Text analytics of social media: Sentiment analysis, event detection and summarization

In the last decade, large numbers of social media services have emerged and been widely used in people's daily life as important information sharing and acquisition tools. With a substantial amount of user-contributed text data on social media, it becomes a necessity to develop methods and tools for text analysis for this emerging data, in order to better utilize it to deliver meaningful information to users. ^ Previous work on text analytics in last several decades is mainly focused on traditional types of text like emails, news and academic literatures, and several critical issues to text data on social media have not been well explored: 1) how to detect sentiment from text on social media; 2) how to make use of social media's real-time nature; 3) how to address information overload for flexible information needs. ^ In this dissertation, we focus on these three problems. First, to detect sentiment of text on social media, we propose a non-negative matrix tri-factorization (tri-NMF) based dual active supervision method to minimize human labeling efforts for the new type of data. Second, to make use of social media's real-time nature, we propose approaches to detect events from text streams on social media. Third, to address information overload for flexible information needs, we propose two summarization framework, dominating set based summarization framework and learning-to-rank based summarization framework. The dominating set based summarization framework can be applied for different types of summarization problems, while the learning-to-rank based summarization framework helps utilize the existing training data to guild the new summarization tasks. In addition, we integrate these techneques in an application study of event summarization for sports games as an example of how to better utilize social media data. ^

Text Analytics of Social Media: Sentiment Analysis, Event Detection and Summarization

In the last decade, large numbers of social media services have emerged and been widely used in people's daily life as important information sharing and acquisition tools. With a substantial amount of user-contributed text data on social media, it becomes a necessity to develop methods and tools for text analysis for this emerging data, in order to better utilize it to deliver meaningful information to users. Previous work on text analytics in last several decades is mainly focused on traditional types of text like emails, news and academic literatures, and several critical issues to text data on social media have not been well explored: 1) how to detect sentiment from text on social media; 2) how to make use of social media's real-time nature; 3) how to address information overload for flexible information needs. In this dissertation, we focus on these three problems. First, to detect sentiment of text on social media, we propose a non-negative matrix tri-factorization (tri-NMF) based dual active supervision method to minimize human labeling efforts for the new type of data. Second, to make use of social media's real-time nature, we propose approaches to detect events from text streams on social media. Third, to address information overload for flexible information needs, we propose two summarization framework, dominating set based summarization framework and learning-to-rank based summarization framework. The dominating set based summarization framework can be applied for different types of summarization problems, while the learning-to-rank based summarization framework helps utilize the existing training data to guild the new summarization tasks. In addition, we integrate these techneques in an application study of event summarization for sports games as an example of how to better utilize social media data.

A multimedia indexing and retrieval framework for multimedia database systems

The main challenges of multimedia data retrieval lie in the effective mapping between low-level features and high-level concepts, and in the individual users' subjective perceptions of multimedia content. ^ The objectives of this dissertation are to develop an integrated multimedia indexing and retrieval framework with the aim to bridge the gap between semantic concepts and low-level features. To achieve this goal, a set of core techniques have been developed, including image segmentation, content-based image retrieval, object tracking, video indexing, and video event detection. These core techniques are integrated in a systematic way to enable the semantic search for images/videos, and can be tailored to solve the problems in other multimedia related domains. In image retrieval, two new methods of bridging the semantic gap are proposed: (1) for general content-based image retrieval, a stochastic mechanism is utilized to enable the long-term learning of high-level concepts from a set of training data, such as user access frequencies and access patterns of images. (2) In addition to whole-image retrieval, a novel multiple instance learning framework is proposed for object-based image retrieval, by which a user is allowed to more effectively search for images that contain multiple objects of interest. An enhanced image segmentation algorithm is developed to extract the object information from images. This segmentation algorithm is further used in video indexing and retrieval, by which a robust video shot/scene segmentation method is developed based on low-level visual feature comparison, object tracking, and audio analysis. Based on shot boundaries, a novel data mining framework is further proposed to detect events in soccer videos, while fully utilizing the multi-modality features and object information obtained through video shot/scene detection. ^ Another contribution of this dissertation is the potential of the above techniques to be tailored and applied to other multimedia applications. This is demonstrated by their utilization in traffic video surveillance applications. The enhanced image segmentation algorithm, coupled with an adaptive background learning algorithm, improves the performance of vehicle identification. A sophisticated object tracking algorithm is proposed to track individual vehicles, while the spatial and temporal relationships of vehicle objects are modeled by an abstract semantic model. ^

An epistemic event -based correlation scheme for pervasive network management

Computer networks produce tremendous amounts of event-based data that can be collected and managed to support an increasing number of new classes of pervasive applications. Examples of such applications are network monitoring and crisis management. Although the problem of distributed event-based management has been addressed in the non-pervasive settings such as the Internet, the domain of pervasive networks has its own characteristics that make these results non-applicable. Many of these applications are based on time-series data that possess the form of time-ordered series of events. Such applications also embody the need to handle large volumes of unexpected events, often modified on-the-fly, containing conflicting information, and dealing with rapidly changing contexts while producing results with low-latency. Correlating events across contextual dimensions holds the key to expanding the capabilities and improving the performance of these applications. This dissertation addresses this critical challenge. It establishes an effective scheme for complex-event semantic correlation. The scheme examines epistemic uncertainty in computer networks by fusing event synchronization concepts with belief theory. Because of the distributed nature of the event detection, time-delays are considered. Events are no longer instantaneous, but duration is associated with them. Existing algorithms for synchronizing time are split into two classes, one of which is asserted to provide a faster means for converging time and hence better suited for pervasive network management. Besides the temporal dimension, the scheme considers imprecision and uncertainty when an event is detected. A belief value is therefore associated with the semantics and the detection of composite events. This belief value is generated by a consensus among participating entities in a computer network. The scheme taps into in-network processing capabilities of pervasive computer networks and can withstand missing or conflicting information gathered from multiple participating entities. Thus, this dissertation advances knowledge in the field of network management by facilitating the full utilization of characteristics offered by pervasive, distributed and wireless technologies in contemporary and future computer networks.

An Epistemic Event-based Correlation Scheme for Pervasive Network Management

Computer networks produce tremendous amounts of event-based data that can be collected and managed to support an increasing number of new classes of pervasive applications. Examples of such applications are network monitoring and crisis management. Although the problem of distributed event-based management has been addressed in the non-pervasive settings such as the Internet, the domain of pervasive networks has its own characteristics that make these results non-applicable. Many of these applications are based on time-series data that possess the form of time-ordered series of events. Such applications also embody the need to handle large volumes of unexpected events, often modified on-the-fly, containing conflicting information, and dealing with rapidly changing contexts while producing results with low-latency. Correlating events across contextual dimensions holds the key to expanding the capabilities and improving the performance of these applications. This dissertation addresses this critical challenge. It establishes an effective scheme for complex-event semantic correlation. The scheme examines epistemic uncertainty in computer networks by fusing event synchronization concepts with belief theory. Because of the distributed nature of the event detection, time-delays are considered. Events are no longer instantaneous, but duration is associated with them. Existing algorithms for synchronizing time are split into two classes, one of which is asserted to provide a faster means for converging time and hence better suited for pervasive network management. Besides the temporal dimension, the scheme considers imprecision and uncertainty when an event is detected. A belief value is therefore associated with the semantics and the detection of composite events. This belief value is generated by a consensus among participating entities in a computer network. The scheme taps into in-network processing capabilities of pervasive computer networks and can withstand missing or conflicting information gathered from multiple participating entities. Thus, this dissertation advances knowledge in the field of network management by facilitating the full utilization of characteristics offered by pervasive, distributed and wireless technologies in contemporary and future computer networks.

Exploring Hidden Coherent Feature Groups and Temporal Semantics for Multimedia Big Data Analysis

Thanks to the advanced technologies and social networks that allow the data to be widely shared among the Internet, there is an explosion of pervasive multimedia data, generating high demands of multimedia services and applications in various areas for people to easily access and manage multimedia data. Towards such demands, multimedia big data analysis has become an emerging hot topic in both industry and academia, which ranges from basic infrastructure, management, search, and mining to security, privacy, and applications. Within the scope of this dissertation, a multimedia big data analysis framework is proposed for semantic information management and retrieval with a focus on rare event detection in videos. The proposed framework is able to explore hidden semantic feature groups in multimedia data and incorporate temporal semantics, especially for video event detection. First, a hierarchical semantic data representation is presented to alleviate the semantic gap issue, and the Hidden Coherent Feature Group (HCFG) analysis method is proposed to capture the correlation between features and separate the original feature set into semantic groups, seamlessly integrating multimedia data in multiple modalities. Next, an Importance Factor based Temporal Multiple Correspondence Analysis (i.e., IF-TMCA) approach is presented for effective event detection. Specifically, the HCFG algorithm is integrated with the Hierarchical Information Gain Analysis (HIGA) method to generate the Importance Factor (IF) for producing the initial detection results. Then, the TMCA algorithm is proposed to efficiently incorporate temporal semantics for re-ranking and improving the final performance. At last, a sampling-based ensemble learning mechanism is applied to further accommodate the imbalanced datasets. In addition to the multimedia semantic representation and class imbalance problems, lack of organization is another critical issue for multimedia big data analysis. In this framework, an affinity propagation-based summarization method is also proposed to transform the unorganized data into a better structure with clean and well-organized information. The whole framework has been thoroughly evaluated across multiple domains, such as soccer goal event detection and disaster information management.

An automated method for the determination of electroreception thresholds in weakly electric fish

Weakly electric fish produce a dual function electric signal that makes them ideal models for the study of sensory computation and signal evolution. This signal, the electric organ discharge (EOD), is used for communication and navigation. In some families of gymnotiform electric fish, the EOD is a dynamic signal that increases in amplitude during social interactions. Amplitude increase could facilitate communication by increasing the likelihood of being sensed by others or by impressing prospective mates or rivals. Conversely, by increasing its signal amplitude a fish might increase its sensitivity to objects by lowering its electrolocation detection threshold. To determine how EOD modulations elicited in the social context affect electrolocation, I developed an automated and fast method for measuring electroreception thresholds using a classical conditioning paradigm. This method employs a moving shelter tube, which these fish occupy at rest during the day, paired with an electrical stimulus. A custom built and programmed robotic system presents the electrical stimulus to the fish, slides the shelter tube requiring them to follow, and records video of their movements. I trained the electric fish of the genus Sternopygus was trained to respond to a resistive stimulus on this apparatus in 2 days. The motion detection algorithm correctly identifies the responses 91% of the time, with a false positive rate of only 4%. This system allows for a large number of trials, decreasing the amount of time needed to determine behavioral electroreception thresholds. This novel method enables the evaluation the evolutionary interplay between two conflicting sensory forces, social communication and navigation.

Design strategies for an artificial neural network based algorithm for automatic incident detection on major arterial streets

Traffic incidents are non-recurring events that can cause a temporary reduction in roadway capacity. They have been recognized as a major contributor to traffic congestion on our nation’s highway systems. To alleviate their impacts on capacity, automatic incident detection (AID) has been applied as an incident management strategy to reduce the total incident duration. AID relies on an algorithm to identify the occurrence of incidents by analyzing real-time traffic data collected from surveillance detectors. Significant research has been performed to develop AID algorithms for incident detection on freeways; however, similar research on major arterial streets remains largely at the initial stage of development and testing. This dissertation research aims to identify design strategies for the deployment of an Artificial Neural Network (ANN) based AID algorithm for major arterial streets. A section of the US-1 corridor in Miami-Dade County, Florida was coded in the CORSIM microscopic simulation model to generate data for both model calibration and validation. To better capture the relationship between the traffic data and the corresponding incident status, Discrete Wavelet Transform (DWT) and data normalization were applied to the simulated data. Multiple ANN models were then developed for different detector configurations, historical data usage, and the selection of traffic flow parameters. To assess the performance of different design alternatives, the model outputs were compared based on both detection rate (DR) and false alarm rate (FAR). The results show that the best models were able to achieve a high DR of between 90% and 95%, a mean time to detect (MTTD) of 55-85 seconds, and a FAR below 4%. The results also show that a detector configuration including only the mid-block and upstream detectors performs almost as well as one that also includes a downstream detector. In addition, DWT was found to be able to improve model performance, and the use of historical data from previous time cycles improved the detection rate. Speed was found to have the most significant impact on the detection rate, while volume was found to contribute the least. The results from this research provide useful insights on the design of AID for arterial street applications.

Design Strategies for an Artificial Neural Network Based Algorithm for Automatic Incident Detection on Major Arterial Streets

Traffic incidents are non-recurring events that can cause a temporary reduction in roadway capacity. They have been recognized as a major contributor to traffic congestion on our national highway systems. To alleviate their impacts on capacity, automatic incident detection (AID) has been applied as an incident management strategy to reduce the total incident duration. AID relies on an algorithm to identify the occurrence of incidents by analyzing real-time traffic data collected from surveillance detectors. Significant research has been performed to develop AID algorithms for incident detection on freeways; however, similar research on major arterial streets remains largely at the initial stage of development and testing. This dissertation research aims to identify design strategies for the deployment of an Artificial Neural Network (ANN) based AID algorithm for major arterial streets. A section of the US-1 corridor in Miami-Dade County, Florida was coded in the CORSIM microscopic simulation model to generate data for both model calibration and validation. To better capture the relationship between the traffic data and the corresponding incident status, Discrete Wavelet Transform (DWT) and data normalization were applied to the simulated data. Multiple ANN models were then developed for different detector configurations, historical data usage, and the selection of traffic flow parameters. To assess the performance of different design alternatives, the model outputs were compared based on both detection rate (DR) and false alarm rate (FAR). The results show that the best models were able to achieve a high DR of between 90% and 95%, a mean time to detect (MTTD) of 55-85 seconds, and a FAR below 4%. The results also show that a detector configuration including only the mid-block and upstream detectors performs almost as well as one that also includes a downstream detector. In addition, DWT was found to be able to improve model performance, and the use of historical data from previous time cycles improved the detection rate. Speed was found to have the most significant impact on the detection rate, while volume was found to contribute the least. The results from this research provide useful insights on the design of AID for arterial street applications.

A framework for automatic feature extraction from airborne light detection and ranging data

Recent advances in airborne Light Detection and Ranging (LIDAR) technology allow rapid and inexpensive measurements of topography over large areas. Airborne LIDAR systems usually return a 3-dimensional cloud of point measurements from reflective objects scanned by the laser beneath the flight path. This technology is becoming a primary method for extracting information of different kinds of geometrical objects, such as high-resolution digital terrain models (DTMs), buildings and trees, etc. In the past decade, LIDAR gets more and more interest from researchers in the field of remote sensing and GIS. Compared to the traditional data sources, such as aerial photography and satellite images, LIDAR measurements are not influenced by sun shadow and relief displacement. However, voluminous data pose a new challenge for automated extraction the geometrical information from LIDAR measurements because many raster image processing techniques cannot be directly applied to irregularly spaced LIDAR points. ^ In this dissertation, a framework is proposed to filter out information about different kinds of geometrical objects, such as terrain and buildings from LIDAR automatically. They are essential to numerous applications such as flood modeling, landslide prediction and hurricane animation. The framework consists of several intuitive algorithms. Firstly, a progressive morphological filter was developed to detect non-ground LIDAR measurements. By gradually increasing the window size and elevation difference threshold of the filter, the measurements of vehicles, vegetation, and buildings are removed, while ground data are preserved. Then, building measurements are identified from no-ground measurements using a region growing algorithm based on the plane-fitting technique. Raw footprints for segmented building measurements are derived by connecting boundary points and are further simplified and adjusted by several proposed operations to remove noise, which is caused by irregularly spaced LIDAR measurements. To reconstruct 3D building models, the raw 2D topology of each building is first extracted and then further adjusted. Since the adjusting operations for simple building models do not work well on 2D topology, 2D snake algorithm is proposed to adjust 2D topology. The 2D snake algorithm consists of newly defined energy functions for topology adjusting and a linear algorithm to find the minimal energy value of 2D snake problems. Data sets from urbanized areas including large institutional, commercial, and small residential buildings were employed to test the proposed framework. The results demonstrated that the proposed framework achieves a very good performance. ^

Ensemble fuzzy belief intrusion detection design

With the rapid growth of the Internet, computer attacks are increasing at a fast pace and can easily cause millions of dollar in damage to an organization. Detecting these attacks is an important issue of computer security. There are many types of attacks and they fall into four main categories, Denial of Service (DoS) attacks, Probe, User to Root (U2R) attacks, and Remote to Local (R2L) attacks. Within these categories, DoS and Probe attacks continuously show up with greater frequency in a short period of time when they attack systems. They are different from the normal traffic data and can be easily separated from normal activities. On the contrary, U2R and R2L attacks are embedded in the data portions of the packets and normally involve only a single connection. It becomes difficult to achieve satisfactory detection accuracy for detecting these two attacks. Therefore, we focus on studying the ambiguity problem between normal activities and U2R/R2L attacks. The goal is to build a detection system that can accurately and quickly detect these two attacks. In this dissertation, we design a two-phase intrusion detection approach. In the first phase, a correlation-based feature selection algorithm is proposed to advance the speed of detection. Features with poor prediction ability for the signatures of attacks and features inter-correlated with one or more other features are considered redundant. Such features are removed and only indispensable information about the original feature space remains. In the second phase, we develop an ensemble intrusion detection system to achieve accurate detection performance. The proposed method includes multiple feature selecting intrusion detectors and a data mining intrusion detector. The former ones consist of a set of detectors, and each of them uses a fuzzy clustering technique and belief theory to solve the ambiguity problem. The latter one applies data mining technique to automatically extract computer users’ normal behavior from training network traffic data. The final decision is a combination of the outputs of feature selecting and data mining detectors. The experimental results indicate that our ensemble approach not only significantly reduces the detection time but also effectively detect U2R and R2L attacks that contain degrees of ambiguous information.

Motion correction algorithm of lung tumors for respiratory gated PET images

Respiratory gating in lung PET imaging to compensate for respiratory motion artifacts is a current research issue with broad potential impact on quantitation, diagnosis and clinical management of lung tumors. However, PET images collected at discrete bins can be significantly affected by noise as there are lower activity counts in each gated bin unless the total PET acquisition time is prolonged, so that gating methods should be combined with imaging-based motion correction and registration methods. The aim of this study was to develop and validate a fast and practical solution to the problem of respiratory motion for the detection and accurate quantitation of lung tumors in PET images. This included: (1) developing a computer-assisted algorithm for PET/CT images that automatically segments lung regions in CT images, identifies and localizes lung tumors of PET images; (2) developing and comparing different registration algorithms which processes all the information within the entire respiratory cycle and integrate all the tumor in different gated bins into a single reference bin. Four registration/integration algorithms: Centroid Based, Intensity Based, Rigid Body and Optical Flow registration were compared as well as two registration schemes: Direct Scheme and Successive Scheme. Validation was demonstrated by conducting experiments with the computerized 4D NCAT phantom and with a dynamic lung-chest phantom imaged using a GE PET/CT System. Iterations were conducted on different size simulated tumors and different noise levels. Static tumors without respiratory motion were used as gold standard; quantitative results were compared with respect to tumor activity concentration, cross-correlation coefficient, relative noise level and computation time. Comparing the results of the tumors before and after correction, the tumor activity values and tumor volumes were closer to the static tumors (gold standard). Higher correlation values and lower noise were also achieved after applying the correction algorithms. With this method the compromise between short PET scan time and reduced image noise can be achieved, while quantification and clinical analysis become fast and precise.

Field programmable gate array based target detection and gesture recognition

The move from Standard Definition (SD) to High Definition (HD) represents a six times increases in data, which needs to be processed. With expanding resolutions and evolving compression, there is a need for high performance with flexible architectures to allow for quick upgrade ability. The technology advances in image display resolutions, advanced compression techniques, and video intelligence. Software implementation of these systems can attain accuracy with tradeoffs among processing performance (to achieve specified frame rates, working on large image data sets), power and cost constraints. There is a need for new architectures to be in pace with the fast innovations in video and imaging. It contains dedicated hardware implementation of the pixel and frame rate processes on Field Programmable Gate Array (FPGA) to achieve the real-time performance. ^ The following outlines the contributions of the dissertation. (1) We develop a target detection system by applying a novel running average mean threshold (RAMT) approach to globalize the threshold required for background subtraction. This approach adapts the threshold automatically to different environments (indoor and outdoor) and different targets (humans and vehicles). For low power consumption and better performance, we design the complete system on FPGA. (2) We introduce a safe distance factor and develop an algorithm for occlusion occurrence detection during target tracking. A novel mean-threshold is calculated by motion-position analysis. (3) A new strategy for gesture recognition is developed using Combinational Neural Networks (CNN) based on a tree structure. Analysis of the method is done on American Sign Language (ASL) gestures. We introduce novel point of interests approach to reduce the feature vector size and gradient threshold approach for accurate classification. (4) We design a gesture recognition system using a hardware/ software co-simulation neural network for high speed and low memory storage requirements provided by the FPGA. We develop an innovative maximum distant algorithm which uses only 0.39% of the image as the feature vector to train and test the system design. Database set gestures involved in different applications may vary. Therefore, it is highly essential to keep the feature vector as low as possible while maintaining the same accuracy and performance^

Event mining for system and service management

Modern IT infrastructures are constructed by large scale computing systems and administered by IT service providers. Manually maintaining such large computing systems is costly and inefficient. Service providers often seek automatic or semi-automatic methodologies of detecting and resolving system issues to improve their service quality and efficiency. This dissertation investigates several data-driven approaches for assisting service providers in achieving this goal. The detailed problems studied by these approaches can be categorized into the three aspects in the service workflow: 1) preprocessing raw textual system logs to structural events; 2) refining monitoring configurations for eliminating false positives and false negatives; 3) improving the efficiency of system diagnosis on detected alerts. Solving these problems usually requires a huge amount of domain knowledge about the particular computing systems. The approaches investigated by this dissertation are developed based on event mining algorithms, which are able to automatically derive part of that knowledge from the historical system logs, events and tickets. ^ In particular, two textual clustering algorithms are developed for converting raw textual logs into system events. For refining the monitoring configuration, a rule based alert prediction algorithm is proposed for eliminating false alerts (false positives) without losing any real alert and a textual classification method is applied to identify the missing alerts (false negatives) from manual incident tickets. For system diagnosis, this dissertation presents an efficient algorithm for discovering the temporal dependencies between system events with corresponding time lags, which can help the administrators to determine the redundancies of deployed monitoring situations and dependencies of system components. To improve the efficiency of incident ticket resolving, several KNN-based algorithms that recommend relevant historical tickets with resolutions for incoming tickets are investigated. Finally, this dissertation offers a novel algorithm for searching similar textual event segments over large system logs that assists administrators to locate similar system behaviors in the logs. Extensive empirical evaluation on system logs, events and tickets from real IT infrastructures demonstrates the effectiveness and efficiency of the proposed approaches.^

Ensemble Fuzzy Belief Intrusion Detection Design

With the rapid growth of the Internet, computer attacks are increasing at a fast pace and can easily cause millions of dollar in damage to an organization. Detecting these attacks is an important issue of computer security. There are many types of attacks and they fall into four main categories, Denial of Service (DoS) attacks, Probe, User to Root (U2R) attacks, and Remote to Local (R2L) attacks. Within these categories, DoS and Probe attacks continuously show up with greater frequency in a short period of time when they attack systems. They are different from the normal traffic data and can be easily separated from normal activities. On the contrary, U2R and R2L attacks are embedded in the data portions of the packets and normally involve only a single connection. It becomes difficult to achieve satisfactory detection accuracy for detecting these two attacks. Therefore, we focus on studying the ambiguity problem between normal activities and U2R/R2L attacks. The goal is to build a detection system that can accurately and quickly detect these two attacks. In this dissertation, we design a two-phase intrusion detection approach. In the first phase, a correlation-based feature selection algorithm is proposed to advance the speed of detection. Features with poor prediction ability for the signatures of attacks and features inter-correlated with one or more other features are considered redundant. Such features are removed and only indispensable information about the original feature space remains. In the second phase, we develop an ensemble intrusion detection system to achieve accurate detection performance. The proposed method includes multiple feature selecting intrusion detectors and a data mining intrusion detector. The former ones consist of a set of detectors, and each of them uses a fuzzy clustering technique and belief theory to solve the ambiguity problem. The latter one applies data mining technique to automatically extract computer users’ normal behavior from training network traffic data. The final decision is a combination of the outputs of feature selecting and data mining detectors. The experimental results indicate that our ensemble approach not only significantly reduces the detection time but also effectively detect U2R and R2L attacks that contain degrees of ambiguous information.