Ever since Russian trolls have been brought to light, their interference in the 2016 US Presidential elections has been monitored and studied. These Russian trolls employ fake accounts registered on several major social media sites to influence public opinion in other countries. Our work involves discovering patterns in these tweets and classifying them by training different machine learning models such as Support Vector Machines, Word2vec, Google BERT, and neural network models, and then applying them to several large Twitter datasets to compare the effectiveness of the different models. Two classification tasks are utilized for this purpose. The first one is used to classify any given tweet as either troll or non-troll tweet. The second model classifies specific tweets as coming from left trolls or right trolls, based on apparent extreme political orientations. On the given data sets, Google BERT provides the best results, with an accuracy of 89.4% for the left/right troll detector and 99% for the troll/non-troll detector. Temporal, geographic, and sentiment analyses were also performed and results were visualized.

Summary Stream processing systems are designed to analyze data arriving in real time and using continuous queries and respond when a specific event or sequence of events are detected. An important aspect of these systems is Streaming Analytics, which facilitates statistical calculations on continuous data within the stream. These systems must be designed to handle high volumes of data, be scalable, and accommodate a multitude of long‐lived concurrently running analytics. The challenges involved in the development of stream processing include on‐the‐fly transformation of data streams to match the query needs of users and the ability to model stream transformations to detect overlaps and possibilities for optimizations and to specify a methodology to deliver optimizations. In particular, this work focuses on exposing data stream application internals in order to detect reusable parts and then consolidate applications to optimize computational resource usage. The Streaming Data Analytics Model presented in this paper adopts a declarative approach that enables processing and manipulation of data streams in a simple manner while facilitating powerful optimizations necessary for managing high volumes of streaming data in real time. An evaluation is provided to demonstrate in both theoretical and quantitative aspects the high performance offered by our approach.

Prescription drug abuse is one of the fastest growing public health problems in the USA. To address this epidemic, a near real-time monitoring strategy, instead of one resorting to a retrospective health records, may improve detecting the prevalence and patterns of abuse of both illegal drugs and prescription medications. In this paper, our primary goals are to demonstrate the possibility of utilizing social media, e.g., Twitter, for automatic monitoring of illegal drug and prescription medication abuse. We use machine learning methods for an automatic classification that can identify tweets that are indicative of drug abuse. We collected tweets associated with well-known illegal and prescription drugs. We manually annotated 300 tweets that are likely to be related to drug abuse. Our experiment compares a set of classification algorithms, and a decision tree classifier J48, and the SVM outperform others for determining whether tweets contain signals of drug abuse. This automatic supervised classification study results illustrate the utility of Twitter in examining patterns of abuse, and show the feasibility of building the drug abuse detection system that can process large volume data from social media sources in a near real-time.

Many organizations require the ability to manage high-volume high-speed streaming data to perform analysis and other tasks in real-time. In this work, we present the Information Streaming Engine, a high-performance data stream processing system capable of scaling to high data volumes while maintaining very low-latency. The Information Streaming Engine adopts a declarative approach which enables processing and manipulation of data streams in a simple manner. Our evaluation demonstrates the high levels of performance achieved when compared to existing systems.

In the U.S., 80% of Medicare spending is for managing patients with multiple coexisting conditions. Predicting potentially correlated diseases for an individual patient and correlated disease progression paths are both important research tasks. For example, obese patients are at an increased risk for developing type-2 diabetes and hypertension. This correlation is called comorbidity relationship. Discovering the comorbidity relationships is complex and difficult due to the limited access to Electronic Health Records by privacy laws. In this paper, we present a framework called Social Data-based Prediction of Incidence and Trajectory to predict potential risks for medical conditions as well as its progression trajectory to identify the comorbidity path. The framework utilizes patients' publicly available social media data and presents a collaborative prediction model to predict the ranked list of potential comorbidity incidences, and a trajectory prediction model to reveal different paths of condition progression. The experimental results show that our framework is able to predict future conditions for online patients with a coverage value of 48% and 75% for a top-20 and a top-100 ranked list, respectively. For risk trajectory prediction, our framework is able to reveal each potential progression trajectory between any two conditions and infer the confidence of the future trajectory, given any observed condition. The predicted trajectories are validated with existing comorbidity relations from the medical literature.

The process of developing an ontology cannot be fully automated at the current state-of-the-art. However, leaving the tedious, time-consuming and error-prone task of ontology development entirely to humans has problems of its own, including limited staff budgets and semantic disagreements between experts. Thus, a hybrid computer/expert approach is advocated. The research challenge is how to minimize and optimally organize the task of the expert(s) while maximally leveraging the power of the computer and of existing computer-readable documents. The purpose of this paper is two-fold. First we present such a hybrid approach by describing a knowledge acquisition tool that we have developed. This tool makes use of an existing Bootstrap Ontology and proposes likely locations of concepts and semantic relationships, based on a text book, to a domain expert who can decide on them. The tool is attempting to minimize the number of interactions. Secondly we are proposing the notion of an augmented ontology specifically for pedagogical use. The application domain of this work is cyber-security education, but the ontology development methods are applicable to any educational topic.

Developing a domain ontology with concepts and relationships between them is a challenge, since knowledge engineering is a labor intensive process that can be a bottleneck and is often not scalable. Developing a cyber-security ontology is no exception. A security ontology can improve search for security learning resources that are scattered in different locations in different formats, since it can provide a common controlled vocabulary to annotate the resources with consistent semantics. In this paper, we present a bootstrapping method for developing a cyber-security ontology using both a security textbook index that provides a list of terms in the security domain and an existing security ontology as a scaffold. The bootstrapping approach automatically extracts the textbook index terms (concepts), derives a relationship to a concept in the security ontology for each and classifies them into the existing security ontology. The bootstrapping approach relies on the exact and approximate similarity matching of concepts as well as the category information obtained from external sources such as Wikipedia. The results show feasibility of our method to develop a more comprehensive and scalable cyber-security ontology with rich concepts from a textbook index. We provide criteria used to select a scaffold ontology among existing ontologies. The current approach can be improved by considering synonyms, deep searching in Wikipedia categories, and domain expert validation.

A business process can be developed as a composition of Web services provided by different service providers. These service providers may have their own policies and constraints for service provisioning and collaboration. In this paper, we focus on secure composition of services, specifically from t

Keyword-based search engines often return an unexpected number of results. Zero hits are naturally undesirable, while too many hits are likely to be overwhelming and of low precision. We present an approach for predicting the number of hits for a given set of query terms. Using word frequencies derived from a large corpus, we construct random samples of combinations of these words as search terms. Then we derive a correlation function between the computed probabilities of search terms and the observed hit counts for them. This regression function is used to predict the hit counts for a user's new searches, with the intention of avoiding information overload. We report the results of experiments with Google, Yahoo! and Bing to validate our methodology. We further investigate the monotonicity of search results for negative search terms by those three search engines.

Geospatial databases include any data with reference to geocoordinate information. The geospatial data can either be digital raster images that represent the data on the earth in the form of pixels or digital vector data that is primarily from satellites. Due to the fact that many of the high-resolution satellites are commercial in nature, uncontrolled dissemination of the high resolution imagery may cause severe threats to national security as well as personal privacy. The severity of the threats is even more significant when this information is combined with vector maps or other publicly available vector data. In this paper, we present a GeoSpatial Authorisation System (GSAS), which is based on a GeoSpatial Authorisation Model (GSAM), for specifying and enforcing access control policies that makes reference to the spatial regions and locational credentials. The specification of authorisations is based on the spatial and temporal attributes associated with the image data, resolution of the images, geospatial credentials associated with users and privilege modes including view, zoom-in, overlay, view-thumbnail, view-annotation, identify, animate and fly-by that are relevant for geospatial image data. We present the GSAS system and its functionalities.

The vision of the Semantic Web is to reduce manual discovery and usage of Web resources (documents and services) and to allow software agents to automatically identify these Web resources, integrate them and execute them for achieving the intended goals of the user. Such a composed Web service may be represented as a workflow, called service flow. Current studies of Web services are not sufficient for automatic composition. This paper presents different types of compositional knowledge required for Web service discovery and composition: syntactic, semantic and pragmatic knowledge. As a proof of concept, we have implemented our framework in a cardiovascular domain which requires advanced service discovery and composition across heterogeneous platforms of multiple organizations. Within this framework, we describe (1) How to represent the compositional knowledge, which plays a role in service discovery and composition, in DAML-S; (2) How heterogeneous medical services intemperate in a composed medical service flow. (3) To solve this knowledge level integration, we build on an ontology integration method called SEMIO (Semantic Interoperability).

Workflow systems are gaining importance as an infrastructure for automating inter-organizational interactions, such as those in Electronic Commerce. In such an environment, a centralized Workflow Management System is not desirable because: (i) it can be a performance bottleneck, and (ii) the system s are inherently distributed, heterogeneous, and autonomous in nature. Decentralized execution of inter-organizational workflows may raise a number of security issues including those related to conflict-of-interest among competing organizations. In this paper, we first provide an approach to realize decentralized workflow execution, in which the workflow is divided into partitions, called self-describing workflows, and handled by a light weight workflow management component, called workflow stub, located at each organizational agent. Second, we identify the limitations of the traditional workflow model with respect to expressing the various types of join dependencies and extend the traditional workflow model suitably. Distinguishing the different types of dependencies among tasks is essential in the efficient execution of self-describing workflows. Finally, we recognize that placing the task execution agents that belong to the same conflict-of-interest class in one self-describing workflow may lead to unfair, and in some cases, undesirable results, akin to being on the wrong side of the Chinese wall. Therefore, to address the conflict-of-interest issues that arise in competitive business environments, we propose a decentralized workflow Chinese wall security model. We propose a restrictive partitioning solution to enforce the proposed model.

The advent of commercial observation satellites in the new millennium provides unprecedented access to timely information, as they produce images of the Earth with the sharpness and quality previously available only from US, Russian, and French military satellites. Due to the fact that they are commercial in nature, a broad range of government agencies (including international), the news media, businesses, and nongovernmental organizations can gain access to this information. This may have grave implications on national security and personal privacy. Formal policies for prohibiting the release of imagery beyond a certain resolution, and notifying when an image crosses an international boundary or when such a request is made, are beginning to emerge. Access permissions in this environment are determined by both the spatial and temporal attributes of the data, such as location, resolution level, and the time of image download, as well as those of the user credentials. Since existing authorization models are not adequate to provide access control based on spatial and temporal attributes, We propose a geospatial data authorization model (GSAM). Unlike the traditional access control models where authorizations are specified using subjects and objects, authorizations in GSAM are specified using credential expressions and object expressions. GSAM supports privilege modes including view, zoom-in, download, overlay, identify, animate, and fly by, among others. We present our access control prototype system that enables subject, object as well as authorization specification via a Web-based interface. When an access request is made, the access control system computes the overlapping region of the authorization and the access request. The zoom-in and zoom-out requests can simply be made through a click of the mouse, and the appropriate authorizations will be evaluated when these access requests are made.

Current art uses metadata associated with satellite images to facilitate their retrieval from image repositories. Typical metadata are geographic location, time, and data type. Because the metadata do not indicate which regions within an image are obscured by clouds, retrieval with such metadata may produce an image within which the region of interest (ROI) for the user is not visible. We report a system that can automatically determine whether an ROI is visible in the image, and can incorporate this into the metadata for individual images to enhance searching capability. The goal is to annotate each image with metadata regarding a number of ROIs. An experiment with the system annotated 236 advanced very high resolution radiometer (AVHRR) images of the North Atlantic from a five-month viewing period with descriptors that expressed the visibility of an ROI centered on Long Island, NY. For ground truth, we used the classifications of three human subjects to determine visibility of the same region of interest, and labeled the ROI with the majority decision of the three subjects. Partial cloud cover made the human determination subjective, and resulted in disagreements among the subjects. Using randomly selected training subsets of the images, we found the two images whose regions were most like those in images for which the Long Island region was visible.

Workflow systems are gaining importance as an infrastructure for automating inter-organizational interactions, such as those in Electronic Commerce. Execution of inter-organiz-ational workflows may raise a number of security issues including those related to conflict-of-interest among competing organizations. Moreover, in such an environment, a centralized Workflow Management System is not desirable because: (i) it can be a performance bottleneck, and (ii) the systems are inherently distributed, heterogeneous and autonomous in nature. In this paper, we propose an approach to realize decentralized workflow execution, in which the workflow is divided into partitions called self-describing workflows, and handled by a light weight workflow management component, called workflow stub, located at each organizational agent. We argue that placing the task execution agents that belong to the same conflict-of-interest class in one self-describing workflow may lead to unfair, and in some cases, undesirable results, akin to being on the wrong side of the Chinese wall. We propose a Chinese wall security model for the decentralized workflow environment to resolve such problems, and a restrictive partitioning solution to enforce the proposed model.