Summary: Low vision seriously impedes people from performing daily tasks especially reading. Readful-U is a mobile application with an attachable stand that helps people with low vision to read easily. It mainly targets the elderly patients since they are the primary group affected. Furthermore, users will be engaged in wider social interactions through inviting people to read for them. Built on current reading assistant technologies, Readful-U steps into the blank space to make audio assistance a vivid interaction between people rather than with a machine generated voice. The user-centered design process is featured with parallel designs, primary user research, contextual inquiry, prototyping, user testing, and iterations. Going beyond the common functions of current reading assistant devices, Readful-U specially caters to the emotional and social needs of low vision patients in an innovative and cost-effective way.

Summary: Human-machine trust is a critical mitigating factor in many HCI instances. Lack of trust in a system can lead to system disuse whilst over-trust can lead to inappropriate use. Whilst human-machine trust has been examined extensively from within a technico-social framework, few efforts have been made to link the dynamics of trust within a steady-state operator-machine environment to the existing literature of the psychology of learning. We set out to recreate a commonly reported learning phenomenon within a trust acquisition environment: Users learning which algorithms can and cannot be trusted to reduce traffic in a city. We failed to replicate (after repeated efforts) the learning phenomena of "blocking", resulting in a finding that people consistently make a very specific error in trust assignment to cues in conditions of uncertainty. This error can be seen as a cognitive bias and has important implications for HCI.

Summary: The increasing online availability of scholarly corpora promises unprecedented opportunities for visualizing and studying scholarly communities. We seek to leverage this with a mixed-method approach that integrates network analysis of features of the online corpora with ethnographic studies of the communities that produce them. In our development of tools and visualizations we seek to support the going back and forth between views of community structures and the perceptions and research trajectories of individual researchers and research groups. We here present results from tracking the temporal evolution of community structures within a research specialty. We explore how the temporal evolution of these maps can be used to provide insights into the historical evolution of a field as well as extract more accurate snapshots of the community structures at a given point in time. We are currently conducting qualitative interviews with experts in this research specialty to assess the validity of the maps.

Summary: Accurate and noise robust multimodal activity and mental state monitoring can be achieved by combining physiological, behavioural and environmental signals. This is especially promising in assistive driving technologies, because vehicles now ship with sensors ranging from wheel and pedal activity, to voice and eye tracking. In practice, however, multimodal user studies are confronted with challenging data collection and synchronisation issues, due to the diversity of sensing, acquisition and storage systems. Referencing current research on cognitive load measurement in a driving simulator, this paper describes the steps we take to consistently collect and synchronise signals, using the Orbit Measurement Library (OML) framework, combined with a multimodal version of a cinema clapperboard. The resulting data is automatically stored in a networked database, in a structured format, including metadata about the data and experiment. Moreover, fine-grained synchronisation between all signals is provided without additional hardware, and clock drift can be corrected post-hoc.

Summary: In the past 30 years, tremendous progress has been achieved in building effective shallow classification models. Despite the success, we come to realize that, for many applications, the key bottleneck is not the qualify of classifiers but that of features. Not being able to automatically get useful features has become the main limitation for shallow models. Since 2006, learning high-level features using deep architectures from raw data has become a huge wave of new learning paradigms. In recent two years, deep learning has made many performance breakthroughs, for example, in the areas of image understanding and speech recognition. In this talk, I will walk through some of the latest technology advances of deep learning within Baidu, and discuss the main challenges, e.g., developing effective models for various applications, and scaling up the model training using many GPUs. In the end of the talk I will discuss what might be interesting future directions.

Summary: This paper describes the Haptic Voice Recognition (HVR) Grand Challenge 2012 and its datasets. The HVR Grand Challenge 2012 is a research oriented competition designed to bring together researchers across multiple disciplines to work on novel multimodal text entry methods involving speech and touch inputs. Annotated datasets were collected and released for this grand challenge as well as future research purposes. A simple recipe for building an HVR system using the Hidden Markov Model Toolkit (HTK) was also provided. In this paper, detailed analyses of the datasets will be given. Experimental results obtained using these data will also be presented.

Summary: Haptic voice recognition (HVR) is a multi-modal text entry method for smart mobile devices. It employs haptic events generated by speakers during speaking to achieve better efficiency and robustness for automatic speech recognition. This paper describes the detailed design of the 2012 SJTU submission for the HVR Grand Challenge. During the design, a new perplexity metric using conditional entropy is proposed to evaluate the potential search space reduction of a haptic event without speech input. A number of new haptic events are evaluated both theoretically and experimentally in detail. The final submission system uses the haptic event of initial letter plus final letter and reduces word error rate by 76% compared to the baseline initial letter event.

Summary: This paper examines several writing features for the evaluation of cognitive load. Our analysis is focused on writing features within and between written strokes, including writing pressure, writing velocity, stroke length and inter-stroke movements. Based on a study of 20 subjects performing a sentence composition task, the reported findings reveal that writing pressure and writing velocity information are very good indicators of cognitive load. A stroke selection threshold was investigated for constraining the feature extraction to long strokes, which resulted in a small further improvement. Differing from most previous research investigating cognitive load during writing based on task performance criteria, this work proposes a new approach to cognitive load measurement using writing dynamics, with the potential to allow new or improve existing handwriting interfaces.

Keywords: rule-base stroke deformation, special nine grid, specific style rendering

Summary: Manifesting the handwriting characters with the specific style of a famous artwork is fascinating. In this paper, a system is built to render the user's handwriting characters with a specific style. A stroke database is established firstly. When rendering a character, the strokes are extracted and recognized, then proper radicals and strokes are filtered, finally these strokes are deformed and the result is generated. The Special Nine Grid (SNG) is presented to help recognize radicals and strokes. The Rule-base Stroke Deformation Algorithm (RSDA) is proposed to deform the original strokes according to the handwriting strokes. The rendering result manifests the specific style with high quality. It is feasible for people to generate the tablet or other artworks with the proposed system.

Keywords: coupled graphical items, labels, linking, menu hierarchy

Summary: This paper proposes Coupa, a novel pen interaction design to support operations of users on portable devices. The design arranges a plurality of labels on the interface, each of which has an identity. The user forms a coupling by linking two graphical items together, and thus performs an action dependent on the identities of the coupled items. During the course of operation, any item on the screen is ready for linking and coupling. To reduce mal-operations, two principles for linking are proposed, with their effectiveness proved in the usability tests. Compared with traditional systems with hierarchical menu structure and point-and-click interaction, the proposed design prominently improves the efficiency and accuracy of pen-based systems with enhanced usability.

Keywords: collaborative filtering, matrix factorization, nonparametric models

Summary: With the sheer growth of online user data, it becomes challenging to develop preference learning algorithms that are sufficiently flexible in modeling but also affordable in computation. In this paper we develop nonparametric matrix factorization methods by allowing the latent factors of two low-rank matrix factorization methods, the singular value decomposition (SVD) and probabilistic principal component analysis (pPCA), to be data-driven, with the dimensionality increasing with data size. We show that the formulations of the two nonparametric models are very similar, and their optimizations share similar procedures. Compared to traditional parametric low-rank methods, nonparametric models are appealing for their flexibility in modeling complex data dependencies. However, this modeling advantage comes at a computational price -- it is highly challenging to scale them to large-scale problems, hampering their application to applications such as collaborative filtering. In this paper we introduce novel optimization algorithms, which are simple to implement, which allow learning both nonparametric matrix factorization models to be highly efficient on large-scale problems. Our experiments on EachMovie and Netflix, the two largest public benchmarks to date, demonstrate that the nonparametric models make more accurate predictions of user ratings, and are computationally comparable or sometimes even faster in training, in comparison with previous state-of-the-art parametric matrix factorization models.

Keywords: structure determination, style evaluation model (SEM), style-consistency calligraphy synthesis

Summary: There are lots of digitized calligraphy works written by ancient famous calligraphists in CADAL (China-America Digital Academic Library) digital library. To make use of these resources, users want to generate a tablet or a piece of calligraphic works written by some ancient famous calligraphist. But some characters in the tablet or the calligraphic work hadn't been written by the calligraphist or though were ever written but are hard to read because of long time weathering. In this paper, a novel approach is proposed to synthesize Chinese calligraphic characters which are in the same style of some calligraphist, and a corresponding system is developed for calligraphy works generation and tablets design.
    Calligraphic character is represented by a three-level hierarchical model. A novel approach for determining the character structure is proposed, which takes advantage of both the structure of the same characters of different styles and the structure of similar characters of the same style. A style evaluation model (SEM) is presented to evaluate whether the calligraphic character generated is in the same style of the specified calligraphist and to adjust the calligraphic character generated. Our experiments show that this system is effective.

Summary: In this paper we propose a non-greedy active learning method for text categorization using least-squares support vector machines (LSSVM). Our work is based on transductive experimental design (TED), an active learning formulation that effectively explores the information of unlabeled data. Despite its appealing properties, the optimization problem is however NP-hard and thus -- like most of other active learning methods -- a greedy sequential strategy to select one data example after another was suggested to find a suboptimum. In this paper we formulate the problem into a continuous optimization problem and prove its convexity, meaning that a set of data examples can be selected with a guarantee of global optimum. We also develop an iterative algorithm to efficiently solve the optimization problem, which turns out to be very easy-to-implement. Our text categorization experiments on two text corpora empirically demonstrated that the new active learning algorithm outperforms the sequential greedy algorithm, and is promising for active text categorization applications.

Keywords: collaborative filtering, recommender systems, semi-supervised learning, social network analysis, spectral clustering

Summary: The Web offers rich relational data with different semantics. In this paper, we address the problem of document recommendation in a digital library, where the documents in question are networked by citations and are associated with other entities by various relations. Due to the sparsity of a single graph and noise in graph construction, we propose a new method for combining multiple graphs to measure document similarities, where different factorization strategies are used based on the nature of different graphs. In particular, the new method seeks a single low-dimensional embedding of documents that captures their relative similarities in a latent space. Based on the obtained embedding, a new recommendation framework is developed using semi-supervised learning on graphs. In addition, we address the scalability issue and propose an incremental algorithm. The new incremental method significantly improves the efficiency by calculating the embedding for new incoming documents only. The new batch and incremental methods are evaluated on two real world datasets prepared from CiteSeer. Experiments demonstrate significant quality improvement for our batch method and significant efficiency improvement with tolerable quality loss for our incremental method.

Summary: The world wide web contains rich textual contents that are interconnected via complex hyperlinks. This huge database violates the assumption held by most of conventional statistical methods that each web page is considered as an independent and identical sample. It is thus difficult to apply traditional mining or learning methods for solving web mining problems, e.g., web page classification, by exploiting both the content and the link structure. The research in this direction has recently received considerable attention but are still in an early stage. Though a few methods exploit both the link structure or the content information, some of them combine the only authority information with the content information, and the others first decompose the link structure into hub and authority features, then apply them as additional document features. Being practically attractive for its great simplicity, this paper aims to design an algorithm that exploits both the content and linkage information, by carrying out a joint factorization on both the linkage adjacency matrix and the document-term matrix, and derives a new representation for web pages in a low-dimensional factor space, without explicitly separating them as content, hub or authority factors. Further analysis can be performed based on the compact representation of web pages. In the experiments, the proposed method is compared with state-of-the-art methods and demonstrates an excellent accuracy in hypertext classification on the WebKB and Cora benchmarks.

Summary: Latent semantic indexing (LSI) is a well-known unsupervised approach for dimensionality reduction in information retrieval. However if the output information (i.e. category labels) is available, it is often beneficial to derive the indexing not only based on the inputs but also on the target values in the training data set. This is of particular importance in applications with multiple labels, in which each document can belong to several categories simultaneously. In this paper we introduce the multi-label informed latent semantic indexing (MLSI) algorithm which preserves the information of inputs and meanwhile captures the correlations between the multiple outputs. The recovered "latent semantics" thus incorporate the human-annotated category information and can be used to greatly improve the prediction accuracy. Empirical study based on two data sets, Reuters-21578 and RCV1, demonstrates very encouraging results.

Summary: Information filtering has made considerable progress in recent years. The predominant approaches are content-based methods and collaborative methods. Researchers have largely concentrated on either of the two approaches since a principled unifying framework is still lacking. This paper suggests that both approaches can be combined under a hierarchical Bayesian framework. Individual content-based user profiles are generated and collaboration between various user models is achieved via a common learned prior distribution. However, it turns out that a parametric distribution (e.g. Gaussian) is too restrictive to describe such a common learned prior distribution. We thus introduce a nonparametric common prior, which is a sample generated from a Dirichlet process which assumes the role of a hyper prior. We describe effective means to learn this nonparametric distribution, and apply it to learn users' information needs. The resultant algorithm is simple and understandable, and offers a principled solution to combine content-based filtering and collaborative filtering. Within our framework, we are now able to interpret various existing techniques from a unifying point of view. Finally we demonstrate the empirical success of the proposed information filtering methods.

Summary: Relevance feedback (RF) has been an effective query modification approach to improving the performance of information retrieval (IR) by interactively asking a user whether a set of documents are relevant or not to a given query concept. The conventional RF algorithms either converge slowly or cost a user's additional efforts in reading irrelevant documents. This paper surveys several RF algorithms and introduces a novel hybrid RF approach using a support vector machine (HRFSVM), which actively selects the uncertain documents as well as the most relevant ones on which to ask users for feedback. It can efficiently rank documents in a natural way for user browsing. We conduct experiments on Reuters-21578 dataset and track the precision as a function of feedback iterations. Experimental results have shown that HRFSVM significantly outperforms two other RF algorithms.

Summary: In order to reduce human efforts, there has been increasing interest in applying active learning for training text classifiers. This paper describes a straightforward active learning heuristic, representative sampling, which explores the clustering structure of 'uncertain' documents and identifies the representative samples to query the user opinions, for the purpose of speeding up the convergence of Support Vector Machine (SVM) classifiers. Compared with other active learning algorithms, the proposed representative sampling explicitly addresses the problem of selecting more than one unlabeled documents. In an empirical study we compared representative sampling both with random sampling and with SVM active learning. The results demonstrated that representative sampling offers excellent learning performance with fewer labeled documents and thus can reduce human efforts in text classification tasks.

Summary: Guaranteed, Reliable, Secure Broadcast (GRSB) - is a protocol that provides reliable and secure broadcast/multicast communications. Four logical nodes are enforced in the network - a Central Retransmitter, a Security Controller, a Designated Acknowledger, a (many when needed) Playback Recorder(s). Through the coordinated service of the four nodes, every user node can be guaranteed to receive all broadcast messages in a secure manner and in the correct temporal order. This paper focuses on the implication of GRSB to office information systems. How GRSB coherently supports several progressive functional requirements, from small number of user nodes to complex, but integrated functions, is elaborated.

Keywords: Models and principles, User/machine systems, Database management, Logical design, Data models, Schema and subschema, Database management, Languages, Data description languages (DDL), Database management, Systems, Distributed systems, Information storage and retrieval, Content analysis and indexing, Information storage and retrieval, Systems and software, Information systems applications, Office automation, Information systems applications, Communications applications, Artificial intelligence, Applications and expert systems, Office automation, Artificial intelligence, Knowledge representation formalisms and methods, Frames and scripts, Representations, Text processing, Document preparation, Format and notation, Design, Economics, Human factors, Management, Computer-supported cooperative work, Hypertext, Information Lens, Intelligent agents, Object-oriented databases, Semiformal systems

Summary: Object Lens allows unsophisticated computer users to create their own cooperative work applications using a set of simple, but powerful, building blocks. By defining and modifying templates for various semistructured objects, users can represent information about people, tasks, products, messages, and many other kinds of information in a form that can be processed intelligently by both people and their computers. By collecting these objects in customizable folders, users can create their own displays which summarize selected information from the objects in table or tree formats. Finally, by creating semiautonomous agents, users can specify rules for automatically processing this information in different ways at different times.
    The combination of these primitives provides a single consistent interface that integrates facilities for object-oriented databases, hypertext, electronic messaging, and rule-based intelligent agents. To illustrate the power of this combined approach, we describe several simple examples of applications (such as task tracking, intelligent message routing, and database retrieval that we have developed in this framework.

reprinted in Baecker, R. M. (1993) Readings in Groupware and Computer Supported Cooperative Work: Assisting Human-Human Collaboration, Morgan Kaufmann Publishers