Abstract Improving health literacy through health information dissemination is one of the most economical and effective mechanisms for improving population health. This process needs to fully accommodate the thematic suitability of health information supply and demand and reduce the impact of information overload and supply–demand mismatch on the enthusiasm of health information acquisition. We propose a health information topic modeling analysis framework that integrates deep learning methods and clustering techniques to model the supply‐side and demand‐side topics of health information and to quantify the thematic alignment of supply and demand. To validate the effectiveness of the framework, we have conducted an empirical analysis on a dataset with 90,418 pieces of textual data from two prominent social networking platforms. The results show that the supply of health information in general has not yet met the demand, the demand for health information has not yet been met to a considerable extent, especially for disease‐related topics, and there is clear inconsistency between the supply and demand sides for the same health topics. Public health policy‐making departments and content producers can adjust their information selection and dissemination strategies according to the distribution of identified health topics, thereby improving the effectiveness of public health information dissemination.

The unique characteristics of online social communities call for a reexamination and adaptation of established behavioral theories of trust decision-making. Guided by relevant social science and computational graph theories, we propose a conceptual model of trust decision-making in online social networks. This is the first study that integrates the existing graph-based view of trust decision-making in social networks into socio-psychological theories of trust to provide a richer understanding of trusting decisions in online social networks. We introduce new behavioral antecedents of trusting decisions, and redefine and integrate existing graph-based concepts to develop our proposed conceptual model. We introduce new behavioral antecedents of trusting decisions that have not been identified in previous research. We also identify novel operationalization methods to measure behavioral trust-inducing factors for online social networks. Our empirical findings indicate that both behavioral and network-specific trust decision-making factors should be considered in studying trusting decisions in online social networks.

Spatio-temporal data mining is finding applications in many domains, such as public health, public safety, financial fraud detection, transportation, and product lifecycle management. Correlation analysis is an important spatio-temporal mining technique for unveiling spatial and temporal relationships among multiple event types. This paper presents a new measure for assessing and analyzing spatio-temporal cross-correlations. This measure extends Ripley’s a widely used measure of spatial correlation, with an additional temporal dimension. Empirical studies using real-world data show that the new measure can lead to a more discriminating and flexible spatio-temporal data analysis framework. In contrast with its predecessor, this measure also allows the discovery of leading (and potentially causal) event types whose occurrences precede those of other event types. Findings from analyses employing this measure may bear important managerial implications.

Tapping into the wisdom of the crowd, social tagging can be considered an alternative mechanism - as opposed to Web search - for organizing and discovering information on the Web. Effective tag-based recommendation of information items, such as Web resources, is a critical aspect of this social information discovery mechanism. A precise understanding of the information structure of social tagging systems lies at the core of an effective tag-based recommendation method. While most of the existing research either implicitly or explicitly assumes a simple tripartite graph structure for this purpose, we propose a comprehensive information structure to capture all types of co-occurrence information in the tagging data. Based on the proposed information structure, we further propose a unified user profiling scheme to make full use of all available information. Finally, supported by our proposed user profile, we propose a novel framework for collaborative filtering in social tagging systems. In our proposed framework, we first generate joint item-tag recommendations, with tags indicating topical interests of users in target items. These joint recommendations are then refined by the wisdom from the crowd and projected to the item space for final item recommendations. Evaluation using three real-world datasets shows that our proposed recommendation approach significantly outperformed state-of-the-art approaches.

Identifying matching attributes across heterogeneous data sources is a critical and time-consuming step in integrating the data sources. In this paper, the author proposes a method for matching the most frequently encountered types of attributes across overlapping heterogeneous data sources. The author uses mutual information as a unified measure of dependence on various types of attributes. An example is used to demonstrate the utility of the proposed method, which is useful in developing practical attribute matching tools.

Several methods have been proposed for cascading other classification algorithms with decision tree learners to alleviate the representational bias of decision trees and, potentially, to improve classification accuracy. Such cascade generalization of decision trees increases the flexibility of the decision boundaries between classes and promotes better fitting of the training data. However, more flexible models may not necessarily lead to more predictive power. Because of potential overfitting problems, the true classification accuracy on test data may not increase. Recently, a generic method for cascade generalization has been proposed. The method uses a parameter — the maximum cascading depth — to constrain the degree that other classification algorithms are cascaded with decision tree learners. A method for efficiently learning a collection (i.e., a forest) of generalized decision trees, each with other classification algorithms cascaded to a particular depth, also has been developed. In this article, we propose several new strategies, including elitist and ensemble (weighted or unweighted), for using the various decision trees in such a collection in the prediction phase. Our empirical evaluation using 32 data sets in the UCI machine learning repository shows that, on average, the elitist strategy outperforms the weighted full ensemble strategy, which, in turn, outperforms the unweighted full ensemble strategy. However, no strategy is universally superior across all applications. Since the same training process can be used to evaluate the various strategies, we recommend that several promising strategies be evaluated and compared before selecting the one to use for a given application.

While decision tree techniques have been widely used in classification applications, a shortcoming of many decision tree inducers is that they do not learn intermediate concepts, i.e., at each node, only one of the original features is involved in the branching decision. Combining other classification methods, which learn intermediate concepts, with decision tree inducers can produce more flexible decision boundaries that separate different classes, potentially improving classification accuracy. We propose a generic algorithm for cascade generalization of decision tree inducers with the maximum cascading depth as a parameter to constrain the degree of cascading. Cascading methods proposed in the past, i.e., loose coupling and tight coupling, are strictly special cases of this new algorithm. We have empirically evaluated the proposed algorithm using logistic regression and C4.5 as base inducers on 32 UCI data sets and found that neither loose coupling nor tight coupling is always the best cascading strategy and that the maximum cascading depth in the proposed algorithm can be tuned for better classification accuracy. We have also empirically compared the proposed algorithm and ensemble methods such as bagging and boosting and found that the proposed algorithm performs marginally better than bagging and boosting on the average.