Search

This paper consists of a study conducted to quantify the traffic congestion problem. The quantification is made in terms of congestion cost and index. To study the effect of congestion, a road stretch of Thiruvananthapuram city in Kerala state was selected. A traffic congestion index was assessed based on travel time model. Congestion index helps to rank the road segments which will be useful for prioritizing the traffic management measures effectively. Another tool used for quantifying the congestion was the congestion cost. From the study, it was inferred that the congestion cost for the study stretch was about 0.03% of GSDP of Kerala and about 25% of the expenditure allocated for the road transport sector during the Kerala state Budget 2020–21. The economic analysis of traffic delay cost will inform the policymakers in monetary terms the actual impact of the traffic congestion and delays that the society bears and helps to allocate the resources efficiently.

One great challenge of modern networks is how to select paths and react to highly variable and demanding traffic patterns. In recent years, emerging networking languages and architectures, such as P4/PISA, have created unprecedented opportunities for rapidly prototyping disruptive solutions in programmable data planes. In this direction, source routing (SR) is a prominent alternative to conventional table-based routing, since it reduces the convergence time of building distributed routing tables. In particular, PolKA explores the polynomial residue number system (RNS) for a fully stateless SR, i.e., no state update on packet headers needs to be conveyed along the route. This paper reports the deployment of PolKA in a continental testbed composed of P4 programmable switches. Results for end-to-end throughput and forwarding latency show that PolKA matches the data plane performance of traditional approaches, while paving the way for further innovative applications by exploring RNS properties in dynamic scenarios.

An algorithm is developed based on Edmund K. Miller’s Model-Based Parameter Estimation (MBPE) technique to mitigate the effects of missing or corrupted data in random regions of wideband linear frequency modulated (LFM) radar signals. Two methods of applying MBPE in the spectral/frequency domain are presented that operate on either the full complex data or separated magnitude/phase data, respectively. The final algorithm iteratively applies MBPE using the latter approach to re-generate results in the corrupted regions of a windowed LFM signal until the difference is minimized relative to un-corrupted data. Several sets of simulations were conducted across many randomized gap parameters where impulse response (IPR) impacts are summarized. Conditions where the algorithm successfully improved the IPR for a single target are provided. The algorithm’s effectiveness on multiple targets, especially when the corrupted regions are relatively large compared to the overall bandwidth of the signal, are also explored.

This article presents a generic structured approach supporting the detection of exploitable software vulnerabilities of given type. Its applicability is illustrated for two weakness types: buffer overflowing and race conditions.

Technological advances continue to produce massive amounts of information from a variety of sources about our everyday lives. The simple use of a smartphone, for example, can generate data on individuals through telephone records (including location data), social media activity, Internet browsing, e-commerce transactions, and email communications. Much attention has been given to expectations of privacy in light of this data collection, especially consumer privacy. Much attention has also been given to how and when government agencies collect and use this data to monitor the activities of individuals.In previous work, we discussed three aspects of data analytics in the context of intelligence agencies that matter, but are often overlooked: stages, numbers, and the human factor. We made the following observations. First, stages of the algorithmic process should be included in what legal constraints apply to these activities. Secondly, the law should take into account the difference between people and machines. Thirdly, the law should think more about numbers. Although each of these topics deserves further discussion, we focus in this chapter in particular on the first issue: stages.

An examination of parallels between the "Day of Infamy" and a major cyberattack reveals lessons for organizations about vulnerability, the nature of survival, and the tenets of protection that can help boost resilience against hackers and other cyberthreats.

The 32 chapters of the 2019 Cambridge Handbook of Computing Education Research synthesize the existing research in computing education and propose new directions for future research. An author from each chapter will summarize their chapter with auto-advancing slides. Attendees will be introduced to the breadth of content in the new handbook and can identify chapters of interest. This fits uniquely as a special session, and will likely be informative, inspiring, and overwhelming.

This paper revisits the challenge of Berman and Hafner's "missing link" paper on representing teleological structure in case-based legal reasoning. It is noted that this was mainly an ontological challenge to represent some of what made legal reasoning distinctive, which was given less attention than factual similarity in the dominant AI and Law paradigm, deriving from HYPO. The response to their paper is noted and briefly evaluated. A parallel is drawn to a new challenge to provide deep structure to the legal context of textual meaning, drawing on the forthcoming work of two Constitutional law scholars who appear to place some faith in the ways of thinking that AI and Law has developed.

The process of creating a photo product, such as a photobook, calendar or collage, from a large personal image collection requires intensive user effort. The primary goal of the current research was to develop an end-to-end solution to the problem of photo product generation that enables the user to complete the process with minimal edits, where the system intelligently selects assets and groups them before presenting the output to the user. The automation is driven by metadata extracted both from individual images as well as from sets of assets in a collection. In particular, we use an automatically detected event hierarchy to establish meaningful groupings in the assets, and to determine an appropriate grouping and pagination for the final product. We propose a novel intermediate construct, called a storyboard, which can be translated to different product types without recomputing the underlying metadata. In addition to chronological storyboards, we also describe a novel hybrid storyboard that joins chronological image presentation with groups of images of a common theme. A pagination algorithm uses the information in the storyboard and the product constraints to generate a product. Finally, the user is provided with a metadata-driven editing mechanism that makes it easy to change the auto-populated product. Given that the proposed system envisions user interaction in creating the final product, user studies are conducted to judge the usefulness of the system, where consumers use the system to generate a photobook with their own images. An end-to-end, metadata-driven intelligent system for automatic photobook creationA flexible intermediate story representation for creating multiple product typesAutomatic generation of representations for event-based and theme-based groupingsA new pagination algorithm for mapping selected images onto photobook pagesMetadata-aided UI enabling users to view groupings and find alternative images

This article compares security fuzzing approaches with respect to different characteristics commenting on their pro and cons concerning both their potential for exposing vulnerabilities and the expected effort required to do so. These preliminary considerations based on abstract reasoning and engineering judgement are subsequently confronted with experimental evaluations based on the application of three different fuzzing tools characterized by diverse data generation strategies on examples known to contain exploitable buffer overflows. Finally, an example inspired by a real-world application illustrates the importance of combining different fuzzing concepts in order to generate data in case fuzzing requires the generation of a plausible sequence of meaningful messages to be sent over a network to a software-based controller as well as the exploitation of a hidden vulnerability by its execution.

Humans can process only small amounts of data, computers can process almost infinitely. But also computers need to act smart or intelligent, because otherwise even they get swamped and/or might produce useless information. Algorithms help in structuring and analyzing vast amounts of data. With the growth of data we have to rely increasingly on algorithms. These algorithms may perform better than alternative approaches we used to rely on. However, algorithms can be opaque, and the danger is that we get obscured by algorithms. Intelligence agencies use algorithms to distinguish between persons of interest and others. Law enforcement uses analytics and data mining to identify suspects and to support investigations. Businesses profile users in all kind of categories. Surveillance is omnipresent. The impact on privacy is not necessarily depending on who does the surveillance. It depends not only on the actors, but on various factors. Sometimes what businesses do impacts severely on the privacy of consumers, sometimes the work of police or intelligence agencies does not. In this paper we focus on surveillance by the US National Security Agency (NSA) and other intelligence agencies. Our aim is to dissect data analytics by intelligence agencies, and to suggest what privacy related law should focus on more than it does today. With an understanding of how big data algorithms usually work we discuss in this chapter the use of algorithms from a privacy and data protection angle. First, we briefly introduce the central concepts of data protection and privacy against the background of the General Data Protection Regulation introduced by the European Union in 2012, published in 2016 and effective as of 25 May 2018. The core of the chapter consists of elaborating upon three issues: 1. The stages of data processing while using of algorithms, how it affects privacy and what safeguards the law should provide; 2. The role of the human factor: how and when should humans be involved in evaluating outcomes, and also under what circumstances human interference is better abstained from; 3. The relevance of scale and scope: in the light of privacy, numbers matter. However, so far in law a discussion on the relevance of numbers (or scale) is largely absent.

The ability to rank the images based on their appearance finds many real-world applications, such as image retrieval or image album creation. Despite the recent dominance of deep learning methods in computer vision which often result in superior performance, they are not always the methods of choice because they lack interpretability. In this paper, we investigate the possibility of improving the image aesthetic inference of the convolutional neural networks with hand-designed features that rely on domain expertise in various fields. We perform a comparison of hand-crafted feature sets in their ability to predict fine-grained aesthetics scores on two image aesthetics data sets. We observe that even feature sets published earlier are able to compete with more recently published algorithms and, by combining the algorithms, a significant improvement in predicting image aesthetics is possible. By using a tree-based learner, we perform the feature elimination to understand the best performing features overall and across different image categories. Only roughly 15% and 8% of the features are needed to achieve full performance in predicting a fine-grained aesthetic score and binary classification, respectively. By combining the hand-crafted features with metafeatures that predict the quality of an image based on convolutional neural network features, the model performs better than a baseline VGG16 model. One can, however, achieve more significant improvement in both aesthetics score prediction and binary classification by fusing the hand-crafted features and the penultimate layer activations. Our experiments indicate an improvement up to 2.2% achieving current state-of-the-art binary classification accuracy on the aesthetic visual analysis data set when the hand-designed features are fused with activations from VGG16 and ResNet50 networks.

This article proposes two heuristic approaches targeted at the optimized generation of test cases capable of triggering buffer overflows resp. underflows. Both testing techniques are based on guiding conditions statically derived by Integer Constraint Analysis. First experimental evaluations confirmed the superiority of local optimization algorithms over global ones.

Gradient control plays an important role in feed-forward networks applied to various computer vision tasks. Previous work has shown that Recurrent Highway Networks minimize the problem of vanishing or exploding gradients. They achieve this by setting the eigenvalues of the temporal Jacobian to 1 across the time steps. In this work, batch normalized recurrent highway networks are proposed to control the gradient flow in an improved way for network convergence. Specifically, the introduced model can be formed by batch normalizing the inputs at each recurrence loop. The proposed model is tested on an image captioning task using MSCOCO dataset. Experimental results indicate that the batch normalized recurrent highway networks converge faster and performs better compared with the traditional LSTM and RHN based models., Comment: 5 pages, 3 figures, Published in 2017 IEEE International Conference on Image Processing (ICIP)

This paper introduces a sentence to vector encoding framework suitable for advanced natural language processing. Our latent representation is shown to encode sentences with common semantic information with similar vector representations. The vector representation is extracted from an encoder-decoder model which is trained on sentence paraphrase pairs. We demonstrate the application of the sentence representations for two different tasks -- sentence paraphrasing and paragraph summarization, making it attractive for commonly used recurrent frameworks that process text. Experimental results help gain insight how vector representations are suitable for advanced language embedding., Comment: 5 pages, 4 figures, IEEE GlobalSIP 2017 Conference

Humans can process only small amounts of data, computers can process almost infinitely. But also computers need to act smart or intelligent, because otherwise even they get swamped and/or might produce useless information. Algorithms help in structuring and analyzing vast amounts of data. With the growth of data we have to rely increasingly on algorithms. These algorithms may perform better than alternative approaches we used to rely on. However, algorithms can be opaque, and the danger is that we get obscured by algorithms. Intelligence agencies use algorithms to distinguish between persons of interest and others. Law enforcement uses analytics and data mining to identify suspects and to support investigations. Businesses profile users in all kind of categories. Surveillance is omnipresent. The impact on privacy is not necessarily depending on who does the surveillance. It depends not only on the actors, but on various factors. Sometimes what businesses do impacts severely on the privacy of consumers, sometimes the work of police or intelligence agencies does not. In this paper we focus on surveillance by the US National Security Agency (NSA) and other intelligence agencies. Our aim is to dissect data analytics by intelligence agencies, and to suggest what privacy related law should focus on more than it does today. With an understanding of how big data algorithms usually work we discuss in this chapter the use of algorithms from a privacy and data protection angle. First, we briefly introduce the central concepts of data protection and privacy against the background of the General Data Protection Regulation introduced by the European Union in 2012, published in 2016 and effective as of 25 May 2018. The core of the chapter consists of elaborating upon three issues: 1. The stages of data processing while using of algorithms, how it affects privacy and what safeguards the law should provide; 2. The role of the human factor: how and when should humans be involved in evaluating outcomes, and also under what circumstances human interference is better abstained from; 3. The relevance of scale and scope: in the light of privacy, numbers matter. However, so far in law a discussion on the relevance of numbers (or scale) is largely absent.

This paper describes how information warfare (IW) is now being carried on the back of cyber warfare (CW). IW is thus amplified so attacks may be deeper, broader, faster, more specific, or more directly causal than in the past. The paper argues that instead of hacking an electrical grid or transportation system, disrupting operations, the new IW-on-CW strategy is a hacking of the knowledge infrastructure (KI). Causing an election-day logistics problem or spreading fake news puts the national knowledge infrastructure at risk. Cyber attack on cyber-physical (CP) information infrastructure (II) is traditionally biased toward the command and control of physical infrastructure. IW traditionally considers scales of time and reach appropriate to pre-internet propagation and points of failure. Critical infrastructure is considered to be power, transportation, food, water, shelter, security, and emergency response, but also (CP) communications, (KI) banking, and now, elections, news, and social media (all KI). The next targets of national knowledge industries might be institutional or industry-wide, including engineering, education, medicine, surveillance, monitoring, investment, advertising, entertainment, and law, with new, heretofore unseen time scales. Knowledge hacking has evolved because pathways are controllable, not just perimeters breachable. IW-on-CW is made possible by the largely voluntary surrender of epistemological checks and balances to the conveniences of cyberspace. Defenses are within the control of a vigilant population that resists trading vulnerability for convenience.

We revisit the issue of maintaining a reasonable amount of privacy for browser users when logs are at risk of being sold and profiles are at risk of being viewed under embarrassing circumstances. A URL that generates dynamic calls in javascript is designed so users have plausible deniability when items show up in profiles, recommendations, or suggested text completions. Three design features are notable: (1) the use of open seed text and an unrestricted user-customizable list of URLs provided externally by the user, and not logged, so broad, indisputable claims of origination are possible; (2) the use of an AI/NLP part-of-speech informed phrase extractor (chunker) to generate queries that cluster semantically; (3) the use of a generative grammar of search-term sequences, or dialogues, to produce more realistic searches. An algorithmic technique is employed to assist in determining what requests were made by the tool post hoc, so forensics are still possible, but only when a specific user's logs have been produced for analysis. We also discuss how an "arms race" between analytics AI and obfuscating AI can be won by combining multiple tools, and how such a race can be avoided simply by deliberately reducing precision, unilaterally, by those who do user profiling analytics.

Concept inventories hold tremendous promise for promoting the rigorous evaluation of teaching methods that might remedy common student misconceptions and promote deep learning. The measurements from concept inventories can be trusted only if the concept inventories are evaluated both by expert feedback and statistical scrutiny (psychometric evaluation). Classical Test Theory and Item Response Theory provide two psychometric frameworks for evaluating the quality of assessment tools. We discuss how these theories can be applied to assessment tools generally and then apply them to the Digital Logic Concept Inventory (DLCI). We demonstrate that the DLCI is sufficiently reliable for research purposes when used in its entirety and as a post-course assessment of students’ conceptual understanding of digital logic. The DLCI can also discriminate between students across a wide range of ability levels, providing the most information about weaker students’ ability levels.

This article proposes approaches supporting the analysis of code vulnerabilities based on overlapping machine instructions of variable length. For the purpose of focusing the search for potential malicious code it is suggested to apply first disassembling techniques allowing for a restriction of potentially exploitable memory space. Successively, testing based on heuristic optimization may be applied in order to evaluate dynamically the practicality of vulnerability exploitation.

In response to increasing threats of malicious activity and data loss on servers, we propose a different and practical strategy for access control modeled after flight plans for pilots, which mixes existing role-based, object-based, and intention-based access models; it supports much finer grained, real-time, sequence-oriented anomaly detection. Users are required to declare their intended “flight path” in advance, a sketch of resource use: this may vary in detail, but could include database tables, file system directories, byte and bandwidth limits, use of encryption and archive creation, command sets, connection time, number and origin of connections, and ports. Sequence information provides especially strong constraint, even if it incomplete. We find an important place for active, on-line human sampling of flight plans, as well as pre-authorization for non-standard paths, and alerts for deviation from path. We also find a place for improved user profiling and a paradigm shift from ex-post log-based reconstruction of user activity to ex-ante declaration.

Ancillary learning opportunities outside the classroom take different forms such as peer-led team learning (PLTL) and supplemental instruction (SI). To gain the benefits of ancillary sessions with less overhead than PLTL and SI, we facilitated optional weekly supervised homework sessions for AAE 35200, an aerospace structural mechanics course for third year engineering students at Purdue University. In these sessions, a graduate teaching assistant used the assigned homework problems to demonstrate key concepts on structural mechanics. In order to understand the attributes of the participants, a survey was administered. Also, session attendance was recorded and compared against the exam and homework results. Based on the collected data during the Fall 2015 and Spring 2016 semesters, we have reached the following conclusions: a) attending the supervised homework sessions does not significantly affect the exam score, whereas the opposite is true for the homework scores and b) SAT-Math scores do not seem to have a strong influence on both the exam and homework scores.

This poster presents an innovative user-support method for structured data entry that resulted from working with a large patient registry reporting group in a leading hospital, tasked with efficient electronic medical record discretization. It is a method of using syntactic constraints on various fields during web-forms coding, to auto-select the fields where values belong, this allows coders to spend more time looking at the clinical narrative documentation and less time attending to the mouse, the browser, and field-selection on the data entry pages. The method is to meta-program a "floating text box" for intermediate, semi-structured notes, with behavior specific to the choices on that page. It bridges the gap between abbreviations, which are easy, and datamining natural language text, which is hard. The ideas can be generalized to auto-filling forms and semi-autonomous forms filling beyond the medical domain.

We describe a novel method of securing services by adding windowed whitelisting to an arbitrary and constantly changing assignment of services to ports (or virtual ports). This is aimed at mitigating port scanning threats and unauthorized intrusion attempts, and to protect a community of known users from data loss. In essence, port numbers, time, and IP address will be used as part of the password/access mechanism; this segregates traffic so that content-based restrictions can be more effective. It also provides a connection-based security wrapper for services that might be vulnerable to software exploits, such as the buffer overruns and backdoors. The method requires a portal to authenticate users and disseminate knowledge of the current port assignment, in addition to permitting users to request a “window” of time to be white-listed. It requires a firewall with dynamic port and whitelist reconfigurability. The method is intended to enhance byte frequency histogram analysis and regexp restriction of traffic. It also requires a policy for keeping alive long-lasting connections. It can be implemented easily with virtual ports using redirection. We discuss some implications for web page rewriting and cgi security, as well as legacy services such as ssh and sftp. The effect is to create a cross-product of IP range, port range, and time specificity, to create a large and sparse search space for any adversary.

The advance of the Internet in the past decade has radically changed the way people communicate and collaborate with each other. Physical distance is no more a barrier in online social networks, but cultural differences (at the individual, community, as well as societal levels) still govern human-human interactions and must be considered and leveraged in the online world. The rapid deployment of high-speed Internet allows humans to interact using a rich set of multimedia data such as texts, pictures, and videos. This position paper proposes to define a new research area called 'connected multimedia', which is the study of a collection of research issues of the super-area social media that receive little attention in the literature. By connected multimedia, we mean the study of the social and technical interactions among users, multimedia data, and devices across cultures and explicitly exploiting the cultural differences. We justify why it is necessary to bring attention to this new research area and what benefits of this new research area may bring to the broader scientific research community and the humanity.

We investigate general concept classification in unconstrained videos by joint audio-visual analysis. An audio-visual grouplet (AVG) representation is proposed based on analyzing the statistical temporal audio-visual interactions. Each AVG contains a set of audio and visual codewords that are grouped together according to their strong temporal correlations in videos, and the AVG carries unique audio-visual cues to represent the video content. By using the entire AVGs as building elements, video concepts can be more robustly classified than using traditional vocabularies with discrete audio or visual codewords. Specifically, we conduct coarse-level foreground/background separation in both audio and visual channels, and discover four types of AVGs by exploring mixed-and-matched temporal audio-visual correlations among the following factors: visual foreground, visual background, audio foreground, and audio background. All of these types of AVGs provide discriminative audio-visual patterns for classifying various semantic concepts. To effectively use the AVGs for improved concept classification, a distance metric learning algorithm is further developed. Based on the AVG structure, the algorithm uses an iterative quadratic programming formulation to learn the optimal distances between data points according to the large-margin nearest-neighbor setting. Various types of grouplet-based distances can be computed using individual AVGs, and through our distance metric learning algorithm these grouplet-based distances can be aggregated for final classification. We extensively evaluate our method over the large-scale Columbia consumer video set. Experiments demonstrate that the AVG-based audio-visual representation can achieve consistent and significant performance improvements compared wth other state-of-the-art approaches.

We provide a retrospective of 25'years of the International Conference on AI and Law, which was first held in 1987. Fifty papers have been selected from the thirteen conferences and each of them is described in a short subsection individually written by one of the 24 authors. These subsections attempt to place the paper discussed in the context of the development of AI and Law, while often offering some personal reactions and reflections. As a whole, the subsections build into a history of the last quarter century of the field, and provide some insights into where it has come from, where it is now, and where it might go.

The authors conducted a qualitative interview-based study to reveal students' misconceptions about state in sequential circuits. This paper documents 16 misconceptions of state, how students' conceptions of state shift and change, and students' methodological weaknesses. These misconceptions can be used to inform and direct instruction. This study revealed a need for the development and adoption of standard terminology and a need to focus digital logic instruction upon a central concept of information encoding. In addition, these misconceptions will serve as the basis for the creation of standard assessments called concept inventories. A concept inventory will provide rigorous and quantitative metrics to assess the effectiveness of new teaching methods.

To improve instruction in computer engineering and computer science, instructors must better understand how their students learn. Unfortunately, little is known about how students learn the fundamental concepts in computing. To investigate student conceptions and misconceptions about digital logic concepts, the authors conducted a qualitative interview-based study. In the interviews, students verbalized their thinking while they solved digital logic problems. The interviews were analyzed to identify students' misconceptions. This paper presents five classes of students' misconceptions and methodological weaknesses concerning medium-scale integrated (MSI) circuits. These misconceptions were used to create a multiple-choice conceptual assessment, called a concept inventory, that is being used to test for the prevalence of the misconceptions discovered through the interviews. Although the misconceptions documented in the paper are primarily intended for the construction of the concept inventory, they can also point to potential ways to improve instruction.

We investigate the challenging issue of joint audio-visual analysis of generic videos targeting at concept detection. We extract a novel local representation, Audio-Visual Atom (AVA), which is defined as a region track associated with regional visual features and audio onset features. We develop a hierarchical algorithm to extract visual atoms from generic videos, and locate energy onsets from the corresponding soundtrack by time-frequency analysis. Audio atoms are extracted around energy onsets. Visual and audio atoms form AVAs, based on which discriminative audio-visual codebooks are constructed for concept detection. Experiments over Kodak's consumer benchmark videos confirm the effectiveness of our approach.

A concept inventory is a standardized assessment tool intended to evaluate a student’s understanding of the core concepts of a topic. In order to create a concept inventory it is necessary to accurately identify these core concepts. A Delphi process is a structured multi-step process that uses a group of experts to achieve a consensus opinion. We present the results of three Delphi processes to identify topics that are important and difficult in each of three introductory computing subjects: discrete mathematics, programming fundamentals, and logic design. The topic rankings can not only be used to guide the coverage of concept inventories, but can also be used by instructors to identify what topics merit special attention.

The academic programming language community continues to reject the change in programming practices brought about by scripting. We need a programming language pragmatics to go past the analysis of syntax and semantics in the same way that linguistics studies perlocution and illocution. Pragmatic questions are not the easiest for mathematically inclined computer scientists to address. They refer by nature to people and their habits, sociology, and the day's technological demands. An industrial psychology literature, apart from computing, has sometimes addressed questions of this kind. But this kind of study must become part of programming language theory within computing. It's the importance of just these kinds of questions that makes programmers choose scripting languages. The author recommends that scripting, not Java, be taught first, asserting that students should learn to love their own possibilities before they learn to loathe other people's restrictions.

The paper I presented at the conference gave graphical form to certain kinds of legal arguments: arguments from precedent cases (Loui, 2007). I did not intend to impose a kind of discipline on those who make such arguments. My research aims at elucidating the various forms of reasoning, proposing rules that capture their patterns and regularities, even if those rules and patterns might seem naive to experts at arguing from precedent. To a formalist, a few good rules are better than no rules, especially since artificial intelligence (AI) depends on having rules for computers to follow. A graphical way of diagramming such arguments is simply an alternative way to present the disciplined arguments and to make them comprehensible to humans, especially when the symbols are being manipulated by automata. For some of my logic colleagues, however, the implicit value of such a graphical mechanism is normative: one presumes to improve the quality of legal arguments by imposing a canonical form. This is the presumption that ran badly afoul at this conference. Why should someone who is already very good at making arguments be interested in following our rules for diagramming their arguments? My idea was simply to improve on the ubiquitous ‘Toulmin form’ for diagramming defeasible arguments (Hitchcock and Verheij, 2006). Analogies are a specific kind of argument requiring a specific kind of Toulmin structure. My theory is itself a putative improvement on existing theories of analogy from precedent. Hence, by claiming to have an improved theory of analogical argument, and claiming to refine Toulmin diagrams with that theory, it is easy to imagine that people could be led to better arguments, better rationality and better decision making, by using new diagramming conventions. This is what many formalists and many research computer scientists engaged in logic and visualization would have claimed. But why should it be true? Why are more diagramming rules better than fewer? I did not explicitly claim that people who use our diagrams make better arguments. I would not even have claimed that people who use Toulmin diagrams make better arguments with those diagrams. My paper was actually a ‘Modest Proposal for Annotating the Dialectical State’ and literary tradition required that such a paper be ironic or satirical, even humorous in the face of serious circumstances (Swift, 1996). So I suggested, whimsically, that the boxes and arrows and recursive structure of argument graphs could be replaced by a single historical image: the face of a political candidate, or a civil war general, perhaps even a football quarterback. Throw out all the rigid logical form, I wrote tongue-in-cheek, and use Barack Obama’s face for a new and untested argument of great popular force, or George W. Bush’s face for an old argument that has gotten progressively more difficult to make and which began on dubious grounds anyway. It was not such a bad idea: replace the Microsoft Windows style summaries that use plus and minus signs (or red and green lights) with a richer set of ‘emoticons.’ Some arguments get a plus sign, or a green light, when many

Video segmentation is the task of grouping pixels in successive video frames into perceptually coherent regions. It is a preliminary step to solve higher level problems such as automated surveillance, object tracking, video summarization, video indexing and retrieval. For consumer videos, video segmentation is a useful tool for extracting relevant and interesting content from such video sequence for further analysis or re-purposing of the visual content. Given an unannotated video sequence captured by either a static or hand-held camera, our graph-based approach first effectively models the data in a high dimensional feature space, which emphasizes the correlation between similar pixels while reducing the inter-class connectivity between different objects. The graph model fuses appearance, spatial, and temporal information to break a volumetric video sequence into semantic spatiotemporal key-segments. By further grouping the key-segments, a binary segmentation is able to extract a moving object of interest from a video sequence based on its unique and distinguishable regional properties. Experiment results show the robustness of our approach, which has achieved comparable or better performance when compared to several unsupervised methods.

Social media, such as Facebook and twitter, received much attention recently especially due to their real-time nature. For example, when an earthquake occurs, people immediately post information related to the earthquake, which enables detection of earthquake occurrence promptly, simply by observing these posts. As described in this paper, we investigate the real-time interaction of events such as earthquakes in Twitter, Facebook and other social media, and propose an algorithm to monitor tweets and to detect a target event. We devised a filter of data based on features such as the keywords, the number of times they are present, and their context. We consider each user feed as a sensor and the collection of such sensors creates a system, which can be used to promptly warn registered users. The posts triggering the detections also provided very short first-impression narratives from people who experienced the shaking. We will also show that the validity of such a process is not bound to a particular context or language but can be used on a variety of other subjects

In core engineering courses, instructors administer multiple examinations as major assessments of students' learning. When a student is unable to take an exam, the instructor must estimate the missing exam score in order to calculate the student's course grade. Using exam score data from multiple offerings of two large engineering courses, we compared the accuracy of several methods to estimate an exam score, including linear regression methods. The standard error of regression of the ordinary least squares (OLS) regression model was consistently about 0.5. For final exam scores, the standard errors of linear models with equal weights were nearly the same as the standard errors of the OLS regression models. For other exam scores, the equal weight model was somewhat less accurate. The results of this study provide practical guidance to instructors who need to estimate missing exam scores.

Structured data and controlled vocabularies in medical records are common goals, but widespread and eager clinical adoption might require more modest, incremental change. One small step along the path to structured data is a voluntary discipline of annotation, supported by simple, user-friendly, dynamic text affordances. A second step is to discipline the metavocabulary rather than the clinical vocabulary: one can target the epistemological attitudes (probability, degree of uncertainty, pedigree of claims) and ontological criteria (fuzziness or vagueness, version of definition, reporting standard) for optional mark up, and even plain text can support disciplined annotation. The methods seem especially useful for summary views of patient records that have multiple authors, contributing at different times, using various criteria. The methods may also support more nuanced outcomes research and reporting, and aid transitions between policies, regulations, and billing codes. It may be too difficult to impose ontologies on so many different participants in medical record-keeping, all at once, with little room for customization. A more organic, evolutionary path to the same end starts with the smaller step of asking medical records keepers to settle on the common annotations. Even as a crowd-sourced meta-language takes shape, the freedom to use epistemic and ontological qualifiers makes controlled terminologies and vocabularies, with discrete and structured data, easier to live with.

We consider the strong possibility that extraverted and intraverted women differ on preference for some aspects of programming, and how this difference might be more significant than its expression among men. We discuss the reasons why we are moved to design a new study confirming a gender-specific relation of Myers-Briggs personality types to programming paradigms and specific language features. We discuss the implications of a significant gender difference on representation in industry, advancement in profession, and curriculum design. We invite hypotheses that explain a significant difference. Past classroom questionnaires and future questions will be shown (but not data, since we are bound by IRB rules not to report them), as well as method of analysis and test for robustness in subcategories. We have a specific method for mitigating ambiguity and non-repeatability in Myers-Briggs type determination. Our cohort is expected to include about 500 Master's students per year, most from the same region in India, with about 25% women respondents, if we sample at past rates.