Aibek Musaev

LITMUS: A Multi-Service Composition System for Landslide Detection

Landslides are an illustrative example of multi-hazards, which can be caused by earthquakes, rainfalls and human activity among other reasons. Detection of landslides presents a significant challenge, since there are no physical sensors that would detect landslides directly. A more recent approach in detection of natural hazards, such as earthquakes, involves the use of social media. We propose a multi-service composition approach and describe LITMUS, which is a landslide detection service that combines data from both physical and social information services by filtering and then joining the information flow from those services based on their spatiotemporal features. Our results show that with such approach LITMUS detects 25 out of 27 landslides reported by USGS in December 2013 and 40 more landslide locations unreported by USGS during this period. LITMUS is a prototype tool that is used to investigate and implement research ideas in the area of disaster detection. We list some of the current work being done on refining the system that allows us to identify 137 landslide locations unreported by USGS during a more recent period of September 2014. Finally, we describe a live demonstration that displays landslide detection results on a web map in real-time.

Landslide Detection Service Based on Composition of Physical and Social Information Services

Social media have been used in the detection and management of natural hazards such as earthquakes. However, disasters often lead to other kinds of disasters, forming multi-hazards. Landslide is an illustrative example of a multi-hazard, which may be caused by earthquakes, rainfalls, water erosion, among other reasons. Detecting such multi-hazards is a significant challenge, since physical sensors designed for specific disasters are insufficient for multi-hazards. We describe LITMUS -- a landslide detection service based on a multi-service composition approach that combines data from both physical and social information services by filtering and then joining the information flow from those services based on their spatiotemporal features. Our results show that with such approach LITMUS detects 25 out of 27 landslides reported by USGS in December and 40 more landslides unreported by USGS. Also, LITMUS provides a live demonstration that displays results on a web map.

Toward a Real-Time Service for Landslide Detection: Augmented Explicit Semantic Analysis and Clustering Composition Approaches

The use of Social Media for event detection, such as detection of natural disasters, has gained a booming interest from research community as Social Media has become an immensely important source of real-time information. However, it poses a number of challenges with respect to high volume, noisy information and lack of geo-tagged data. Extraction of high quality information (e.g., Accurate locations of events) while maintaining good performance (e.g., Low latency) are the major problems. In this paper, we propose two approaches for tackling these issues: an augmented Explicit Semantic Analysis approach for rapid classification and a composition of clustering algorithms for location estimation. Our experiments demonstrate over 98% in precision, recall and F-measure when classifying Social Media data while producing a 20% improvement in location estimation due to clustering composition approach. We implement these approaches as part of the landslide detection service LITMUS, which is live and openly accessible for continued evaluation and use.

Detection of Damage and Failure Events of Critical Public Infrastructure using Social Sensor Big Data

Public infrastructure systems provide many of the services that are critical to the health, functioning, and security of society. Many of these infrastructures, however, lack continuous physical sensor monitoring to be able to detect failure events or damage that has occurred to these systems. We propose the use of social sensor big data to detect these events. We focus on two main infrastructure systems, transportation and energy, and use data from Twitter streams to detect damage to bridges, highways, gas lines, and power infrastructure. Through a three-step filtering approach and assignment to geographical cells, we are able to filter out noise in this data to produce relevant geolocated tweets identifying failure events. Applying the strategy to real-world data, we demonstrate the ability of our approach to utilize social sensor big data to detect damage and failure events in these critical public infrastructures.

Fast Text Classification Using Randomized Explicit Semantic Analysis

Document classification or document categorization is one of the most studied areas in computer science due to its importance. The problem is to assign a document using its text to one or more classes or categories from a predefined set. We propose a new approach for fast text classification using randomized explicit semantic analysis (RS-ESA). It is based on a state of the art approach for word sense disambiguation based on Wikipedia, the largest encyclopedia in existence. Our method reduces Wikipedia repository using a random sample approach resulting in a throughput, which is an order of magnitude faster than the original explicit semantic analysis. RS-ESA approach has been implemented as part of the LITMUS project due to a need in classifying data from Social Media into relevant and irrelevant items with respect to landslide as a natural disaster. We demonstrate that our approach achieves 96% precision when classifying Social Media landslide data collected in December 2014. We also demonstrate the genericity of the proposed approach by using it for separating factual texts from fictional based on Wikipedia articles and fan fiction stories, where we achieve 97% in precision.

Multi-hazard Detection by Integrating Social Media and Physical Sensors

Disaster Management is one of the most important functions of the government. FEMA and CDC are two examples of government agencies directly charged with handling disasters, whereas USGS is a scientific agency oriented towards disaster research. But regardless of the type or purpose, each of the mentioned agencies utilizes Social Media as part of its activities. One of the uses of Social Media is in detection of disasters, such as earthquakes. But disasters may lead to other kinds of disasters, forming multi-hazards such as landslides. Effective detection and management of multi-hazards cannot rely only on one information source. In this chapter, we describe and evaluate a prototype implementation of a landslide detection system LITMUS, which combines multiple physical sensors and Social Media to handle the inherent varied origins and composition of multi-hazards. Our results demonstrate that LITMUS detects more landslides than the ones reported by an authoritative source.

Landslide Information Service Based on Composition of Physical and Social Sensors

Abstract-Modern world data come from an increasing numberof sources, including data from physical sensors like weathersatellites and seismographs as well as social networks and weblogs. While progress has been made in the filtering of individualsocial networks, there are significant advantages in the integrationof big data from multiple sources. For physical events, theintegration of physical sensors and social network data canimprove filtering efficiency and quality of results beyond whatis feasible in each individual data stream.Disasters are representative physical events with real worldimpact. As illustration and demonstration, we have built theLITMUS landslide information service that combines data fromboth physical sensors and social networks in real-time. LITMUSfilters and combines reliable but indirect physical data with directreport social media data on landslides to achieve high quality andwide coverage of landslide information.

Information Diffusion Analysis of Rumor Dynamics over a Social-Interaction Based Model

Rumors may potentially cause undesirable effect such as the widespread panic in the general public. Especially, with the unprecedented growth of different types of social and enterprise networks, rumors could reach a larger audience than before. Many researchers have proposed different approaches to analyze and detect rumors in social networks. However, most of them either study on theoretical models without real data experiments or use content-based analysis and limited information diffusion analysis without fully considering social interactions. In this paper, we propose a social interaction based model FAST by taking four major properties of social interactions into account including familiarity, activeness, similarity, and trustworthiness. Also, we evaluate our model on real data from Sina Weibo (Twitter-like social network in China), which contains around 200 million tweets and 14 million Weibo users. Based on our model, we create a new metrics Fractional Directed Power Community Index (FD-PCI) derived from μ-PCI to identify the influential spreaders in social networks. FD-PCI shows better performance than conventional metrics such as K-core index and PageRank. Moreover, we obtain interesting influential features to detect rumors by the comparison between rumor and real news dynamics.

Towards Multilingual Automated Classification Systems

In this paper we propose and evaluate three approaches for automated classification of texts in over 60 languages without the need for a manually annotated dataset in those languages. All approaches are based on the randomized Explicit Semantic Analysis method using multilingual Wikipedia articles as their knowledge repository. We evaluate the proposed approaches by classifying a Twitter dataset in English and Portuguese into relevant and irrelevant items with respect to landslide as a natural disaster, where the highest achieved F1-score is 0.93. These approaches can be used in various applications where multilingual classification is needed, including multilingual disaster reporting using Social Media to improve coverage and increase confidence. As illustration, we present a demonstration that combines data from physical sensors and social networks to detect landslide events reported in English and Portuguese.

Detection of Damage and Failure Events of Road Infrastructure Using Social Media.

We study the problem of estimating the state of road infrastructure, which is the backbone of transportation system. Road infrastructure can suffer from various issues, including structural failures, such as potholes, and non-structural issues, such as broken traffic lights. However, it is infeasible to cover all roads with physical sensors for monitoring purposes. Instead, we propose to use social sensor big data to detect and estimate these issues based on the public’s activity. As a demonstration, we generate a map of detected road problems based on tweets. The map displays the currently detected hotspots, where for each hotspot we compute the overall sentiment provided by the public. In addition, we identify the peak of public activity during the evaluation period and investigate the key drivers of that peak. Finally, we analyze the most influential users using an extension of PageRank. The proposed approach adds a novel perspective on the state of road infrastructure and may be used to help guide decisions related to road infrastructure funding.