Yiming Ma

CAMAS: a citizen awareness system for crisis mitigation

1 ITR-Rescue, 5251 California Avenue, Suite 210, Irvine CA 92612-2815, Phone: 949-824-1147 http://www.itr-rescue.org 2 Union Bank Building, 400 Oceangate, Suite 1050, Long Beach CA 90802 Phone: 562-628-1675, http://imagecatinc.com ABSTRACT This demo paper provides a brief description of the intuition and design philosophy of CAMAS, one of the main testbeds being developed in the context of the RESCUE (Responding to the Unexpected) project [9]. The goal of our work is to enhance the mitigation capabilities of first responders in the event of a crisis by dramatically transforming their ability to collect, store, analyze, interpret, share and disseminate data. CAMAS specifically is a system designed to allow a variety of users, including the average citizen to report incidents and potentially hazardous situations, analyze and evaluate these reports and notify appropriate personnel for further action. The multidisciplinary approach incorporates a variety of information technologies: networks; distributed systems; databases; image and video processing; GIS; and machine learning, together with subjective information obtained through social science. Besides providing an overview of the CAMAS architecture, we describe the demonstration platform and specific experiments that will illustrate issues of event extraction, ranking, access control and visualization.

Index for fast retrieval of uncertain spatial point data

Location information gathered from a variety of sources in the form of sensor data, video streams, human observations, and so on, is often imprecise and uncertain and needs to be represented approximately. To represent such uncertain location information, the use of a probabilistic model that captures the imprecise location as a probability density function (pdf) has been recently proposed. The pdfs can be arbitrarily complex depending on the type of application and the source of imprecision. Hence, efficiently representing, storing and querying pdfs is a very challenging task. While the current state of the art indexing approaches treat the representation and storage of pdfs as a black box, in this paper, we take the challenge of representing and storing any complex pdf in an efficient way. We further develop techniques to index such pdfs to support the efficient processing of location queries. Our extensive experiments demonstrate that our indexing techniques significantly outperform the best existing solutions.

Modeling and querying uncertain spatial information for situational awareness applications

Situational awareness (SA) applications monitor the real world and the entities therein to support tasks such as rapid decision-making, reasoning, and analysis. Raw input about unfolding events may arrive from variety of sources in the form of sensor data, video streams, human observations, and so on, from which events of interest are extracted. Location is one of the most important attributes of events, useful for a variety of SA tasks. In this paper, we propose an approach to model and represent (potentially uncertain) event locations described by human reporters in the form of free text. We analyze several types of spatial queries of interest in SA applications. Our experimental evaluation demonstrates the effectiveness of our approach.

SAT: spatial awareness from textual input

Recent events (WTC attacks, Southeast Asia Tsunamis, Hurricane Katrina, London bombings) have illustrated the need for accurate and timely situational awareness tools in emergency response. Developing effective situational awareness (SA) systems has the potential to radically improve decision support in crises by improving the accuracy and reliability of the information available to the decision-makers. In an evolving crisis, raw situational information comes from a variety of sources in the form of situational reports, live radio transcripts, sensor data, video streams. Much of the data resides (or can be converted) in the form of free text, from which events of interest are extracted. Spatial or location information is one of the fundamental attributes of the events, and is useful for a variety of situational awareness (SA) tasks.

RAF: an activation framework for refining similarity queries using learning techniques

In numerous applications that deal with similarity search, a user may not have an exact specification of his information need and/or may not be able to formulate a query that exactly captures his notion of similarity. A promising approach to mitigate this problem is to enable the user to submit a rough approximation of the desired query and use relevance feedback on retrieved objects to refine the query. In this paper, we explore such a refinement strategy for a general class of structured similarity queries. Our approach casts the refinement problem as that of learning concepts using the tuples on which the user provides feedback as a labeled training set. Under this setup, similarity query refinement consists of two learning tasks: learning the structure of the query and learning the relative importance of query components. The paper develops machine learning approaches suitable for the two learning tasks. The primary contribution of the paper is the Refinement Activation Framework (RAF) that decides when each learner is invoked. Experimental analysis over many real life datasets shows that our strategy significantly outperforms existing approaches in terms of retrieval quality.

On-Demand Information Portals for Disaster Situations

This paper describes our work on developing technology for rapidly assembling information portals that provide integrated access to and analysis of information from multiple sources in the case of any disaster. Many recent disasters (the S.E. Asian Tsunamis, the London subway bombings, the Katrina hurricane, to name a few) have demonstrated that a lot of valuable information becomes available in the hours and days immediately following the disaster, and such information is indeed valuable to disaster managers or even citizens in their response. In this paper we describe our work on developing information portals for disasters in general; we describe many key information processing capabilities and challenges that we consider important in such portals and also describe our approach to developing such capabilities.

Integrating similarity retrieval and skyline exploration via relevance feedback

Similarity retrieval have been widely used in many practical search applications. A similarity query model can be viewed as a logical combination of a set of similarity predicates. A user can initialize a query model, but model parameters or the model itself may be inadequately specified. As a result, a retrieval system cannot guarantee that it has presented all the relevant tuples to the user. In this paper, we propose a framework that integrates the similarity retrieval and skyline exploration. Using the relevance feedback as a way to constrain the search space, our framework can intelligently explore only a necessary portion of data that contains all the relevant tuples. Our framework is also flexible enough to incorporate model refinement techniques to retrieving relevant results as early as possible.

A framework for refining similarity queries using learning techniques

In numerous applications that deal with similarity search, a user may not have an exact idea of his information need and/or may not be able to construct a query that exactly captures his notion of similarity. A promising approach to mitigate this problem is to enable the user to submit a rough approximation of the desired query and use the feedback on the relevance of the retrieved objects to refine the query. In this paper, we explore such a refinement strategy for a general class of SQL similarity queries. Our approach casts the refinement problem as that of learning concepts using examples. This is achieved by viewing the tuples on which a user provides feedback as a labeled training set for a learner. Under this setup, SQL query refinement consists of two learning tasks, namely learning the structure of the SQL query and learning the relative importance of the query components. The paper develops appropriate machine learning approaches suitable for these two learning tasks. The primary contribution of the paper is a general refinement framework that decides when each learner is invoked in order to quickly learn the user query. Experimental analyses over many real life datasets and queries show that our strategy outperforms the existing approaches significantly in terms of retrieval accuracy and query simplicity.