Computer Science

Hardware specialization is a promising direction for the future of digital computing. Reconfigurable technologies enable hardware specialization with modest non-recurring engineering cost. In this paper, we use FPGAs to evaluate the benefits of building specialized hardware for numerical kernels found in scientific applications. In order to properly evaluate performance, we not only compare Intel Arria 10 and Xilinx U280 performance against Intel Xeon, Intel Xeon Phi, and NVIDIA V100 GPUs, but we also extend the Empirical Roofline Toolkit (ERT) to FPGAs in order to assess our results in terms of the Roofline Model. Although FPGA performance is known to be far less than that of a GPU, we also benchmark the energy efficiency of each platform for the scientific kernels comparing to microbenchmark and technological limits. Results show that while FPGAs struggle to compete in absolute terms with GPUs on memory-and compute-intensive kernels, they require far less power and can deliver nearly the same energy efficiency.

Many non-pharmaceutical interventions (NPI) have been implemented in several countries to prevent and mitigate the COVID-19 pandemic. In spite of the emergence of vaccines that mitigate illness and death, no effective measures exist to control the contagion. Testing, tracing, and isolation (TTI) in conjunction with disciplined social distancing were expected to achieve negligible-contagion levels and avoid, for example, superspreading events 1 . However, the logic sustaining these strategies has proven insufficient and a commitment to implementing them leads to controversial ethical challenges. This paper focuses on the effectiveness of contagion control policies that are independent of the ethical challenges that the chosen strategies pose, and of the complex realities of different countries, that are applicable both country-wide and at the local level. Even more relevant is that our performance measure indicators are given by variables that provide an early warning as to when the epidemic might expand, despite some temporal case decrements. We show that if case reduction is not sustained for more than 7-14 days, the risk growth is considerable, warranting immediate attention and action. To draw these conclusions, our research does not only follow the contagion change but its velocity, and applies these indicators to a sample of prototypical countries, finding strong evidence for our conclusions. The latter are independent of different applied NPI and country differences, suggesting that they can be used as "universal" indicators for epidemic control policies. Our main conclusion is that if the spread reduction is lost during one, or at most two weeks, immediate measures have to be implemented to avoid undesirable scenarios. 45 we use tools developed by Isaac Newton during a long 46 past pandemic. To do so, we consider, in addition to 47 the number of cases and their time evolution, how the 48 increase or decrease of cases varies in time. In other 49 words, and in close analogy to the dynamics of motion, 50 we incorporate in our model the speed of change of the 51 time evolution of the number of infections.

Wireless Sensor Networks (WSN) employ duty-cycles as the main strategy to save energy. The sleep-delay arises as a drawback from this operating mode, increasing end-to-end latency. This paper presents Split-MAC: a low-energy, low-latency asynchronous protocol for WSNs. It combines medium access control (MAC) and routing functions. Some strategies are employed by Split-MAC to achieve low latency: an anycast communication pattern, a medium reservation scheme to transmit data in a pipelined fashion and simultaneous transmission of data segments. Simulations were performed to compare our protocol with others from literature. Split-MAC outperforms all other tested protocols. There is at least 30% decrease in latency. Resumo. Redes de Sensores Sem Fio (RSSF) usam ciclos de trabalho como estratégia para economizar energia. O atraso de propagac ¸ão (sleep-delay) é uma consequência negativa desse modo de operac ¸ão, aumentando a latência fim-afim. Apresenta-se aqui o Split-MAC: um protocolo assíncrono de baixa energia e latência reduzida para RSSFs. Ele reúne func ¸ões de controle de acesso ao meio e de roteamento, e emprega estratégias para obter baixa latência: comunicac ¸ão anycast, um esquema de reserva de canal para transmitir dados de maneira encadeada e transmissão simultânea de segmentos de dados. Simulac ¸ões foram realizadas para comparar este protocolo com outros da literatura. O Split-MAC demonstrou superioridade frente aos outros protocolos testados, reduzindo a latência em pelo menos 30% .

Redes de Sensores Sem Fio (RSSF) usam ciclos de trabalho como estratégia para economizar energia. O atraso de propagação (sleep-delay) é uma consequência negativa desse modo de operação, aumentando a latência fim-a-fim. Apresenta-se aqui o Split-MAC: um protocolo assíncrono de baixa energia e latência reduzida para RSSFs. Ele reúne funções de controle de acesso ao meio e de roteamento, e emprega estratégias para obter baixa latência: comunicação anycast, um esquema de reserva de canal para transmitir dados de maneira encadeada e transmissão simultânea de segmentos de dados. Simulações foram realizadas para comparar este protocolo com outros da literatura. O Split-MAC demonstrou superioridade frente aos outros protocolos testados, reduzindo a latência em pelo menos 30%.

Hadoop Remote Procedure Call (RPC) is increasingly used in many data centers of known technology companies such as Facebook and Yahoo together with other data middleware centers like MapReduce, HDFS (Hadoop Distributed File System) and Hbase. Due to its simplicity, efficiency, high performance use of RPC systems, it is essential to achieve dense data storage and low latency and high throughput. This paper describes how to improve data storage, throughput, and latency performance for big data. E.Sivaraman Dr.R.Manickachezian(2014) said systems are becoming increasingly data-intensive due to the explosion of data and the need to process it, therefore ,the application performance is vital to storage management in data-intensive computing systems. The author mentioned above highligted that researchers believe that applications sensitive areas are advancing in storing large volume of data. As data volumes continue to grow, the movement of data from memory to storage systems has become a critical performance bottleneck for many data centers. However, data management on such multi-server storage systems is yet under studies to ascertain on how to utilize various storages for efficient data movement is an important research topic in RPC data perception. In this research, we propose an adoptive Heartbeat system, data-aware distribution that efficiently manages storage systems to coordinate multiple concurrent data-intensive applications. Extensive experiments are conducted and the results show that this method can significantly improve the network throughput performance and data-intensive storage of Hadoop in HDFS clusters. The thesis discusses the background of RPC, huge data application storage systems, related work and technologies, Hadoop installation, adaptive model mechanisms, evaluation and performance analysis of big data architecture and its server interfaces. The reason to focus on storage improvements is because of the great need to store information globally. A large class of communication-intensive distributed applications and software components have been ported to virtual machines, such as highperformance storage systems. This method dynamically adjusts the RPC configuration between the NameNode and the DataNode by sensing the data characters stored in the DataNode. This method can effectively reduce the processing pressure of the NameNode and improve the network throughput generated by the information transmission between the NameNode and the DataNode.

Wireless sensor networks (WSNs) are networks of distributed autonomous devices that can sense or monitor physical or environmental conditions cooperatively. WSNs face many challenges, mainly caused by communication failures, storage and computational constraints and limited power supply. Paradigms of computational intelligence (CI) have been successfully used in recent years to address various challenges such as data aggregation and fusion, energy aware routing, task scheduling, security, optimal deployment and localization. CI provides adaptive mechanisms that exhibit intelligent behavior in complex and dynamic environments like WSNs. CI brings about flexibility, autonomous behavior, and robustness against topology changes, communication failures and scenario changes. However, WSN developers are usually not or not completely aware of the potential CI algorithms offer. On the other side, CI researchers are not familiar with all real problems and subtle requirements of WSNs. This mismatch makes collaboration and development difficult. This paper intends to close this gap and foster collaboration by offering a detailed introduction to WSNs and their properties. An extensive survey of CI applications to various problems in WSNs from various research areas and publication venues is presented in the paper. Besides, a discussion on advantages and disadvantages of CI algorithms over traditional WSN solutions is offered. In addition, a general evaluation of CI algorithms is presented, which will serve as a guide for using CI algorithms for WSNs.

Despite recent advances in diagnostic radiology, current techniques for radiographic evaluation of the temporomandibular joint continue to present the clinician with difficult problems in interpretation and diagnosis. The use of three-dimensional images reconstructed from computed tomographic (CT) data improves the diagnostic value of conventional CT at no additional risk to the patient and may provide new insights into this complex anatomic structure. A sequence of 1.5 mm CT sections is made with a slice-to-slice spacing of 1 .O mm. These are used to construct a new sequence of slices by mathematical interpolation in which the new slice spacing equals the size of the pixels. Structures to be imaged separately are then masked in the interpolated sections prior to "windowing" for the appropriate tissue density. A special algorithm detects the boundary surface of the selected structure. The surface pixels are assigned gray levels on the basis of their distance and attitude from the observer. When displayed, this produces a simulated three-dimensional image. The image can be rotated and sectioned. Rotations permit otherwise hidden surfaces to be examined. Images of a human temporomandibular joint in vitro are presented to demonstrate (1) the bony components of the joint and their relationships within the joint; (2) the bony components separated to display hidden surfaces; and (3) the joint meniscus in situ and as a separate component. (ORAL SURG. 58~461-474, 1984) I n current dental practice, radiographic interpretation of temporomandibular joint disorders is usually based on data obtained from one or more of several available procedures.'-3 These include various combinations of conventional radiography, arthrography, tomography, and computed tomography (CT). The criteria for selection of radiographic procedures are based on the available clinical findings, the type, quantity, and quality of radiographic information required, the equipment and expertise available, and the amount of radiation exposure considered justifiable (risk/benefit ratio).4s5 Madsen6 has outlined the goals of the ideal TMJ radiographic technique to include demonstration of all parts of the joint with minimal distortion, simultaneous depiction of both joints, patient comfort and natural posture, and accuracy sufficient to permit reproducible images

The accuracy of deformable image registration (DIR) techniques has an important role in adaptive radiation therapy. DIR algorithms are in fact used to adapt the contours outlined in planning CT in order to follow anatomical change occurring during radiotherapy treatment. In this study, we propose a quantitative validation of the deformable image registration algorithm implemented in MIM-Maestro using real phantoms and synthetic images. The MIM software uses a free-form intensitybased algorithm to perform CT-CT DIR; it also implements the RegRefine tool to manually correct deformed images. The DIR algorithm was tested considering deformations that typically occur in head and neck area: bending and tumour shrinking. Two phantoms mimicking these two situations were prepared and consecutive CT acquisitions were performed applying increasing degrees of deformation. DIR performances on images with different artificially changed contrasts and using RegRefine tool were also evaluated. DIR accuracy was estimated using DICE similarity index (DSI) between corresponding ROIs; ROI volume variation and distances (R) between corresponding markers. Results: Increase bending -DSI range between 0.96 and 0.94 for all the bending considered; the worst result corresponds to the maximum flection. Maximum R (1.8 mm) was measured in correspondence of rigid structures close to big deformations. Tumour shrinking -DSIs of rigid structures, close to the varying volume, range between 0.94 and 0.97 for maximum deformations. Decrease contrast -DSIs decrease with the contrast between objects down to 0.91 for the lower contrast. In this situation an R up to 2.2 mm was measured. RegRefine -DSIs improve when the tool is used; differences are greater at higher strains. Conclusions: The performance of MIM DIR algorithm remains reasonably good even in extreme bending, volume variations and poor contrast situations. The use of RegRefine improves registration results.

Illicit drug epidemics are infamous for their unexpected arrival and their speed of onset. Based on ethnographic work with youthful heroin experimenters in the Baltimore metropolitan area, an explanation was constructed based on circulating stories of drug reputation. An agent‐based model was built in SWARM to evaluate the explanation with good and bad stories about the drug passed among agents, which changed their initial attitudes. On repeated runs with different initial attitudes the model shows wide variation in outcomes and a dampening effect of increased social connections, contrary to epidemiologic expectations. The conclusion spells out implications for drug intervention and social research that relies on single case studies. © 2002 Wiley Periodicals, Inc.

Today the most cities in the world face different problems caused by vehicular traffic due to the increasing number of vehicles on road, such as traffic congestion, environmental pollution, excessive noise, increased accidents on the road, etc. Therefore it is important to represent the phenomenon by applying models so as to find better traffic flow patterns in a given city. In this work we introduce a simple macroscopic model to describe traffic flow in traffic congestion and we obtain the equation that relates the characteristic behavior between traffic flow and traffic density, known in the literature as fundamental relationship. Finally, we find an analytical solution for optimal flow passing vehicles on a congested road section.

The paper explores the possibilities of using information technologies based on the principles of machine learning and artificial intelligence in the process of digitizing Cyrillic manuscripts for the purposes of creating a historical dictionary of the Serbian language. Empirical research is based on the use of the Transkribus software platform in the creation of a model for automatic text recognition of the manuscripts by Gavril Stefanović Venclović, the most significant and prolific Serbian cultural enthusiast of the 18th century, whose extensive manuscript legacy in Serbian vernacular represents the most significant primary source for the historical dictionary of the Serbian language of this period. Following the results of conducted research, it can be concluded that the process of digitizing Cyrillic manuscripts for the purposes of creating a historical dictionary of the Serbian language can be significantly accelerated using Transkribus by creating specific and generic models for automatic text recognition. The advantage of automatic text recognition compared to the traditional methods is particularly reflected in the possibility of continuous improvement of the performance of specific and generic models in accordance with the progress of the transcription process and the increase in the amount of digitized text that can be used to train a new version of the model.

Introduction. Distributed Data Networks (DDNs) offer infrastructure solutions for sharing electronic health data from across disparate data sources to support comparative effectiveness research. Data sharing mechanisms must address technical and governance concerns stemming from network security and data disclosure laws and best practices, such as HIPAA.Methods. The Scalable Architecture for Federated Translational Inquiries Network (SAFTINet) deploys TRIAD grid technology, a common data model, detailed technical documentation, and custom software for data harmonization to facilitate data sharing in collaboration with stakeholders in the care of safety net populations. Data sharing partners host TRIAD grid nodes containing harmonized clinical data within their internal or hosted network environments. Authorized users can use a central web-based query system to request analytic data sets.Discussion. SAFTINet DDN infrastructure achieved a number of data sharing objectives, including s...

Reports on errors can be a rich source for understanding their causes, cascades, and consequences, leading to interventions for improvement. There are national and international calls for the development of appropriate error reporting and taxonomy systems that are useful at the point of care. The current momentum and urgency for these developments present an opportunity to harness the benefits of computer visualization that helps structure and illustrate the "story" of an error. This visualization process could help overcome the shortcomings of current reporting methods and could aid in creating an unambiguous international error taxonomy. We present a concept for a visual error reporting interface. The ambulatory care domain is used for illustration. The system has the potential to provide a user friendly, efficient means of reporting errors. Errors reported in this way would populate a "visual database," providing the ability to disseminate patient safety information in a straightforward, structured format that will be useful to a variety of stakeholders.

Background: Multiple taxonomies are used to classify medical errors. Most are conceptually based with limited empirical evidence on their utility to characterize processes leading to errors. We analyzed the utility of the Dimensions of Medical Outcomes taxonomy to describe medical ...

In order to achieve results and effi ciency, those in charge of the organization make decisions based on based on concrete data. With this, the controllership has the obligation to deliver up-to-date information, with future projections and easy to understand. This work is delivered by the contributions that the controller presents to the business management, because with its application, administrators will be able to obtain reliable information, accurate and reliable data that give greater support in decision making. The general objective of this article is to present the importance of the controller in decision making for improvement in the process of business management. A bibliographical research was elaborated, in order to mention the concepts and defi nitions. It is highlighted in the research, the various contributions that the controller can provide the organization, provided it is in a way properly and always giving greater support in their decision-making.

Процес післязбиральної обробки зерна, особливо, сушіння, є одним із найважливіших у технології виробництва сільськогосподарської продукції. Ця обробка сприяє одержанню високоякісної продукції, її тривалому та ефективному збереженню до збирання нового врожаю. Для отримання зернової сировини кондиційної вологості та зниження енерговитрат у сільськогосподарському виробництві використовують різні прийоми інтенсифікації сушіння: вібраційне сушіння, рециркуляційне сушіння, попереднє нагрівання, вакуумне сушіння, сушіння із застосуванням інфрачервоного випромінювання, сушіння струмами високої частоти, сушіння зерна під впливом електричного поля та багато інших. Перспективним з точки зору енерговитрат та якості отриманої сировини є використання вібраційних сушарок у поєднанні із технологією озонування, яка виступає додатковим фактором інтенсифікації процесу сушіння. Статтю присвячено вирішенню актуальної задачі інтенсифікації процесу сушіння зернової сировини та розробці нового віброозонуючого комплексу, що реалізує комплексний технологічний вплив на оброблюване середовище. Експериментально досліджено енергетичні параметри розробленого віброозонуючого комплексу в залежності від режимних параметрів досліджуваного процесу сушіння зернової сировини.

The importance of service delivery technology and online service adoption and usage in the banking industry has received an increased discussion in the literature in recent years. Owing to the fact that Strong online banking services are important drivers for bank performance and customer service delivery; several studies have been carried out on online banking service adoption or acceptance where services are already deployed and on the factors that influence customers' adoption and use or intention to use those services. However, despite the increasing discussion in the literatures, no attempt has been made to look at consumers' preference in terms of future online banking service adoption. This study used conjoint analysis and stated preference methods with discrete choice model to analyze the technology adoption pattern regarding consumers' preference for potential future online banking services in the Nigerian banking industry. The result revealed that to increase efficiency and strengthen competitiveness, banks need to promote smart and practical branded services especially self-services at the same time promote a universal adoption of e-banking system services that add entertainment or extra convenience to customers such as ease of usage including digital wallet, real-time interaction (video banking), ATMs integrated with smart phones, website customization, biometric services, and digital currency. These services can contribute to an increasing adoption of online services.

Null Convention Logic (NCL) is a robust asynchronous technique that poses new challenges to test and testability strategies due to the lack of a clock signal and the state-holding behavior of the NCL gates. The lack of deterministic timing in NCL complicates the management of test timing, and stuck-at faults on gate internal feedback (GIF) of the NCL gates exhibit a totally different effect compared to that of stuck-at faults on the gate inputs. Stuck-at faults on gate internal feedback of NCL gates do not always cause an incorrect output and therefore are considered hard-to-detect or undetectable by automatic test pattern generation (ATPG) algorithms. Such faults could leave the primary outputs of the circuit completely unaffected or sometimes they only affect the circuit by early detection of completeness. This work first proposes a clock-less self-timed ATPG, with no added design for test (DFT), that detects all of the faults on the gate inputs and a share of those on the GIF of gates. Then, this work investigates the effectiveness of I DDQ (quiescent current) test for detecting stuck-at faults on GIF of NCL gates. Hspice is used for implementing static and semi-static transistor-level NCL gates in (45nm, 1.1V) technology, for which the supply current is measured and compared for fault-free and faulty circuits. The experimental results show that the faulty current is orders of magnitude higher than the fault-free leakage current. This considerable difference shows that I DDQ testing might be an efficient and low-cost candidate for detecting stuck-at faults on GIF of NCL gates. The proposed I DDQ test method along with the self-timed ATPG has resulted in average 98.16% and 98.04% fault coverage for static and semi-static implementations of several NCL circuits, respectively. To the extent of our knowledge, this is the first work that has addressed clock-less, self-timed ATPG for NCL with no area overhead, and also the first work conducted on I DDQ test for NCL. • N. Nemati and M. C. Reed are with the

NULL Convention Logic™ (NCL™) is a symbolically complete logic which expresses process completely in terms of the logic itself and inherently and conveniently expresses asynchronous digital circuits. The traditional form of Boolean logic is not symbolically complete in the sense that it requires the participation of a fundamentally different form of expression, time in the form of the clock, which has to be very carefully coordinated with the logic part of the expression to completely and effectively express a process. We introduce NULL Convention Logic™ in relation to Boolean logic as a four value logic, and as a three value logic and finally as two value logic quite different from traditional Boolean logic. We then show how systems can be constructed entirely in terms of NULL Convention Logic™.

The advantage of a two-pass spatial transform tech- nique over a one-pass technique has been that it offers a more effective and efficient match with current implementation technology.'2 But only part of this advantage has been recognized. Previous thinking has focused on the mathematical notion of point, or coordinate, transforms mapping points of the output image to points in the input image. Resampling and interpolating input intensities about the mapped point to generate the output pixel intensity is a secondary concern. Distinct advantages can be gained simply by reversing this order of significance, that is, by considering first the intensity interpolation technique and The development of the technique described in this article was fully funded by Honeywell, Inc., and is being patented by the company.

Quasi-pseudo-MV algebras (quasi-pMV algebras, for short) arising from quantum computational logics are the generalizations of both quasi-MV algebras and pseudo-MV algebras. In this paper, we introduce the notions of states, state-morphisms, state operators and state-morphism operators to quasi-pMV algebras. First, we present the related properties of states on quasi-pMV algebras and show that states and Bosbach states coincide on any quasi-pMV algebra. And then we investigate the relationship between state-morphisms and the normal and maximal ideals of quasi-pMV algebras. We prove state-morphisms and extremal states are equivalent. The existence of states on quasi-pMV algebras is also discussed. Finally, state operators and state-morphism operators are introduced to quasi-pMV algebras, and the corresponding structures are called state quasi-pMV algebras and state-morphism quasi-pMV algebras, respectively. We investigate the related properties of ideals under state operators and state-morphism operators. Meanwhile, we show that there is a bijective correspondence between normal σ -ideals and ideal congruences on state quasi-pMV algebras. Ideals • Quasi-pseudo-MV algebras • States • State operators • State quasi-pMV algebras Communicated by A.

To begin, I must acknowledge my gratitude to José M. Martínez Sánchez who has been giving his generous support to this research for the last six years. He has aided me in many ways: by helping me obtain financial resources, inviting me to collaborate on very high-level research European projects and reviewing each and every one of my writings. In this regard, I would also like to thank Narciso García Santos for his expert advice, revisions and for believing in and supporting the research presented in this thesis. Thank you both for guiding me during these years. This thesis would not exist without your continuous help, advice, patience and professionalism. Second, I have to thank the Video Processing and Understanding (VPU) Lab, the research group in which my research interest has progressively grown, and its members: Jesús, Miguel Angel Javi, Luis I, Luis II, Fabrizio, Víctor Fernández, Víctor Valdés, Juan Carlos, Álvaro I, Álvaro II, Álvaro III, Marcos and Virginia. I thank Jesús Bescós Cano in particular for letting me attend his classes and sharing his knowledge, ideas, and resources. Third, I would like to thank Hermann Hellwagner, Christian Timmerer, and Dietmar Jannach for their support and help during my stay in Klagenfurt. Back in the summer of 2007 I initiated the innovative multimedia adaptation decision method that CAIN-21 uses. I thank them for revising all the versions of the papers in which these approaches have been put forth and explained. Fourth, I would like to thank Safak Dogan, Gokce Nur, Marta Mrak, Hemantha Kodikara Arachchi for their support during my second stay in their lab. They have expert knowledge of the quality-based decision methods, now incorporated into CAIN-21.

A partir de um mapeamento sistemático buscou-se verificar as contribuições de Learning Analytics e Mineração de Dados Educacionais no contexto educacional brasileiro. Optou-se por fontes de busca em três revistas na área de Informática na Educação e Anais de dois eventos de relevância nacional onde foram verificados 136 artigos entre janeiro de 2008 e março de 2020. Após a aplicação dos critérios de exclusão e de qualidade foram selecionados 71 artigos. Os resultados apresentam a maior ocorrência de projetos com finalidade de analisar desempenho acadêmico e prevenção de evasão escolar, embora nos últimos anos os assuntos estejam apresentando diversificação temática. Grande parte são voltados ao Ensino Superior e na modalidade de Educação a Distância. Há variedade de tecnologias e recursos utilizados no desenvolvimento de soluções em LA ressaltando o uso de Linguagem R, MySQL e a ferramenta Weka.

AimsClassification of vegetation is an essential tool to describe, understand, predict and manage biodiversity. Given the multiplicity of approaches to classify vegetation, it is important to develop international consensus around a set of general guidelines and purpose‐specific standard protocols. Before these goals can be achieved, however, it is necessary to identify and understand the different choices that are made during the process of classifying vegetation. This paper presents a framework to facilitate comparisons between broad‐scale plot‐based classification approaches.ResultsOur framework is based on the distinction of four structural elements (plot record, vegetation type, consistent classification section and classification system) and two procedural elements (classification protocol and classification approach). For each element we describe essential properties that can be used for comparisons. We also review alternative choices regarding critical decisions of classific...

Full bibliographic details must be given when referring to, or quoting from full items including the author's name, the title of the work, publication details where relevant (place, publisher, date), pagination, and for theses or dissertations the awarding institution, the degree type awarded, and the date of the award. If you believe that any material held in the repository infringes copyright law, please contact the

Full bibliographic details must be given when referring to, or quoting from full items including the author's name, the title of the work, publication details where relevant (place, publisher, date), pagination, and for theses or dissertations the awarding institution, the degree type awarded, and the date of the award. If you believe that any material held in the repository infringes copyright law, please contact the

In addressing the question of the design of technologies for e-discovery it is essential to recognise that such work takes place through a system in which both people and technology interact as a complex whole. Technology can promote discovery and insight and support human sensemaking in this context, but the question hangs on the extent to which it naturally extends the way legal practitioners think and work. We describe research at UCL which uses this as a starting point for empirical studies to inform the design of supporting technologies. We report aspects of an interview field study with lawyers who worked on a large regulatory investigation. Using data from the study we describe document review and analysis in terms of a sequence of transitions between different kinds of representation. We then focus on one particular transition: the creation of chronology records from documents. We develop the idea that investigators make sense of evidence by the application of conceptual 'frames' , but whilst the investigator 'sees' the situation in terms of these frames, the system 'sees' the situation in terms of documents, textual tokens and metadata. We conclude that design leverage can be obtained through the development of technologies that aggregate content around investigators' frames. We outline further research to explore this further.

We report an in-depth, longitudinal study of a freelance music journalist writing a feature article. Our analysis attends to the participant's activities from initiation to completion, and the ways in which she established structure using tools and artefacts to support cognitive effort. We observed five work stages: establishing an initial idea; preparing for an interview; interviewing; planning the article; and writing. Each resulted in the production of a working document embodying ideas and commitments which provided a key resource for the next stage. Stages began with phases of idea generation during which ideas were spontaneously triggered through intense engagement with information resources. They finished with phases of material consolidation when intermediate artifacts were configured to facilitate generation during the next stage. We examine these in detail and use our findings to motivate a discussion of working document overview representations and specific requirements related to idea generation and material consolidation.

Sensemaking occurs when people face the problem of forming an understanding of a situation. One scenario in which technology has a particularly significant impact on sensemaking and its success is in legal investigations. Legal investigations extend over time, are resource intensive, and require the sifting and re-representation of large collections of electronic evidence and close collaboration between multiple investigators. In this paper, we present an account of sensemaking in three corporate legal investigations. We summarise information interaction processes in the form of a model which conceptualises processes as resource transformations triggered and shaped by both bottom-up and top-down resources. The model both extends upon and validates aspects of a previous account of investigative sensemaking and brings to the fore two kinds of focusing. Data focusing involves identifying and structuring information to draw out facts relevant to a given set of investigation issues. Issue focusing involves revising the issues in the light of new insights. Both are essential in sensemaking. We draw this distinction through detailed accounts of two activities in the investigations: reviewing documents for relevance and the creation and use of external representations. This provides a basis for a number of requirements for sensemaking support systems, particularly in collaborative settings, including: document annotation, dynamically associating documents of a given type; interacting with documents in fluid ways; linking external representation elements to evidence; filtering external representations in flexible ways; and viewing external representations at different levels of scale and fidelity. Finally, we use our data to analyse the conceptual elements within a 'line of enquiry'. This provides a framework which can form the basis for partitioning information into hierarchically embedded enquiry 'contexts' within collaborative sensemaking systems.

Interest in sensemaking has recently gained prominence in human-computer interaction research. Efforts are being made to better understand sensemaking to help inform the design of appropriate computer-based tools. In this position paper we address the problem of design by presenting a propositional classification of sensemaking situations from a representational perspective. When people engage in sensemaking they organize information into structures which they use as a basis for guiding interpretation and the search for further data. It is these structures that we look at in terms of representations and classify them accordingly. This position paper argues that such a classification can be used as a starting point which would inform the design of sensemaking support tools.

Sensemaking has been described as a process involving information structuring. However, there are few detailed accounts of how this manifests in practice, particularly in relation to the creation and use of external representations such as data visualisations, and how such structuring aids sensemaking. To explore these questions in depth, we present an interview study of police crime analysts from which a model of their analysis process is developed. We describe the model focusing on the notion of 'think-steps', which for the analysts acted as a primary structuring concept. We describe how 'think-steps' propagate throughout the analysis process captured in the model. For the analysts, 'think-steps' are extensible templates that decompose a case into elements, provide a way of storing and visually structuring data, support generation of requests for information, focus research, simulate a case, and shape reporting. We reflect on the implications that our findi...

Sense-making plays an important role in Intelligence Analysis, but can be difficult to study in situ. Thus, it is common to use training exercises to study this phenomenon. In this paper, an exercise was undertaken by Military Intelligence personnel. The behaviour of groups of analysts is considered in terms the Data / Frame Model of Sense-making. The paper illustrates how Intelligence Analysis need not follow a linear process but often involves parallel and overlapping explorations of data, with multiple frames that might be minimal and sketchy. The use of representations, such as link diagrams, provides a means of externalizing frames and it is suggested that this shifts reasoning from deductive reasoning to abductive reasoning as the exercise progresses.

The reconstruction of analysts’ reasoning processes ( reasoning provenance) during complex sensemaking tasks can support reflection and decision making. One potential approach to such reconstruction is to automatically infer reasoning from low-level user interaction logs. We explore a novel method for doing this using machine learning. Two user studies were conducted in which participants performed similar intelligence analysis tasks. In one study, participants used a standard web browser and word processor; in the other, they used a system called INVISQUE (Interactive Visual Search and Query Environment). Interaction logs were manually coded for cognitive actions based on captured think-aloud protocol and posttask interviews based on Klein, Phillips, Rall, and Pelusos’s data/frame model of sensemaking as a conceptual framework. This analysis was then used to train an interaction frame mapper, which employed multiple machine learning models to learn relationships between the interac...

Full bibliographic details must be given when referring to, or quoting from full items including the author's name, the title of the work, publication details where relevant (place, publisher, date), pagination, and for theses or dissertations the awarding institution, the degree type awarded, and the date of the award. If you believe that any material held in the repository infringes copyright law, please contact the

The focus of sensemaking research is often on process and resources such as "schemas" and "frames". Less attention has been paid to the conceptual structures that make up the schema or frame, or how visualisations can be designed to support users' conceptual structures. In this chapter, we present an approach to gathering user requirements based on the conceptual structures that people are working with when making sense of a domain. We illustrate the approach with examples drawn from our own experience of designing, prototyping and testing an interactive visualisation tool for making sense of academic literature and of studies of sensemaking by lawyers and journalists. We discuss how to move from requirements to design, drawing on a classification of visualisations that highlights their principal conceptual structuring basis. Since each individual makes sense in their own way, it is beneficial to include features that enable people to work with a representation in their own way; for this, appropriation tools are helpful. We discuss the design of such features. Finally, we present an approach to evaluating interactive visualisations in terms of their support for sensemaking, focusing on the quality of the fit between users and system.

Information seeking does not occur in a vacuum but invariably is motivated by some wider task. It is well accepted that to understand information seeking we must understand the task context within which it takes place. Writing is amongst the most common tasks within which information seeking is embedded. This paper considers how writing can be understood in order to account for embedded information seeking. Following Sharples, the paper treats writing as a design activity and explore parallels between the psychology of design and information seeking. Significant parallels can be found and ideas from the psychology of design offer explanations for a number of information seeking phenomena. Next, a design‐oriented representation of writing tasks as a means of providing an account of phenomena such as information seeking uncertainty and focus refinement is developed. The paper illustrates the representation with scenarios describing the work of newspaper journalists.

There is need for user-centred visualisation research to engage with the activity context, information needs, knowledge and abilities of target user-groups. With a focus on the work of journalists, we first argue for information-retrieval results visualisation as a suitable browsing framework for journalists' frequently ill- defined needs and high-recall searches. We then describe the design and rationale for a histogram- based visualisation for journalists. We also describe the integration of this idea within a system that structures searching as a two-step query-and-filter operation. This approach is intended to support initial exploratory browsing and refinement in a way that is sympathetic to the systematic focusing that naturally occurs during complex, unstructured task performance. Further, the use of an enduring 'base' results set is intended to encourage structural familiarity with the broader results and therefore to enhance navigation.

This paper presents findings from a study of information seeking behaviour by National Health Service patients which explored motivational triggers for infor- mation needs. Previous research has highlighted the importance of contextual elements in users' changing information needs. This paper highlights how those needs may centre on specific events: in particular, a patient's consultation with their doctor. Patients initiate information seeking to assess whether they need clinical intervention, in preparation for the patient–doctor consultation and to verify the diagnosis or treatment stemming from that consultation. The study has revealed that having confidence in health practitioners is one key motiva- tion for information seeking. Another is a desire to use health service resources judiciously, efficiently and effectively.

We report an in-depth, longitudinal study of a freelance music journalist writing a feature article. Our analysis attends to the participant's activities from initiation to completion, and the ways in which she established structure using tools and artefacts to support cognitive effort. We observed five work stages: establishing an initial idea; preparing for an interview; interviewing; planning the article; and writing. Each resulted in the production of a working document embodying ideas and commitments which provided a key resource for the next stage. Stages began with phases of idea generation during which ideas were spontaneously triggered through intense engagement with information resources. They finished with phases of material consolidation when intermediate artifacts were configured to facilitate generation during the next stage. We examine these in detail and use our findings to motivate a discussion of working document overview representations and specific requirements related to idea generation and material consolidation.

As part of the 'Making Sense of Information'project, at UCL Interaction Centre in London [1], we are currently performing field-studies in order to better understand the sensemaking challenges faced by lawyers and journalists. We are using our findings to explore implications for systems to support their work better. As part of our project, we have undertaken a case-study of a large corporate fraud investigation. This involved indepth, retrospective interviews with 9 of the 30 lawyers who worked on the investigation, and ...

During extended sensemaking tasks people typically create external representations that integrate information and support their thinking. Understanding the variety, role and use of these is important for understanding sensemaking and how to support it effectively. We report a case-study of a large, document-based fraud investigation undertaken by a law firm. We focus on the construction and use of integrated representations in the form of chronologies. We show how these supported conjecture recording, focussing on time- ...