Data Mining. Analyzing Social Roles Based on a Hierarchical Model and Data Mining for Collective Decision-Making Support

Transcription

1 IEEE PROJECT LIST(JAVA) Data Mining CODE 18ANSP-DM ANSP-DM-002 TITLE AND ABSTRACT Analyzing Social Roles Based on a Hierarchical Model and Data Mining for Collective Decision-Making Support With the popularity of social networking services (SNSs) and the increase of users, individuals social roles in a social network have become more and more important in terms of the recommendation of personalized services and the collective decision-making process. Usually, in an SNS system, active users may not represent the major opinions among the whole users, and most of the users opinions may be multifarious. In this paper, we focus on analyzing and identifying users dynamical social roles to facilitate the collective decision-making process. After introducing the social choice theory and an improved collective decisionmaking model, we present a three-layer model to analyze users social roles in a hierarchical way and develop an integrated mechanism to utilize the identification of social roles to support the collective decision making. Based on a developed NetLogo-based tool, a case study for the course-offering determination with an application scenario is demonstrated to show the process of using users social roles to support the collective decision making. The comparison experiment conducted between our method and the Delphi method shows the usefulness of our proposed method to help users achieve the decision consensus in a more efficient way. A Reversible Watermarking Technique for Social Network Data Sets for Enabling Data Trust in Cyber, Physical, and Social Computing Social network data are being mined for extracting interesting patterns. Such data are collected by different researchers and organizations and are usually also shared via different channels. These data usually have huge volume because there are millions of social network users throughout the world. In this context, ownership protection of such data sets with huge volume becomes relevant. Digital watermarking is a more demanding solution than any other technique for ensuring rights protection and integrity of the original data sets. The objective of this paper is to devise a reversible watermarking technique for the social network data to prove ownership rights and also provide a mechanism for data recovery. Robustness of the proposed technique is evaluated through attack analysis using experimental study. In this paper, Z notation-based formal specification is also provided to show the working of the proposed reversible watermarking technique for social network data sets for enabling data trust in Cyber, Physical, and Social Computing (CPSCom).

2 18ANSP-DM ANSP-DM ANSP-DM-005 Dynamic Weight-Based Individual Similarity Calculation for Information Searching in Social Computing In the social computing environment, the complete information about an individual is usually distributed in heterogeneous social networks, which are presented as linked data. Synthetically recognizing and integrating these distributed and heterogeneous data for efficiently information searching is an important but challenging work. In this paper, a dynamic weight (DW)-based similarity calculation is proposed to recognize and integrate similar individuals from distributed data environments. First, each link of an individual is weighted by applying DW. Then, a semantic similarity metric is proposed to combine the DW into similarity calculation. Then, a searching system framework for a similarity-based individual is designed and tested in real-world data sets. Finally, massive experiments are conducted both in benchmark and real-world social community data sets. The results show that our approach can produce a good result in similar individual searching in social networks. In addition, it performs significantly better than the existing state-of-the-art approaches in similar individual searching. An Internal Intrusion Detection and Protection System by Using Data Mining and Forensic Techniques Currently, most computer systems use user IDs and passwords as the login patterns to authenticate users. However, many people share their login patterns with coworkers and request these coworkers to assist co-tasks, thereby making the pattern as one of the weakest points of computer security. Insider attackers, the valid users of a system who attack the system internally, are hard to detect since most intrusion detection systems and firewalls identify and isolate malicious behaviors launched from the outside world of the system only. In addition, some studies claimed that analyzing system calls (SCs) generated by commands can identify these commands, with which to accurately detect attacks, and attack patterns are the features of an attack. Therefore, in this paper, a security system, named the Internal Intrusion Detection and Protection System (IIDPS), is proposed to detect insider attacks at SC level by using data mining and forensic techniques. The IIDPS creates users personal profiles to keep track of users usage habits as their forensic features and determines whether a valid login user is the account holder or not by comparing his/her current computer usage behaviors with the patterns collected in the account holder s personal profile. The experimental results demonstrate that the IIDPS s user identification accuracy is 94.29%, whereas the response time is less than 0.45 s, implying that it can prevent a protected system from insider attacks effectively and efficiently. Target-Based, Privacy Preserving, and Incremental Association Rule Mining We consider a special case in association rule mining where mining is conducted by a third party over data located at a central location that is updated from several source locations. The data at the central location is at rest while that flowing in through source locations is in motion. We impose some limitations on the source locations, so that the central target location tracks and privatizes changes and a

3 18ANSP-DM-006 third party mines the data incrementally. Our results show high efficiency, privacy and accuracy of rules for small to moderate updates in large volumes of data. We believe that the framework we develop is therefore applicable and valuable for securely mining big data. Modeling Urban Behavior by Mining Geotagged Social Data Data generated on location-based social networks provide rich information on the whereabouts of urban dwellers. Specifically, such data reveal who spends time where, when, and on what type of activity (e.g., shopping at a mall, or dining at a restaurant). That information can, in turn, be used to describe city regions in terms of activity that takes place therein. For example, the data might reveal that citizens visit one region mainly for shopping in the morning, while another for dining in the evening. Furthermore, once such a description is available, one can ask more elaborate questions. For example, one might ask what features distinguish one region from another some regions might be different in terms of the type of venues they host and others in terms of the visitors they attract. As another example, one might ask which regions are similar across cities. In this paper, we present a method to answer such questions using publicly shared Foursquare data. Our analysis makes use of a probabilistic model, the features of which include the exact location of activity, the users who participate in the activity, as well as the time of the day and day of week the activity takes place. Compared to previous approaches to similar tasks, our probabilistic modeling approach allows us to make minimal assumptions about the data which relieves us from having to set arbitrary parameters in our analysis (e.g., regarding the granularity of discovered regions or the importance of different features). We demonstrate how the model learned with our method can be used to identify the most likely and distinctive features of a geographical area, quantify the importance features used in the model, and discover similar regions across different cities. Finally, we perform an empirical comparison with previous work and discuss insights obtained through our findings. 18ANSP-DM-007 Mining Sequential Risk Patterns From Large-Scale Clinical Databases for Early Assessment of Chronic Diseases: A Case Study on Chronic Obstructive Pulmonary Disease Chronic diseases have been among the major concerns in medical fields since they may cause a heavy burden on healthcare resources and disturb the quality of life. In this paper, we propose a novel framework for early assessment on chronic diseases by mining sequential risk patterns with time interval information from diagnostic clinical records using sequential rules mining, and classification modeling techniques. With a complete workflow, the proposed framework consists of four phases namely data preprocessing, risk pattern mining, classification modeling, and post analysis. For empiricasl evaluation, we

4 18ANSP-DM ANSP-DM-009 demonstrate the effectiveness of our proposed framework with a case study on early assessment of COPD. Through experimental evaluation on a large-scale nationwide clinical database in Taiwan, our approach can not only derive rich sequential risk patterns but also extract novel patterns with valuable insights for further medical investigation such as discovering novel markers and better treatments. To the best of our knowledge, this is the first work addressing the issue of mining sequential risk patterns with time-intervals as well as classification models for early assessment of chronic diseases. CHRS: Cold Start Recommendation Across Multiple Heterogeneous Information Networks Nowadays, people are overwhelmingly exposed to various kinds of information from different information networks. In order to recommend users with the information entities that match their interests, many recommendation methods have been proposed so far. And some of these methods have explored different ways to utilize different kinds of auxiliary information to deal with the information sparsity problem of user feedbacks. However, as a special kind of information sparsity problem, the ``cold start'' problem is still a big challenge not well-solved yet in the recommendation problem. In order to tackle the ``cold start'' challenge, in this paper, we propose a novel recommendation model, which integrates the auxiliary information in multiple heterogeneous information networks (HINs), namely the Cross- HIN Recommendation System (CHRS). By utilizing the rich heterogeneous information from meta-paths, the CHRS is able to calculate the similarities of information entities and apply the calculated similarity scores in the recommendation process. For the information entities shared among multiple information networks, CHRS transfers item latent information from other networks to help the recommendation task in a given network. During the information transfer process, CHRS applies a domain adaptation matrix to tackle the domain difference problem. We conduct experiments to compare our CHRS method with several widely employed or the state-of-art recommendation models, and the experimental results demonstrate that our method outperforms the baseline methods in addressing the ``cold start'' recommendation problem. EHAUPM: Efficient High Average-Utility Pattern Mining With Tighter Upper Bounds High-utility itemset mining (HUIM) has become a popular data mining task, as it can reveal patterns that have a high-utility, contrarily to frequent pattern mining, which focuses on discovering frequent patterns. High average-utility itemset mining (HAUIM) is a variation of HUIM that provides an alternative measure, called the average utility, to select patterns by considering both their utilities and lengths. In the last decades, several algorithms have been developed to mine high average-utility itemsets (HAUIs). But most of them consume large amounts of memory and have long execution times, since they generally utilize the averageutility upper-bound (auub) model to overestimate the average utilities of itemsets. To improve the performance of HAUIM, this paper proposes two novel tighter upper-bound models as alternative to the traditional auub model for mininghauis. The looser upper-bound model considers the remaining-maximum

5 18ANSP-DM ANSP-DM ANSP-DM-012 utility in transactions to reduce the upper bound on the utilities of itemsets. The second upper-bound model ignores irrelevant items in transactions to further tighten the upper bound. Three pruning strategies are also designed to reduce the search space for mining HAUIs by a greater amount compared with the state-ofthe-art HAUI-Miner algorithm. Experiments conducted on several benchmark data sets show that the designed algorithm integrating the two novel upper-bound models outperforms the traditional HAUI-Miner algorithm in terms of runtime, memory usage, number of join operations, and scalability. Privacy-Preserving Data Mining: Methods, Metrics, and Applications The collection and analysis of data are continuously growing due to the pervasiveness of computing devices. The analysis of such information is fostering businesses and contributing beneficially to the society in many different fields. However, this storage and flow of possibly sensitive data poses serious privacy concerns. Methods that allow the knowledge extraction from data, while preserving privacy, are known as privacy-preserving data mining (PPDM) techniques. This paper surveys the most relevant PPDM techniques from the literature and the metrics used to evaluate such techniques and presents typical applications of PPDM methods in relevant fields. Furthermore, the current challenges and open issues in PPDM are discussed. A Two-Layer Clustering Model for Mobile Customer Analysis Customer segmentation provides an effective way to obtain insights into customer characteristics and behavioral preferences. The authors two-layer clustering model for mobile telecom customer analysis enhances customer relationship management and focuses on a dynamically changing marketplace. With the rise of big data and the evolution of data mining technology, the mass storage of internal enterprise information can be analyzed effectively for hidden customer value. The promotion of marketing activities and customer relationship support are also based on an extensive precision-marketing model that evolved to target the customer base and obtain in-depth understanding to fit that base s needs. An increasingly important issue is how to integrate marketing resources and properly distribute and match individual customer interests and preferences with the most effective marketing activities, as well as mine data to determine those products or services most attractive to customers. Mining Coherent Topics With Pre-Learned Interest Knowledge in Twitter Discovering semantic coherent topics from the large amount of user-generated content (UGC) in social media would facilitate many downstream applications of intelligent computing. Topic models, as one of the most powerful algorithms, have been widely used to discover the latent semantic patterns in text collections. However, one key weakness of topic models is that they need documents with certain length to provide reliable statistics for generating coherent topics. In Twitter, the users' tweets are mostly short and noisy. Observations of word cooccurrences are incomprehensible for topic models. To deal with this problem, previous work tried to incorporate prior knowledge to obtain better results. However, this strategy is not practical for the fast evolving UGC in Twitter. In this

6 18ANSP-DM ANSP-DM-014 paper, we first cluster the users according to the retweet network, and the users' interests are mined as the prior knowledge. Such data are then applied to improve the performance of topic learning. The potential cause for the effectiveness of this approach is that users in the same community usually share similar interests, which will result in less noisy sub-data sets. Our algorithm pre-learns two types of interest knowledge from the data set: the interest-word-sets and a tweetinterest preference matrix. Furthermore, a dedicated background model is introduced to judge whether a word is drawn from the background noise. Experiments on two real life twitter data sets show that our model achieves significant improvements over state-of-the-art baselines. Mining Frequent Route Patterns Based on Personal Trajectory Abstraction Frequent route pattern mining from personal trajectory data is the basis of location awareness and location services. However, because personal trajectory data is highly uncertain, most existing approaches are only capable of finding short and incomplete route patterns. In this paper, a novel approach is proposed for the discovery of frequent route patterns based on trajectory abstraction. First, trajectory partition, location extraction, data simplification, and common segment discovery are used to abstract trajectory data, convert these trajectories into common segment temporal sequences (STS) and generate 1-frequent itemsets. Then, a pattern mining algorithm is proposed based on the spatial-temporal adjacency relationship. This algorithm uses the constraint mechanism and bidirectional projected database to mine frequent route patterns from STS. Based on the real GeoLife trajectory data, the experimental results indicate that the proposed method has better performance and can find longer route patterns than other currently available methods. AdScope: Search Campaign Scoping Using Relevance Feedback Advertisers use online advertising for branding and direct response. To raise brand awareness, for example, they target a broad set of online users whether or not these users have an intention to buy something is of secondary importance. In contrast, in direct response campaigns, advertisers are keenly interested in users intention to buy, so they specifically match up ads and search words accordingly (an ad for the search phrase, how to create a social network for cheap, for example, gives advertisers access to bargain hunters searching the Internet for tools to create social networks). To advertise on a search network, the sponsored ad networks for search engine queries, advertisers must determine a set of keywords to represent how users will search for their products in a search engine. Once these keywords are determined, the campaign can go live. Because it s difficult in the beginning to know exactly which queries users will pose, advertisers often identify the key phrases that are most likely to occur in queries, as well as keywords that are broadly similar to queries. If an advertiser is certain that a specific keyword can be one of the queries, then that keyword can be used as is. It s crucial to have the right set of keywords to target the right set of users. Otherwise, the campaign budget will be wasted on users who are less likely to convert into sales.

7 18ANSP-DM-015 Aspect-Based Extraction and Analysis of Affective Knowledge from Social Media Streams This article introduces an approach to analyze emotional values associated with brands and companies. Online media coverage about products and services typically refers to a wide range of aspects to which such emotional values apply. These aspects can include product features (such as a digital camera s maximum resolution), common applications (such as a smartphone used as a car navigation system), or perceptions in conjunction with a specific event (for example, as part of a sponsorship agreement). Our approach integrates affective and factual knowledge extraction to capture opinions related to specific aspects along multiple emotional dimensions. We use the automotive industry as a sample domain to demonstrate the proposed approach, given the large number of aspects that characterize its complex technical products.