem of identifying community evolution. In this work, we try to address this issue by employing the temporal information to analyze a multi-mode network. A spectral framework and its scalability issue are carefully studied. Experiments on both synthetic data and real-world large scale networks demonstrate the efficacy of our algorithm and suggest its generality in solving problems with complex relationships.

Blogging becomes a popular way for a Web user to publish information on the Web. Bloggers write blog posts, share their likes and dislikes, voice their opinions, provide suggestions, report news, and form groups in Blogosphere. Bloggers form their virtual communities of similar interests. Activities happened in Blogosphere affect the external world. One way to understand the development on Blogosphere is to find influential blog sites. There are many non-influential blog sites which form the "the long tail". Regardless of a blog site being influential or not, there are influential bloggers. Inspired by the high impact of the influentials in a physical community, we study a novel problem of identifying influential bloggers at a blog site. Active bloggers are not necessarily influential. Influential bloggers can impact fellow bloggers in various ways. In this paper, we discuss the challenges of identifying influential bloggers, investigate what constitutes influential bloggers, present a preliminary model attempting to quantify an influential blogger, and pave the way for building a robust model that allows for finding various types of the influentials. To illustrate these issues, we conduct experiments with data from a real-world blog site, evaluate multi-facets of the problem of identifying influential bloggers, and discuss unique challenges. We conclude with interesting findings and future work

A topic taxonomy is an effective representation that describes salient features of virtual groups or online communities. A topic taxonomy consists of topic nodes. Each internal node is defined by its vertical path (i.e., ancestor and child nodes) and its horizonal list of attributes (or terms). In a text-dominant environment, a topic taxonomy can be used to flexibly describe a group's interests with varying granularity. However, the stagnant nature of a taxonomy may fail to timely capture the dynamic change of a group's interest. This article addresses the problem of how to adapt a topic taxonomy to the accumulated data that reflects the change of a group's interest to achieve dynamic group profiling. We first discuss the issues related to topic taxonomy. We next formulate taxonomy adaptation as an optimization problem to find the taxonomy that best fits the data. We then present a viable algorithm that can efficiently accomplish taxonomy adaptation. We conduct extensive experiments to evaluate our approach's efficacy for group profiling, compare the approach with some alternatives, and study its performance for dynamic group profiling. While pointing out various applications of taxonomy adaption, we suggest some future work that can take advantage of burgeoning Web 2.0 services for online targeted marketing, counterterrorism in connecting dots, and community tracking.

Hierarchical models have been shown to be effective in content classification. However, we observe through empirical study that the performance of a hierarchical model varies with given taxonomies; even a semantically sound taxonomy has potential to change its structure for better classification. By scrutinizing typical cases, we elucidate why a given semantics-based hierarchy does not work well in content classification, and how it could be improved for accurate hierarchical classification. With these understandings, we propose effective localized solutions that modify the given taxonomy for accurate hierarchical classification. We conduct extensive experiments on both toy and real-world data sets, report improved performance and interesting findings, and provide further analysis of algorithmic issues such as time complexity, robustness, and sensitivity to the number of features.

Feature selection is often applied to high-dimensional data as a preprocessing step in text classification. When dealing with highly skewed data, we observe that typical feature selection metrics like information gain or chi-squared are biased toward selecting features for the minor class, and the metric of bi-normal separation can select features for both minor and major classes. In this work, we investigate how these feature selection metrics impact on the performance of frequently used classifiers such as decision trees, naive bayes, and support vector machines via bias analysis for highly skewed data. Three types of biases are metric bias, class bias, and classifier bias. Extensive experiments are designed to understand how these biases can be employed in concert and efficiently to achieve good classification performance. We report our findings and present recommended approaches to text classification based on bias analysis and the empirical study.