With the growth of social networking website, social media has become a major platform for marketing, such as social business, political manipulation, influence and brand management, etc. However, social media marketing is very different to traditional marketing. Social media marketing needs to face large number of users and need to repeat same process frequently. It is therefore a very human power consuming task. Under this situation, it is the reason why Robotic Process Automation and Social-bots is now very popular in many social networking websites. However, there are many negative effects when applying social-bots for social marketing. Therefore, more and more researchers are devoting on propose efficient ways to detect social-bots. In this paper, we proposed an approach to detect social-bots by considering the content that users posted as well as the behavior and features when using social networking website. In this approach, we adopt the concept of word embedding and social network embedding. Convolutional neural network is used as the main techniques to train the model for social-bots detection. The experimental results show that the proposed combination approach has better detection accuracy than only social network embedding or word embedding approach as well as it reaches 92% detection accuracy by using our dataset.

Web flow now is a very important success factor for social media marketing and thus more and more approaches for creating high web flow have been proposed in recent years. Automatic content generation (ACG) website is one of the possible approaches which can help to create web flow. In order to achieve the idea of automatic content generation website, web article classification has been considered the most important task. Therefore, we have development an empirical study to test the content labeling and article classification performance, which is based on the technique of BERT neural network. The performance evaluation including accuracy performance and time performance that are important for us to understand the possibility for implementing the ACG website in real environment, especially the possibility when dealing with large amount of data.

In recent years, social media has becoming a battle field, not only for online marketing but also for politic, etc. Such as Facebook, online advertisement is now the main revenue of their company and the main idea is to attract users to particular fans page and to create flow. Flow is king is now the important concept for who want to manage their business online. Thus, in this paper, we intend to develop a website based on the concept of auto-article generation (AAG), which can gather useful information or news from other resources from WWW. The techniques that used for the AAG website including web crawler, cloud storage and computing, content classification, etc. The main idea is to attract users to visit the website and by this to create website flow.

Hot topics analysis is one of the important task for users to organize information from WWW and especially for social networking websites. Therefore, how to design an efficient approach for users to extract those hot topics is very essential. Thus, we proposed a so-called 2-layers keywords extraction approach and three empirical analyses are then be applied to validate the usability and performance of the proposed approach.

The present article aims to disclose the role of task–individual–social software fit (TISF) in knowledge creation in the context of the manufacturing and service industries and research institutes. The methodology used is an empirical study that proposes and examines the proposed research model. The mediation effect of structural social exchange is also explored. Results of the data analysis of 279 valid samples reveal the following findings. First, the effect of TISF is confirmed. Second, structural exchanges do not mediate the role of TISF toward creation performance. Third, TISF is significantly associated with the social software, creation task, and individual cognition variables. Lastly, goal-free and goal-frame creation modes and analytical and intuitive cognition styles significantly influence the fit of features of creation task, individual cognition, and social software. The article provides domain scholars and practitioners with value of the task–individual–social software fit in the context of knowledge creation. Discussion and implications are also presented in this article.

Mobile user data are collected by service providers around the clock and through intelligent data analysis in which it can offer great services for health cares, business activities, and other personal or social services, etc. However, data could be misused and privacy could potentially be breached which might lead to harmful consequences. Many privacy-preserving techniques have been proposed in the past decade for anonymizing relational and social data. But only a handful of privacy-preserving techniques have been proposed to anonymize sensitive mobile context before releasing data to service providers. Unfortunately, these techniques also reduce the utility of data that are supposed to provide helpful services. As such, the effectiveness of these anonymization techniques cannot be easily justified and compared. In this work, we propose a unified approach to define privacy gain and utility loss due to anonymizing sensitive context on mobile user data. We further perform extensive numerical evaluation on various well-known anonymization techniques, compare their performances and trade-offs between privacy and utility, and also provide a framework of analysis which serves a reference for adopting suitable anonymization technique for different user requirements.

In recent years, with rapid growth of social networking websites, users are very active in these platforms and large amount of data are aggregated. Among those social networking websites, Facebook is the most popular website that has most users. However, in Facebook, the abusing problem is a very critical issue, such as Hate Groups. Therefore, many researchers are devoting on how to detect potential hate groups, such as using the techniques of social networks analysis. However, we believe content is also a very important factors for hate groups detection. Thus, in this paper, we will propose an architecture to for hate groups detection which is based on the technique of Social Networks Analysis and Web Mining (Text Mining; Natural Language Processing). From the experiment result, it shows that content plays an critical role for hate groups detection and the performance is better than the system that just applying social networks analysis.

Efficient shortest path calculation has been studied extensively, in particular, in the distributed environment. However, preserving privacy in the cloud environment has just attracted latest attention. To preserve fixed-pattern one-neighborhood privacy in the cloud, current approach requires the calculation of all-pairs shortest paths in advance, which is time consuming for large graphs. In addition, specific paths that are sensitive and require hiding the source and destination vertices are not well addressed. In this work, we propose a new flexible k-neighborhood privacy-protection and efficient shortest distance computation scheme for sensitive shortest paths in the cloud environment. Combining the construction of k-skip shortest path sub-graphs, sensitive vertex adjustment, vertex hierarchy labeling and bottom-up partitioning techniques, the proposed approach not only subsumes one-neighborhood privacy but also provides efficient partitioning and query processing for sensitive shortest paths. Numerical experiments demonstrating the characteristics of proposed approach are presented.

Preserving privacy in social networking environment has been studied extensively in recent years. Although more works have adopted un-weighted graphs to model network relationships, weighted graph modeling can provide deeper analysis of the degree of relationships. Previous works on weighted graph privacy have concentrated on preserving the shortest path characteristic between pairs of vertices. Two common types of privacy have been proposed. One type of privacy tried to add random noise edge weights to the graph but still maintain the same shortest path. The other privacy, k-shortest path privacy, minimally perturbed edge weights so that there exist k shortest paths. However, the k-shortest path privacy did not consider degree attacks on the nodes of anonymized shortest paths. For example, if the adversary possesses background knowledge of node degrees on the shortest path, the true shortest path can be identified. In this work, we present a new concept called (k1, k2)-shortest path privacy to prevent such privacy breach. A published network graph with (k1, k2)-shortest path privacy has at least k1 indistinguishable shortest paths between the source and destination vertices. In addition, for the non-overlapping vertices on the k1 shortest paths, there exist at least k2 vertices with same node degree and lie on more than one shortest path. Three heuristic algorithms are proposed and experimental results showing the feasibility and characteristics of the proposed approaches are presented.

Due to the proliferation of online social networking, a large number of personal data are publicly available. As such, personal attacks, reputational, financial, or family losses might occur once this personal and sensitive information falls into the hands of malicious hackers. Research on Privacy-Preserving Network Publishing has attracted much attention in recent years. But most work focus on node de-identification and link protection. In academic social networks, business transaction networks, and transportation networks, etc, node identities and link structures are public knowledge but weights and shortest paths are sensitive. In this work, we study the problem of k-anonymous path privacy. A published network graph with k-anonymous path privacy has at least k indistinguishable shortest paths between the source and destination vertices [21]. In order to achieve such privacy, three different strategies of modification on edge weights of directed graphs are proposed. Numerical comparisons show that weight-proportional-based strategy is more efficient than PageRank-based and degree-based strategies. In addition, it is also more efficient and causes less information loss than running on un-directed graphs.

In recent years, social networking sites have becoming important platforms for users to establish the relationships between each other. As time goes by, the links between people will form the so-called "Strong Links". For those users, information provided by the friends with strong link is considered as more interesting and useful. Most of recent search engines are designed based on only measuring the similarity between keywords and articles. However, the social relations between authors of articles and searcher have not been taken into account in recent research. Therefore, in order to improve the performance of recent search engines, we include the measurement of social relationships in search engine and expect the search quality can be improved. In this study, we collected the data from Facebook to calculate the social relationship. About the content, the data will be processed by using CKIP (Chinese word net) and TF-IDF. Finally, we combine key-word frequency and social relations as a value, which is called the Social Ranking vaule. The value will be used as the key to rank the search results. In this paper, we will also demonstrate a real example to explain the proposed methodology as well as a system interface.

Location and local service is one of the hottest bunches of applications in recent years, due to the proliferation of Global Position System (GPS) and mobile web search technology. Spatial queries retrieving neighboring Point-Of-Interests (POI) require actual user locations for services. However, exposing the physical location of querier to service system may pose privacy threat to users, if malicious adversary has access to the system. To hinder the service system from obtaining the "true" location of querier, current obfuscation-based approach requires a trusted third party anonymizer. As for the data-encryption-based and cPIR-based approaches, they incur costly computation overheads. Although the secure hardware-aided PIR-based technique has been shown to be superior to formers, it did not consider the characteristics of data distribution of searching domain. To deal with the problem of non-uniform data distribution and efficient retrieval, we propose a scheme, MHBL, based on Hilbert space-filling curve and flexible multi-layer grids for efficient storage and retrieval of POI data, so that improved performance of PIR-based techniques could be achieved. Numerical experiments demonstrate that the proposed technique indeed deliver better efficiency under various criteria.

Due to the proliferation of Global Position System (GPS) and smart phone technology, Location-Based Service (LBS) has attained tremendous growth in recent years. Spatial queries retrieving nearest Point-Of-Interests (POI) require actual user locations for services. However, sharing such sensitive personal location information with potentially malicious servers may cause concerns about user privacy. The current obfuscation-based approach addressing this problem cannot provide binding privacy guarantees as a trusted third-party anonymizer is required. On the other hand, the data-encryption-based and cPIR-based approaches incur costly computation overheads. Recently, the secure hardware-aided PIR-based technique has been shown to be superior to formers, but it did not consider the characteristics of data distribution of searching domain. In this work, we propose two schemes: MSQL, NSQL, based on flexible multi-layer grids and non-empty lookup table for efficient storage and retrieval on non-uniform distribution of POI data, so that improved performance of PIR-based techniques could be achieved. Numerical experiments demonstrate that the proposed techniques indeed deliver better efficiency under various criteria.

With the rapid growth of Internet and social networking websites, there are various services that provided in these platforms. For instance, Face book focuses on social activities, Twitter and Plurk are both focus on the interaction of users through short messages (which are so-called microblogs). Therefore, there are more than millions of users registered in these websites and become places where full of marketing possibilities. Thus, it is an important issue to assist companies to understand the users in the social networking websites in order to enhance the accuracy and efficiency of target marketing. In this paper, we have proposed the architecture of a social recommendation system based on the data from microblogs. The social recommendation system is conducted according to the messages and social structure of target users. The similarity of the discovered features of users and products will then be calculated as the essence of the recommendation engine. A case study will be included to present how the recommendation system works based on real data that collected from Plurk.

The research area of Social networks analysis has been recognized as extremely time-consuming tasks as well as large storage space is always necessary in order to store the social data, especially to deal with the data in the World Wide Web. Therefore, how to design an architecture and environment for performing social networks analysis is very essential. In this paper, we proposed a data warehousing and analyzing system which is based on the concept of cloud computing. The system has also been implemented and evaluated under the proposed environment with different cloud computing approaches.

Academic network (AN) is quite different network and hard to acquire relative information for inexperience researchers. To realize AN is an important but difficult work because of the constituents of AN are complicated and dynamic variation. In this research, two Taiwan academic networks are established from the perspectives of academic conference topic and participant committee. Each academic conference is regarded as an AN node to establish the conference academic network. In addition, there are two connection types between AN nodes, including: the relationship among CFP topics (academic conference topic perspective) and the relationship between committee member (participant committee perspective). Finally, 15 IT/IM relative target conferences in Taiwan 2009 are analyzed via social network analysis methodology. According to experiment result, the values of degree and betweeness reveal some interesting findings, such as TANET, OOTA and IMP are top three important bridge conferences in Taiwan. Additionally, as the cluster analyzed result shown, academic network somehow represents domain expertise because of the committee are grouped according to their research domains. These finding is consistence with the reality academic network in Taiwan and helpful information for Taiwan academic network understanding.

Social networks are extremely important for word-of-mouth (WOM) marketing. However, marketers often ignore network structures when developing WOM marketing strategies. Specifically, there is clearly a lack of research on looking into overall structures and structural heterogeneities of social networks. This research investigates structural heterogeneities between online social networks in different product categories. We collected data from four online networks in different product categories from the most popular social networking site in Taiwan. Social network analysis was performed to understand the network structures. The findings demonstrate the structural heterogeneities between these networks and we also provide managerial implications for practitioners.

Large amount of social (communication) data have been generated in many applications for personal communication purpose. However, these data have not been used well currently. In this paper, we will introduce a methodology to collect and analyze those personal data, and by this for extracting social networks from the data. A system architecture will also be presented and implemented to show how the data can be collected, pre-processed, analyzed, which can also be used for personal decision support.

Global economic recession has been causing the unpaid leave and massive layoffs in major high-tech firms of Taiwan, both factors present great potential hardship to many employees according to the reports from industry. Therefore, layoff prediction and management have become great concerns of employees and managers. Employees wish to retain their jobs and keep their work for a long time. Hence, they need to predict the possible layoff and then utilize their resources to retain their job. In response to the difficulty of layoff prediction, this study applies social networks and data mining techniques to build a model for layoff prediction. This study compares various techniques to propose a better approach to generate a possible layoff list for employees. Through an empirical study, the results indicate that the proposed approach has pretty good prediction accuracy by using organizational networks, employee databases and layoff records to build the layoff prediction model.

In this paper, a web usage mining based approach is proposed to discover potential browsing problems. Two web usage mining techniques in the approach are introduced, including Automatic Pattern Discovery (APD) and Co-occurrence Pattern Mining with Distance Measurement (CPMDM). A combination method is also discussed to show how potential browsing problems can be identified

This paper describes a novel Web usage mining approach to discover patterns in the navigation of Web sites known as unexpected browsing behaviours (UBBs). By reviewing these UBBs, a Web site designer can choose to modify the design of their Web site or redesign the site completely. UBB mining is based on the continuous common subsequence (CCS), a special instance of common subsequence (CS), which is used to define a set of expected routes. The predefined expected routes are then treated as rules and stored in a rule base. By using the predefined route and the UBB mining algorithm, interesting browsing behaviours can be discovered. This paper introduces the format of the expected route and describes the UBB algorithms. The paper also describes a series of experiments designed to evaluate how well UBB mining algorithms work.