Using text-spotting to query the world

The world we live in is well labeled for the benefit of humans but to date robots have made little use of this resource. In this paper we describe a system that allows robots to read and interpret visible text and use it to understand the content of the scene. We use a generative probabilistic model that explains spotted text in terms of arbitrary search terms. This allows the robot to understand the underlying function of the scene it is looking at, such as whether it is a bank or a restaurant. We describe the text spotting engine at the heart of our system that is able to detect and parse wild text in images, and the generative model, and present results from images obtained with a robot in a busy city setting.

Why publish?

Anybody who has attempted to publish some aspect of their work in an academic journal will know that it isn’t as easy as it may seem. The amount of preparation required of a manuscript can be quite daunting. Besides actually writing the manuscript, the authors are faced with a number of technical requirements. Each journal has their own formatting requirements, relating not only to section headings and text layout, but also to very small details such as placement of commas in reference lists. Then, if presenting data in the form of figures, this must be formatted so that it can be understood by the readership, and most journals still require that the data be in a format which can be read when printed in black-and-white. Most daunting (and important) of all, for the article to be scientifically valid it must be absolutely true in the representation of the work reported (i.e. all data must be shown unless a strong justification exists for removing data points), and this might cause angst in the mind of the authors when the results aren’t clear or possibly contradict the expected or desired result.

Making sense of an everyday science text : linguistic, visual and spatial design

In this article, we take a close look at the literacy demands of one task from the ‘Marvellous Micro-organisms Stage 3 Life and Living’ Primary Connections unit (Australian Academy of Science, 2005). One lesson from the unit, ‘Exploring Bread’, (pp 4-8) asks students to ‘use bread labels to locate ingredient information and synthesise understanding of bread ingredients’. We draw upon a framework offered by the New London Group (2000), that of linguistic, visual and spatial design, to consider in more detail three bread wrappers and from there the complex literacies that students need to interrelate to undertake the required task. Our findings are that although bread wrappers are an example of an everyday science text, their linguistic, visual and spatial designs and their interrelationship are not trivial. We conclude by reinforcing the need for teachers of science to also consider how the complex design elements of everyday science texts and their interrelated literacies are made visible through instructional practice.

Abstracting and correlating heterogeneous events to detect complex scenarios

The research presented in this thesis addresses inherent problems in signaturebased intrusion detection systems (IDSs) operating in heterogeneous environments. The research proposes a solution to address the difficulties associated with multistep attack scenario specification and detection for such environments. The research has focused on two distinct problems: the representation of events derived from heterogeneous sources and multi-step attack specification and detection. The first part of the research investigates the application of an event abstraction model to event logs collected from a heterogeneous environment. The event abstraction model comprises a hierarchy of events derived from different log sources such as system audit data, application logs, captured network traffic, and intrusion detection system alerts. Unlike existing event abstraction models where low-level information may be discarded during the abstraction process, the event abstraction model presented in this work preserves all low-level information as well as providing high-level information in the form of abstract events. The event abstraction model presented in this work was designed independently of any particular IDS and thus may be used by any IDS, intrusion forensic tools, or monitoring tools. The second part of the research investigates the use of unification for multi-step attack scenario specification and detection. Multi-step attack scenarios are hard to specify and detect as they often involve the correlation of events from multiple sources which may be affected by time uncertainty. The unification algorithm provides a simple and straightforward scenario matching mechanism by using variable instantiation where variables represent events as defined in the event abstraction model. The third part of the research looks into the solution to address time uncertainty. Clock synchronisation is crucial for detecting multi-step attack scenarios which involve logs from multiple hosts. Issues involving time uncertainty have been largely neglected by intrusion detection research. The system presented in this research introduces two techniques for addressing time uncertainty issues: clock skew compensation and clock drift modelling using linear regression. An off-line IDS prototype for detecting multi-step attacks has been implemented. The prototype comprises two modules: implementation of the abstract event system architecture (AESA) and of the scenario detection module. The scenario detection module implements our signature language developed based on the Python programming language syntax and the unification-based scenario detection engine. The prototype has been evaluated using a publicly available dataset of real attack traffic and event logs and a synthetic dataset. The distinct features of the public dataset are the fact that it contains multi-step attacks which involve multiple hosts with clock skew and clock drift. These features allow us to demonstrate the application and the advantages of the contributions of this research. All instances of multi-step attacks in the dataset have been correctly identified even though there exists a significant clock skew and drift in the dataset. Future work identified by this research would be to develop a refined unification algorithm suitable for processing streams of events to enable an on-line detection. In terms of time uncertainty, identified future work would be to develop mechanisms which allows automatic clock skew and clock drift identification and correction. The immediate application of the research presented in this thesis is the framework of an off-line IDS which processes events from heterogeneous sources using abstraction and which can detect multi-step attack scenarios which may involve time uncertainty.

Historical timelines : analyzing multimodal text design

The recent focus on literacy in Social Studies has been on linguistic design, particularly that related to the grammar of written and spoken text. When students are expected to produce complex hybridized genres such as timelines, a focus on the teaching and learning of linguistic design is necessary but not sufficient to complete the task. Theorizations of new literacies identify five interrelated meaning making designs for text deconstruction and reproduction: linguistic, spatial, visual, gestural, and audio design. Honing in on the complexity of timelines, this paper casts a lens on the linguistic, visual, spatial, and gestural designs of three pairs of primary school aged Social Studies learners. Drawing on a functional metalanguage, we analyze the linguistic, visual, spatial, and gestural designs of their work. We also offer suggestions of their effect, and from there consider the importance of explicit instruction in text design choices for this Social Studies task. We conclude the analysis by suggesting the foci of explicit instruction for future lessons.

Semantic oscillations : encoding context and structure in complex valued holographic vectors

In computational linguistics, information retrieval and applied cognition, words and concepts are often represented as vectors in high dimensional spaces computed from a corpus of text. These high dimensional spaces are often referred to as Semantic Spaces. We describe a novel and efficient approach to computing these semantic spaces via the use of complex valued vector representations. We report on the practical implementation of the proposed method and some associated experiments. We also briefly discuss how the proposed system relates to previous theoretical work in Information Retrieval and Quantum Mechanics and how the notions of probability, logic and geometry are integrated within a single Hilbert space representation. In this sense the proposed system has more general application and gives rise to a variety of opportunities for future research.

Ageing and use of complex product interfaces

This paper presents an experiment designed to investigate if redundancy in an interface has any impact on the use of complex interfaces by older people and people with low prior-experience with technology. The important findings of this study were that older people (65+ years) completed the tasks on the Words only based interface faster than on Redundant (text and symbols) interface. The rest of the participants completed tasks significantly faster on the Redundant interface. From a cognitive processing perspective, sustained attention (one of the functions of Central Executive) has emerged as one of the important factors in completing tasks on complex interfaces faster and with fewer of errors.

Process based synthesis to evaluate design flexibility in airport terminal layout

Flexible design concept is a relatively new trend in airport terminal design which is believed to facilitate the ever changing needs of a terminal. Current architectural design processes become more complex every day because of the introduction of new building technologies where the concept of flexible airport terminal would apparently make the design process even more complex. Previous studies have demonstrated that ever growing aviation industry requires airport terminals to be planned, designed and constructed in such a way that should allow flexibility in design process. In order to adopt the philosophy of ‘design for flexibility’ architects need to address a wide range of differing needs. An appropriate integration of the process models, prior to the airport terminal design process, is expected to uncover the relationships that exist between spatial layout and their corresponding functions. The current paper seeks to develop a way of sharing space adjacency related information obtained from the Business Process Models (BPM) to assist in defining flexible airport terminal layouts. Critical design parameters are briefly investigated at this stage of research whilst reviewing the available design alternatives and an evaluation framework is proposed in the current paper. Information obtained from various design layouts should assist in identifying and defining flexible design matrices allowing architects to interpret and to apply those throughout the lifecycle of the terminal building.

Approaches to designing for older adults' intuitive interaction with complex devices

Many older people have difficulties using modern consumer products due to increased product complexity both in terms of functionality and interface design. Previous research has shown that older people have more difficulty in using complex devices intuitively when compared to the younger. Furthermore, increased life expectancy and a falling birth rate have been catalysts for changes in world demographics over the past two decades. This trend also suggests a proportional increase of older people in the work-force. This realisation has led to research on the effective use of technology by older populations in an effort to engage them more productively and to assist them in leading independent lives. Ironically, not enough attention has been paid to the development of interaction design strategies that would actually enable older users to better exploit new technologies. Previous research suggests that if products are designed to reflect people's prior knowledge, they will appear intuitive to use. Since intuitive interfaces utilise domain-specific prior knowledge of users, they require minimal learning for effective interaction. However, older people are very diverse in their capabilities and domain-specific prior knowledge. In addition, ageing also slows down the process of acquiring new knowledge. Keeping these suggestions and limitations in view, the aim of this study was set to investigate possible approaches to developing interfaces that facilitate their intuitive use by older people. In this quest to develop intuitive interfaces for older people, two experiments were conducted that systematically investigated redundancy (the use of both text and icons) in interface design, complexity of interface structure (nested versus flat), and personal user factors such as cognitive abilities, perceived self-efficacy and technology anxiety. All of these factors could interfere with intuitive use. The results from the first experiment suggest that, contrary to what was hypothesised, older people (65+ years) completed the tasks on the text only based interface design faster than on the redundant interface design. The outcome of the second experiment showed that, as expected, older people took more time on a nested interface. However, they did not make significantly more errors compared with younger age groups. Contrary to what was expected, older age groups also did better under anxious conditions. The findings of this study also suggest that older age groups are more heterogeneous in their capabilities and their intuitive use of contemporary technological devices is mediated more by domain-specific technology prior knowledge and by their cognitive abilities, than chronological age. This makes it extremely difficult to develop product interfaces that are entirely intuitive to use. However, by keeping in view the cognitive limitations of older people when interfaces are developed, and using simple text-based interfaces with flat interface structure, would help them intuitively learn and use complex technological products successfully during early encounter with a product. These findings indicate that it might be more pragmatic if interfaces are designed for intuitive learning rather than for intuitive use. Based on this research and the existing literature, a model for adaptable interface design as a strategy for developing intuitively learnable product interfaces was proposed. An adaptable interface can initially use a simple text only interface to help older users to learn and successfully use the new system. Over time, this can be progressively changed to a symbols-based nested interface for more efficient and intuitive use.

Mapping semantic knowledge for unsupervised text categorisation

Text categorisation is challenging, due to the complex structure with heterogeneous, changing topics in documents. The performance of text categorisation relies on the quality of samples, effectiveness of document features, and the topic coverage of categories, depending on the employing strategies; supervised or unsupervised; single labelled or multi-labelled. Attempting to deal with these reliability issues in text categorisation, we propose an unsupervised multi-labelled text categorisation approach that maps the local knowledge in documents to global knowledge in a world ontology to optimise categorisation result. The conceptual framework of the approach consists of three modules; pattern mining for feature extraction; feature-subject mapping for categorisation; concept generalisation for optimised categorisation. The approach has been promisingly evaluated by compared with typical text categorisation methods, based on the ground truth encoded by human experts.

Monitoring of complex systems of interacting dynamic systems

Increases in functionality, power and intelligence of modern engineered systems led to complex systems with a large number of interconnected dynamic subsystems. In such machines, faults in one subsystem can cascade and affect the behavior of numerous other subsystems. This complicates the traditional fault monitoring procedures because of the need to train models of the faults that the monitoring system needs to detect and recognize. Unavoidable design defects, quality variations and different usage patterns make it infeasible to foresee all possible faults, resulting in limited diagnostic coverage that can only deal with previously anticipated and modeled failures. This leads to missed detections and costly blind swapping of acceptable components because of one’s inability to accurately isolate the source of previously unseen anomalies. To circumvent these difficulties, a new paradigm for diagnostic systems is proposed and discussed in this paper. Its feasibility is demonstrated through application examples in automotive engine diagnostics.

Text messaging to support off-campus clinical facilitators of undergraduate student nurses

Managing large cohorts of undergraduate student nurses during off-campus clinical placement is complex and challenging. Clinical facilitators are required to support and assess nursing students during clinical placement. Therefore clear communication between university academic coordinators and clinical facilitators is essential for consistency and prompt management of emerging issues. Increasing work demands require both coordinators and facilitators to have an efficient and effective mode of communication. The aim of this study was to explore the use of Short Message Service (SMS) texts, sent between mobile phones, for communication between university Unit Coordinators and off-campus Clinical Facilitators. This study used an after-only design. During a two week clinical placement 46 clinical facilitators working with first and second year Bachelor of Nursing students from a large metropolitan Australian university were regularly sent SMS texts of relevant updates and reminders from the university coordinator. A 15 item questionnaire comprising x of 5 point likert scale and 3 open-ended questions was then used to survey the clinical facilitators. The response rate was 47.8% (n=22). Correlations were found between the approachability of the coordinator and facilitator perception of a) that the coordinator understood issues on clinical placement (r=0.785, p<0.001,), and b) being part of the teaching team (r=0.768, p<0.001). Analysis of responses to qualitative questions revealed three themes: connection, approachability and collaboration. Results indicate that SMS communication is convenient and appropriate in this setting. This quasi-experimental after-test study found regular SMS communication improves a sense of connection, approachability and collaboration.

Text mining for screening efficiency? Testing within a Cochrane public health review

Background The requirement for dual screening of titles and abstracts to select papers to examine in full text can create a huge workload, not least when the topic is complex and a broad search strategy is required, resulting in a large number of results. An automated system to reduce this burden, while still assuring high accuracy, has the potential to provide huge efficiency savings within the review process. Objectives To undertake a direct comparison of manual screening with a semi‐automated process (priority screening) using a machine classifier. The research is being carried out as part of the current update of a population‐level public health review. Methods Authors have hand selected studies for the review update, in duplicate, using the standard Cochrane Handbook methodology. A retrospective analysis, simulating a quasi‐‘active learning’ process (whereby a classifier is repeatedly trained based on ‘manually’ labelled data) will be completed, using different starting parameters. Tests will be carried out to see how far different training sets, and the size of the training set, affect the classification performance; i.e. what percentage of papers would need to be manually screened to locate 100% of those papers included as a result of the traditional manual method. Results From a search retrieval set of 9555 papers, authors excluded 9494 papers at title/abstract and 52 at full text, leaving 9 papers for inclusion in the review update. The ability of the machine classifier to reduce the percentage of papers that need to be manually screened to identify all the included studies, under different training conditions, will be reported. Conclusions The findings of this study will be presented along with an estimate of any efficiency gains for the author team if the screening process can be semi‐automated using text mining methodology, along with a discussion of the implications for text mining in screening papers within complex health reviews.

A new constraint handling method for wind farm layout optimization with lands owned by different owners

For wind farm optimizations with lands belonging to different owners, the traditional penalty method is highly dependent on the type of wind farm land division. The application of the traditional method can be cumbersome if the divisions are complex. To overcome this disadvantage, a new method is proposed in this paper for the first time. Unlike the penalty method which requires the addition of penalizing term when evaluating the fitness function, it is achieved through repairing the infeasible solutions before fitness evaluation. To assess the effectiveness of the proposed method on the optimization of wind farm, the optimizing results of different methods are compared for three different types of wind farm division. Different wind scenarios are also incorporated during optimization which includes (i) constant wind speed and wind direction; (ii) various wind speed and wind direction, and; (iii) the more realisticWeibull distribution. Results show that the performance of the new method varies for different land plots in the tested cases. Nevertheless, it is found that optimum or at least close to optimum results can be obtained with sequential land plot study using the new method for all cases. It is concluded that satisfactory results can be achieved using the proposed method. In addition, it has the advantage of flexibility in managing the wind farm design, which not only frees users to define the penalty parameter but without limitations on the wind farm division.

Dynamic Bayesian network inferencing for non-homogeneous complex systems

Dynamic Bayesian Networks (DBNs) provide a versatile platform for predicting and analysing the behaviour of complex systems. As such, they are well suited to the prediction of complex ecosystem population trajectories under anthropogenic disturbances such as the dredging of marine seagrass ecosystems. However, DBNs assume a homogeneous Markov chain whereas a key characteristics of complex ecosystems is the presence of feedback loops, path dependencies and regime changes whereby the behaviour of the system can vary based on past states. This paper develops a method based on the small world structure of complex systems networks to modularise a non-homogeneous DBN and enable the computation of posterior marginal probabilities given evidence in forwards inference. It also provides an approach for an approximate solution for backwards inference as convergence is not guaranteed for a path dependent system. When applied to the seagrass dredging problem, the incorporation of path dependency can implement conditional absorption and allows release from the zero state in line with environmental and ecological observations. As dredging has a marked global impact on seagrass and other marine ecosystems of high environmental and economic value, using such a complex systems model to develop practical ways to meet the needs of conservation and industry through enhancing resistance and/or recovery is of paramount importance.