Search

Distributed dataflow systems such as Apache Spark or Apache Flink enable parallel, in-memory data processing on large clusters of commodity hardware. Consequently, the appropriate amount of memory to allocate to the cluster is a crucial consideration. In this paper, we analyze the challenge of efficient resource allocation for distributed data processing, focusing on memory. We emphasize that in-memory processing with in-memory data processing frameworks can undermine resource efficiency. Based on the findings of our trace data analysis, we compile requirements towards an automated solution for efficient cluster resource allocation., Comment: 4 pages, 3 Figures; ACM SSDBM 2023

Aggregated time series are generated effortlessly everywhere, e.g., "total confirmed covid-19 cases since 2019" and "total liquor sales over time." Understanding "how" and "why" these key performance indicators (KPI) evolve over time is critical to making data-informed decisions. Existing explanation engines focus on explaining one aggregated value or the difference between two relations. However, this falls short of explaining KPIs' continuous changes over time. Motivated by this, we propose TSEXPLAIN, a system that explains aggregated time series by surfacing the underlying evolving top contributors. Under the hood, we leverage prior works on two-relations diff as a building block and formulate a K-Segmentation problem to segment the time series such that each segment after segmentation shares consistent explanations, i.e., contributors. To quantify consistency in each segment, we propose a novel within-segment variance design that is explanation-aware; to derive the optimal K-Segmentation scheme, we develop an efficient dynamic programming algorithm. Experiments on synthetic and real-world datasets show that our explanation-aware segmentation can effectively identify evolving explanations for aggregated time series and outperform explanation-agnostic segmentation. Further, we proposed an optimal selection strategy of K and several optimizations to speed up TSEXPLAIN for interactive user experience, achieving up to 13X efficiency improvement., Comment: 17 pages; Accepted by ICDE 2023

It is widely anticipated that the use of health-related big data will enable further understanding and improvements in human health and wellbeing. Our current project, funded through the Dutch National Research Agenda, aims to explore the relationship between the development of diabetes and socio-economic factors such as lifestyle and health care utilization. The analysis involves combining data from the Maastricht Study (DMS), a prospective clinical study, and data collected by Statistics Netherlands (CBS) as part of its routine operations. However, a wide array of social, legal, technical, and scientific issues hinder the analysis. In this paper, we describe these challenges and our progress towards addressing them., Comment: 6 pages, 1 figure, preliminary result, project report

The collection of time series data increases as more monitoring and automation are being deployed. These deployments range in scale from an Internet of things (IoT) device located in a household to enormous distributed Cyber-Physical Systems (CPSs) producing large volumes of data at high velocity. To store and analyze these vast amounts of data, specialized Time Series Management Systems (TSMSs) have been developed to overcome the limitations of general purpose Database Management Systems (DBMSs) for times series management. In this paper, we present a thorough analysis and classification of TSMSs developed through academic or industrial research and documented through publications. Our classification is organized into categories based on the architectures observed during our analysis. In addition, we provide an overview of each system with a focus on the motivational use case that drove the development of the system, the functionality for storage and querying of time series a system implements, the components the system is composed of, and the capabilities of each system with regard to Stream Processing and Approximate Query Processing (AQP). Last, we provide a summary of research directions proposed by other researchers in the field and present our vision for a next generation TSMS., Comment: 20 Pages, 15 Figures, 2 Tables, Accepted for publication in IEEE TKDE

Efficient Nearest Neighbor (NN) search in high-dimensional spaces is a foundation of many multimedia retrieval systems. Because it offers low responses times, Product Quantization (PQ) is a popular solution. PQ compresses high-dimensional vectors into short codes using several sub-quantizers, which enables in-RAM storage of large databases. This allows fast answers to NN queries, without accessing the SSD or HDD. The key feature of PQ is that it can compute distances between short codes and high-dimensional vectors using cache-resident lookup tables. The efficiency of this technique, named Asymmetric Distance Computation (ADC), remains limited because it performs many cache accesses. In this paper, we introduce Quick ADC, a novel technique that achieves a 3 to 6 times speedup over ADC by exploiting Single Instruction Multiple Data (SIMD) units available in current CPUs. Efficiently exploiting SIMD requires algorithmic changes to the ADC procedure. Namely, Quick ADC relies on two key modifications of ADC: (i) the use 4-bit sub-quantizers instead of the standard 8-bit sub-quantizers and (ii) the quantization of floating-point distances. This allows Quick ADC to exceed the performance of state-of-the-art systems, e.g., it achieves a Recall@100 of 0.94 in 3.4 ms on 1 billion SIFT descriptors (128-bit codes)., Comment: 8 pages, 5 figures, published in Proceedings of ICMR'17, Bucharest, Romania, June 06-09, 2017

A geometrical pattern is a set of points with all pairwise distances (or, more generally, relative distances) specified. Finding matches to such patterns has applications to spatial data in seismic, astronomical, and transportation contexts. For example, a particularly interesting geometric pattern in astronomy is the Einstein cross, which is an astronomical phenomenon in which a single quasar is observed as four distinct sky objects (due to gravitational lensing) when captured by earth telescopes. Finding such crosses, as well as other geometric patterns, is a challenging problem as the potential number of sets of elements that compose shapes is exponentially large in the size of the dataset and the pattern. In this paper, we denote geometric patterns as constellation queries and propose algorithms to find them in large data applications. Our methods combine quadtrees, matrix multiplication, and unindexed join processing to discover sets of points that match a geometric pattern within some additive factor on the pairwise distances. Our distributed experiments show that the choice of composition algorithm (matrix multiplication or nested loops) depends on the freedom introduced in the query geometry through the distance additive factor. Three clearly identified blocks of threshold values guide the choice of the best composition algorithm. Finally, solving the problem for relative distances requires a novel continuous-to-discrete transformation. To the best of our knowledge this paper is the first to investigate constellation queries at scale.

Facility location queries identify the best locations to set up new facilities for providing service to its users. Majority of the existing works in this space assume that the user locations are static. Such limitations are too restrictive for planning many modern real-life services such as fuel stations, ATMs, convenience stores, cellphone base-stations, etc. that are widely accessed by mobile users. The placement of such services should, therefore, factor in the mobility patterns or trajectories of the users rather than simply their static locations. In this work, we introduce the TOPS (Trajectory-Aware Optimal Placement of Services) query that locates the best k sites on a road network. The aim is to optimize a wide class of objective functions defined over the user trajectories. We show that the problem is NP-hard and even the greedy heuristic with an approximation bound of (1-1/e) fails to scale on urban-scale datasets. To overcome this challenge, we develop a multi-resolution clustering based indexing framework called NetClus. Empirical studies on real road network trajectory datasets show that NetClus offers solutions that are comparable in terms of quality with those of the greedy heuristic, while having practical response times and low memory footprints. Additionally, the NetClus framework can absorb dynamic updates in mobility patterns, handle constraints such as site-costs and capacity, and existing services, thereby providing an effective solution for modern urban-scale scenarios., Comment: ICDE 2017 poster

Huge amount of data with both space and text information, e.g., geo-tagged tweets, is flooding on the Internet. Such spatio-textual data stream contains valuable information for millions of users with various interests on different keywords and locations. Publish/subscribe systems enable efficient and effective information distribution by allowing users to register continuous queries with both spatial and textual constraints. However, the explosive growth of data scale and user base has posed challenges to the existing centralized publish/subscribe systems for spatio-textual data streams. In this paper, we propose our distributed publish/subscribe system, called PS2Stream, which digests a massive spatio-textual data stream and directs the stream to target users with registered interests. Compared with existing systems, PS2Stream achieves a better workload distribution in terms of both minimizing the total amount of workload and balancing the load of workers. To achieve this, we propose a new workload distribution algorithm considering both space and text properties of the data. Additionally, PS2Stream supports dynamic load adjustments to adapt to the change of the workload, which makes PS2Stream adaptive. Extensive empirical evaluation, on commercial cloud computing platform with real data, validates the superiority of our system design and advantages of our techniques on system performance improvement., Comment: 13 pages, 16 figures, this paper has been accepted by ICDE2017

Subjective questions such as `does neymar dive', or `is clinton lying', or `is trump a fascist', are popular queries to web search engines, as can be seen by autocompletion suggestions on Google, Yahoo and Bing. In the era of cognitive computing, beyond search, they could be handled as hypotheses issued for evaluation. Our vision is to leverage on unstructured data and metadata of the rich user-generated multimedia that is often shared as material evidence in favor or against hypotheses in social media platforms. In this paper we present two preliminary experiments along those lines and discuss challenges for a cognitive computing system that collects material evidence from user-generated multimedia towards aggregating it into some form of collective decision on the hypothesis., Comment: 6 pages, 6 figures, 3 tables in Proc. of the 1st Workshop on Multimedia Support for Decision-Making Processes, at IEEE Intl. Symposium on Multimedia (ISM'16), San Jose, CA, 2016

A constantly growing amount of information is available through the web. Unfortunately, extracting useful content from this massive amount of data still remains an open issue. The lack of standard data models and structures forces developers to create adhoc solutions from the scratch. The figure of the expert is still needed in many situations where developers do not have the correct background knowledge. This forces developers to spend time acquiring the needed background from the expert. In other directions, there are promising solutions employing machine learning techniques. However, increasing accuracy requires an increase in system complexity that cannot be endured in many projects. In this work, we approach the web knowledge extraction problem using an expertcentric methodology. This methodology defines a set of configurable, extendible and independent components that permit the reutilisation of large pieces of code among projects. Our methodology differs from similar solutions in its expert-driven design. This design, makes it possible for subject-matter expert to drive the knowledge extraction for a given set of documents. Additionally, we propose the utilization of machine assisted solutions that guide the expert during this process. To demonstrate the capabilities of our methodology, we present a real use case scenario in which public procurement data is extracted from the web-based repositories of several public institutions across Europe. We provide insightful details about the challenges we had to deal with in this use case and additional discussions about how to apply our methodology.

Considering a group of users, each specifying individual preferences over categorical attributes, the problem of determining a set of objects that are objectively preferable by all users is challenging on two levels. First, we need to determine the preferable objects based on the categorical preferences for each user, and second we need to reconcile possible conflicts among users' preferences. A naive solution would first assign degrees of match between each user and each object, by taking into account all categorical attributes, and then for each object combine these matching degrees across users to compute the total score of an object. Such an approach, however, performs two series of aggregation, among categorical attributes and then across users, which completely obscure and blur individual preferences. Our solution, instead of combining individual matching degrees, is to directly operate on categorical attributes, and define an objective Pareto-based aggregation for group preferences. Building on our interpretation, we tackle two distinct but relevant problems: finding the Pareto-optimal objects, and objectively ranking objects with respect to the group preferences. To increase the efficiency when dealing with categorical attributes, we introduce an elegant transformation of categorical attribute values into numerical values, which exhibits certain nice properties and allows us to use well-known index structures to accelerate the solutions to the two problems. In fact, experiments on real and synthetic data show that our index-based techniques are an order of magnitude faster than baseline approaches, scaling up to millions of objects and thousands of users., Comment: To appear in Knowledge and Information Systems Journal (KAIS), Springer 2015

We describe FactorBase, a new SQL-based framework that leverages a relational database management system to support multi-relational model discovery. A multi-relational statistical model provides an integrated analysis of the heterogeneous and interdependent data resources in the database. We adopt the BayesStore design philosophy: statistical models are stored and managed as first-class citizens inside a database. Whereas previous systems like BayesStore support multi-relational inference, FactorBase supports multi-relational learning. A case study on six benchmark databases evaluates how our system supports a challenging machine learning application, namely learning a first-order Bayesian network model for an entire database. Model learning in this setting has to examine a large number of potential statistical associations across data tables. Our implementation shows how the SQL constructs in FactorBase facilitate the fast, modular, and reliable development of highly scalable model learning systems., Comment: 14 pages, 10 figures, 10 tables, Published on 2015 IEEE International Conference on Data Science and Advanced Analytics (IEEE DSAA'2015), Oct 19-21, 2015, Paris, France

The position we advocate in this paper is that relational algebra can provide a unified language for both representing and computing with statistical-relational objects, much as linear algebra does for traditional single-table machine learning. Relational algebra is implemented in the Structured Query Language (SQL), which is the basis of relational database management systems. To support our position, we have developed the FACTORBASE system, which uses SQL as a high-level scripting language for statistical-relational learning of a graphical model structure. The design philosophy of FACTORBASE is to manage statistical models as first-class citizens inside a database. Our implementation shows how our SQL constructs in FACTORBASE facilitate fast, modular, and reliable program development. Empirical evidence from six benchmark databases indicates that leveraging database system capabilities achieves scalable model structure learning., Comment: 3 pages, 1 figure, Position Paper of the Fifth International Workshop on Statistical Relational AI at UAI 2015

Formalisms for specifying statistical models, such as probabilistic-programming languages, typically consist of two components: a specification of a stochastic process (the prior), and a specification of observations that restrict the probability space to a conditional subspace (the posterior). Use cases of such formalisms include the development of algorithms in machine learning and artificial intelligence. We propose and investigate a declarative framework for specifying statistical models on top of a database, through an appropriate extension of Datalog. By virtue of extending Datalog, our framework offers a natural integration with the database, and has a robust declarative semantics. Our Datalog extension provides convenient mechanisms to include numerical probability functions; in particular, conclusions of rules may contain values drawn from such functions. The semantics of a program is a probability distribution over the possible outcomes of the input database with respect to the program; these outcomes are minimal solutions with respect to a related program with existentially quantified variables in conclusions. Observations are naturally incorporated by means of integrity constraints over the extensional and intensional relations. We focus on programs that use discrete numerical distributions, but even then the space of possible outcomes may be uncountable (as a solution can be infinite). We define a probability measure over possible outcomes by applying the known concept of cylinder sets to a probabilistic chase procedure. We show that the resulting semantics is robust under different chases. We also identify conditions guaranteeing that all possible outcomes are finite (and then the probability space is discrete). We argue that the framework we propose retains the purely declarative nature of Datalog, and allows for natural specifications of statistical models., Comment: 14 pages, 4 figures

Databases contain information about which relationships do and do not hold among entities. To make this information accessible for statistical analysis requires computing sufficient statistics that combine information from different database tables. Such statistics may involve any number of {\em positive and negative} relationships. With a naive enumeration approach, computing sufficient statistics for negative relationships is feasible only for small databases. We solve this problem with a new dynamic programming algorithm that performs a virtual join, where the requisite counts are computed without materializing join tables. Contingency table algebra is a new extension of relational algebra, that facilitates the efficient implementation of this M\"obius virtual join operation. The M\"obius Join scales to large datasets (over 1M tuples) with complex schemas. Empirical evaluation with seven benchmark datasets showed that information about the presence and absence of links can be exploited in feature selection, association rule mining, and Bayesian network learning., Comment: 11pages, 8 figures, 8 tables, CIKM'14,November 3--7, 2014, Shanghai, China

Differential privacy is achieved by the introduction of Laplacian noise in the response to a query, establishing a precise trade-off between the level of differential privacy and the accuracy of the database response (via the amount of noise introduced). Multiple queries may improve the accuracy but erode the privacy budget. We examine the case where we submit just a single counting query. We show that even in that case a Bayesian approach may be used to improve the accuracy for the same amount of noise injected, if we know the size of the database and the probability of a positive response to the query.

We study three-way joins on MapReduce. Joins are very useful in a multitude of applications from data integration and traversing social networks, to mining graphs and automata-based constructions. However, joins are expensive, even for moderate data sets; we need efficient algorithms to perform distributed computation of joins using clusters of many machines. MapReduce has become an increasingly popular distributed computing system and programming paradigm. We consider a state-of-the-art MapReduce multi-way join algorithm by Afrati and Ullman and show when it is appropriate for use on very large data sets. By providing a detailed experimental study, we demonstrate that this algorithm scales much better than what is suggested by the original paper. However, if the join result needs to be summarized or aggregated, as opposed to being only enumerated, then the aggregation step can be integrated into a cascade of two-way joins, making it more efficient than the other algorithm, and thus becomes the preferred solution., Comment: 6 pages

Traditional gazetteers are built and maintained by authoritative mapping agencies. In the age of Big Data, it is possible to construct gazetteers in a data-driven approach by mining rich volunteered geographic information (VGI) from the Web. In this research, we build a scalable distributed platform and a high-performance geoprocessing workflow based on the Hadoop ecosystem to harvest crowd-sourced gazetteer entries. Using experiments based on geotagged datasets in Flickr, we find that the MapReduce-based workflow running on the spatially enabled Hadoop cluster can reduce the processing time compared with traditional desktop-based operations by an order of magnitude. We demonstrate how to use such a novel spatial-computing infrastructure to facilitate gazetteer research. In addition, we introduce a provenance-based trust model for quality assurance. This work offers new insights on enriching future gazetteers with the use of Hadoop clusters, and makes contributions in connecting GIS to the cloud computing environment for the next frontier of Big Geo-Data analytics., Comment: 45 pages, 10 figures

This paper introduces ENFrame, a unified data processing platform for querying and mining probabilistic data. Using ENFrame, users can write programs in a fragment of Python with constructs such as bounded-range loops, list comprehension, aggregate operations on lists, and calls to external database engines. The program is then interpreted probabilistically by ENFrame. The realisation of ENFrame required novel contributions along several directions. We propose an event language that is expressive enough to succinctly encode arbitrary correlations, trace the computation of user programs, and allow for computation of discrete probability distributions of program variables. We exemplify ENFrame on three clustering algorithms: k-means, k-medoids, and Markov Clustering. We introduce sequential and distributed algorithms for computing the probability of interconnected events exactly or approximately with error guarantees. Experiments with k-medoids clustering of sensor readings from energy networks show orders-of-magnitude improvements of exact clustering using ENFrame over na\"ive clustering in each possible world, of approximate over exact, and of distributed over sequential algorithms., Comment: 12 pages

This paper outlines certain scenarios from the fields of astrophysics and fluid dynamics simulations which require high performance data warehouses that support array data type. A common feature of all these use cases is that subsetting and preprocessing the data on the server side (as far as possible inside the database server process) is necessary to avoid the client-server overhead and to minimize IO utilization. Analyzing and summarizing the requirements of the various fields help software engineers to come up with a comprehensive design of an array extension to relational database systems that covers a wide range of scientific applications. We also present a working implementation of an array data type for Microsoft SQL Server 2008 to support large-scale scientific applications. We introduce the design of the array type, results from a performance evaluation, and discuss the lessons learned from this implementation. The library can be downloaded from our website at http://voservices.net/sqlarray/

We propose a protocol for secure mining of association rules in horizontally distributed databases. The current leading protocol is that of Kantarcioglu and Clifton (TKDE 2004). Our protocol, like theirs, is based on the Fast Distributed Mining (FDM) algorithm of Cheung et al. (PDIS 1996), which is an unsecured distributed version of the Apriori algorithm. The main ingredients in our protocol are two novel secure multi-party algorithms --- one that computes the union of private subsets that each of the interacting players hold, and another that tests the inclusion of an element held by one player in a subset held by another. Our protocol offers enhanced privacy with respect to the protocol of Kantarcioglu and Clifton. In addition, it is simpler and is significantly more efficient in terms of communication rounds, communication cost and computational cost.

Mining association rules is a task of data mining, which extracts knowledge in the form of significant implication relation of useful items (objects) from a database. Mining multilevel association rules uses concept hierarchies, also called taxonomies and defined as relations of type 'is-a' between objects, to extract rules that items belong to different levels of abstraction. These rules are more useful, more refined and more interpretable by the user. Several algorithms have been proposed in the literature to discover the multilevel association rules. In this article, we are interested in the problem of discovering multi-level frequent itemsets under constraints, involving the user in the research process. We proposed a technique for modeling and interpretation of constraints in a context of use of concept hierarchies. Three approaches for discovering multi-level frequent itemsets under constraints were proposed and discussed: Basic approach, "Test and Generate" approach and Pruning based Approach., Comment: 20 pages

We mark up a corpus of LaTeX lecture notes semantically and expose them as Linked Data in XHTML+MathML+RDFa. Our application makes the resulting documents interactively browsable for students. Our ontology helps to answer queries from students and lecturers, and paves the path towards an integration of our corpus with external sites., Comment: 7th Extended Semantic Web Conference (http://www.eswc2010.org), Demo Track

The HL7 standard is widely used to exchange medical information electronically. As a part of the standard, HL7 defines scalar communication data types like physical quantity, point in time and concept descriptor but also complex types such as interval types, collection types and probabilistic types. Typical HL7 applications will store their communications in a database, resulting in a translation from HL7 concepts and types into database types. Since the data types were not designed to be implemented in a relational database server, this transition is cumbersome and fraught with programmer error. The purpose of this paper is two fold. First we analyze the HL7 version 3 data type definitions and define a number of conditions that must be met, for the data type to be suitable for implementation in a relational database. As a result of this analysis we describe a number of possible improvements in the HL7 specification. Second we describe an implementation in the PostgreSQL database server and show that the database server can effectively execute scientific calculations with units of measure, supports a large number of operations on time points and intervals, and can perform operations that are akin to a medical terminology server. Experiments on synthetic data show that the user defined types perform better than an implementation that uses only standard data types from the database server., Comment: 12 pages, 9 figures, 6 tables

Astrophysics has become a domain extremely rich of scientific data. Data mining tools are needed for information extraction from such large datasets. This asks for an approach to data management emphasizing the efficiency and simplicity of data access; efficiency is obtained using multidimensional access methods and simplicity is achieved by properly handling metadata. Moreover, clustering and classification techniques on large datasets pose additional requirements in terms of computation and memory scalability and interpretability of results. In this study we review some possible solutions.

We analyse the issues involved in the management and mining of astrophysical data. The traditional approach to data management in the astrophysical field is not able to keep up with the increasing size of the data gathered by modern detectors. An essential role in the astrophysical research will be assumed by automatic tools for information extraction from large datasets, i.e. data mining techniques, such as clustering and classification algorithms. This asks for an approach to data management based on data warehousing, emphasizing the efficiency and simplicity of data access; efficiency is obtained using multidimensional access methods and simplicity is achieved by properly handling metadata. Clustering and classification techniques, on large datasets, pose additional requirements: computational and memory scalability with respect to the data size, interpretability and objectivity of clustering or classification results. In this study we address some possible solutions., Comment: 10 pages, Latex

In the context of the ATLAS experiment there is growing evidence of the importance of different kinds of Meta-data including all the important details of the detector and data acquisition that are vital for the analysis of the acquired data. The Online BookKeeper (OBK) is a component of ATLAS online software that stores all information collected while running the experiment, including the Meta-data associated with the event acquisition, triggering and storage. The facilities for acquisition of control data within the on-line software framework, together with a full functional Web interface, make the OBK a powerful tool containing all information needed for event analysis, including an electronic log book. In this paper we explain how OBK plays a role as one of the main collectors and managers of Meta-data produced on-line, and we'll also focus on the Web facilities already available. The usage of the web interface as an electronic run logbook is also explained, together with the future extensions. We describe the technology used in OBK development and how we arrived at the present level explaining the previous experience with various DBMS technologies. The extensive performance evaluations that have been performed and the usage in the production environment of the ATLAS test beams are also analysed.

The configuration database is one of the vital systems in the BaBar on-line system. It provides services for the different parts of the data acquisition system and control system, which require run-time parameters. The original design and implementation of the configuration database played a significant role in the successful BaBar operations since the beginning of experiment. Recent additions to the design of the configuration database provide better means for the management of data and add new tools to simplify main configuration tasks. We describe the design of the configuration database, its implementation with the Objectivity/DB object-oriented database, and our experience collected during the years of operation., Comment: Talk from the 2003 Computing in High Energy and Nuclear Physics (CHEP03), La Jolla, Ca, USA, March 2003, 5 pages, 4 figures, PDF. PSN MOKT004

The design and implementation of two software systems introduced to improve the efficiency of offline analysis of event data taken with the ZEUS Detector at the HERA electron-proton collider at DESY are presented. Two different approaches were made, one using a set of event directories and the other using a tag database based on a commercial object-oriented database management system. These are described and compared. Both systems provide quick direct access to individual collision events in a sequential data store of several terabytes, and they both considerably improve the event analysis efficiency. In particular the tag database provides a very flexible selection mechanism and can dramatically reduce the computing time needed to extract small subsamples from the total event sample. Gains as large as a factor 20 have been obtained., Comment: Accepted for publication in Computer Physics Communications

The Microsoft TerraServer stores aerial and satellite images of the earth in a SQL Server Database served to the public via the Internet. It is the world's largest atlas, combining five terabytes of image data from the United States Geodetic Survey, Sovinformsputnik, and Encarta Virtual Globe. Internet browsers provide intuitive spatial and gazetteer interfaces to the data. The TerraServer is also an E-Commerce application. Users can buy the right to use the imagery using Microsoft Site Servers managed by the USGS and Aerial Images. This paper describes the TerraServer's design and implementation., Comment: Original file at http://research.microsoft.com/~gray/TerraServer_TR.doc