http://www.computer.org/intelligent IEEE Intelligent Systems, a bimonthly publication of the IEEE Computer Society, covers new tools, techniques, concepts, and current research and development activities in intelligent systems. The magazine serves software engineers, systems designers, information managers, knowledge engineers, and professionals in finance, manufacturing, medicine, law, and geophysical sciences. en-us Sun, 19 May 2013 10:00:07 GMT http://csdl.computer.org/common/images/logos/intelligent.gif List of recently published journal articles http://www.computer.org/intelligent http://doi.ieeecomputersociety.org/10.1109/MIS.2013.38 Advanced satellite tracking technologies have collected huge amounts of wild birds’ migration data. These data are very useful for biologists to understand birds’ dynamic migration patterns, to study correlations between the habitats, and to predict global spread trends of avian influenza. We transform the biological problem into a machine learning problem by converting the migratory paths of wild birds into graphs. Our first step of H5N1 outbreak prediction is to discover weighted closed cliques from the graphs by our mining algorithm HELEN (short for High-wEight cLosed cliquE miNing), which are then used by our learning algorithm HELEN-p to predict potential H5N1 outbreaks at habitats. We show that the prediction is more accurate in comparison with that by the traditional method on a migration data set obtained through a real satellite bird-tracking system. It is also confirmed by our empirical analysis that H5N1 spreads in a manner of high-weight closed cliques and frequent cliques. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.38 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.28 In recent years several efforts were devoted to automatically mining opinions and sentiments from natural language in social media messages, news and commercial product reviews. Since this task involves a deep understanding of the explicit and implicit information conveyed by the language, most of the approaches refer to annotated corpora. However, the development of this kind of resource raises several new challenges due both to the specificity of the data from such domains and text genres, and to the knowledge to be annotated. This paper focusses on the main issues related to the development of a corpus for opinion and sentiment analysis, with a special attention to irony, and presents as a case study Senti-TUT, an ongoing project for Italian aimed at investigating sentiment and irony about politics in social media. We introduce and analyze the Senti-TUT corpus, a collection of texts from Twitter annotated morpho-syntactically and with sentiment polarity. We describe the dataset, the annotation, the methodologies applied and our investigations on two important features of irony: polarity reversing and emotion expressions. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.28 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.26 Optimal design and control of smart space structures are computational intensive. In traditional design methods, numbers of sensors and actuators required, their positions, and parameters of controllers are selected and optimized sequentially. Hence, only local optimality can be achieved by these methods. In order to reach the global optimal performance, a new approach is proposed in this article that will implement the concurrent design for smart space structures. In this approach, the quantity and placement of sensors/actuators and parameters of controllers are simultaneously optimized by a memetic evolutionary algorithm. A solar array smart structure has been used for computational experiments and the corresponding results indicate that the proposed concurrent design can obtain better performance than the sequential one. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.26 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.9 The number of videos available online and elsewhere is continuously growing, and with this the need for effective methods to process the vast amount of multimodal information shared through this media. This paper addresses the task of multimodal sentiment analysis, and presents a method that integrates linguistic, audio, and visual features for the purpose of identifying sentiment in online videos. We focus our experiments on a new dataset consisting of Spanish videos collected from the social media website YouTube and annotated for sentiment polarity. Through comparative experiments, we show that the joint use of visual, audio, and textual features greatly improves over the use of only one modality at a time. Moreover, we also test the portability of our multimodal method, and run evaluations on a second dataset of English videos. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.9 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.37 In this paper, we introduce a new methodology for the retrieval of product features and opinions from a collection of free-text customer reviews about a product or service. The proposal relies on a language modeling framework that can be applied to reviews in any domain and language provided with a minimal knowledge source of sentiments or opinions (e.g., a seed set of opinion words). The methodology combines both a kernel- based model of opinion words (learned from the knowledge source of sentiments or opinions) and a statistical mapping between words to approximate a model of product features from which the retrieval is carried out. To validate the usefulness of the proposal, we carried out experiments over several collections of customer reviews about products from different industry domains and languages (specifically, English and Spanish). We also compare the obtained results on the retrieval of product features to closely related work on extracting product features from customer reviews. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.37 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.36 Reviews about products and services are abundantly available online. However, selecting information relevant to a potential buyer involves a significant amount of time reading user's reviews and weeding out comments unrelated to the important aspects of the reviewed entity. In this work, we present Starlet, a novel approach to extractive multi-document summarization for evaluative text that considers aspect rating distributions and language modeling as summarization features. These features encourage the inclusion of sentences in the summary that preserve the overall opinion distribution expressed across the original reviews and whose language best reflects the language of reviews. We demonstrate how this method offers improvements over traditional summarization techniques and other approaches to multi-document summarization of evaluative text. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.36 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.35 Most severe disasters cause large human population movements and evacuations. Understanding and predicting these movements is critical for planning effective humanitarian relief, disaster management, and long-term societal reconstruction. In this paper, we present an intelligent system called DBAPRS for analyzing and simulating of human evacuation behaviors during large-scale disasters in Japan. DBAPRS stores the GPS records from mobile devices used by approximately 1.6 million people throughout Japan from 1 August 2010 to 31 July 2011. By mining this enormous set of Auto-GPS mobile sensor data, the short-term and long-term evacuation behaviors during the Great East Japan Earthquake and the Fukushima nuclear accident throughout the whole country are able to be automatically discovered and analyzed. Meanwhile, DBAPRS utilizes the discovered evacuations to effectively learn a probabilistic model to better understand and simulate human mobility during the disasters. Based on the training model, population mobility in various cities impacted by the disasters throughout Japan is able to be automatically simulated or predicted. On the basis of such kind of intelligent system, it is easy for us to find some new features or population mobility patterns after the recent severe earthquake, tsunami and release of radioactivity in Japan, which are likely to play a vital role in future disaster relief and management worldwide. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.35 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.34 In this contribution we focus on the task of automatically analyzing a speaker's sentiment in on-line videos containing movie reviews. In addition to textual information, we consider adding audio features as typically used in speech-based emotion recognition as well as video features encoding valuable valence information conveyed by the speaker. We combine this multi-modal experimental setup with a detailed analysis of different methods for linguistic sentiment analysis by gradually increasing the level of domain-independence: First, we consider in-domain analysis by examining a cross-validation setup applied on a novel database named Multi-Modal Movie Opinion (ICT-MMMO) corpus. Next, we concentrate on cross-domain analysis by using a large corpus of written movie reviews for training. Finally, we explore the application of on-line knowledge sources for inferring the speaker's sentiment. Our experimental results indicate that training on written movie reviews is a promising alternative to exclusively using (spoken) in-domain data for building a system that analyses spoken movie review videos and that language-independent audiovisual analysis can compete with linguistic analysis. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.34 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.32 This paper studies a machine learning strategy called co-transfer learning. Unlike many previous transfer learning problems, we focus on how to use labeled data of different feature spaces to enhance the classification of different learning spaces simultaneously. Our idea is to model the problem as a coupled Markov chain with restart. The transition probabilities in the coupled Markov chain can be constructed by using the intra-relationships based on affinity metric among instances in the same space, and the inter-relationships based on co-occurrence information among instances from different spaces. The learning algorithm computes ranking of labels to indicate the importance of a set of labels to an instance by propagating the ranking score of labeled instances via the coupled Markov chain with restart. Experimental results on benchmark data (multi-class image-text and English-Spanish-French classification data sets) have shown that the learning algorithm is computationally efficient, and effective in learning across different spaces. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.32 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.31 This paper describes a framework that allows the collaborative development and deployment of procedural knowledge for task support in emergency situations. In this framework, procedural knowledge is represented in a wiki using an informal, textual description that is marked up with formal tags based on the <I-N-C-A> representation for hierarchical task networks used in AI planning. Procedural knowledge in the wiki can be used for task support by way of enhanced browsing facilities and the planning capabilities of an HTN planner. The latter supports the automatic composition of procedures to form plans for specific tasks. The tight integration of collaborative editing with deployment is new in this system and advances knowledge engineering for planning domain knowledge, that is, procedural knowledge. An experimental evaluation has shown that the explicit availability of procedural knowledge in emergency situations can reduce procedural uncertainty. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.31 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.29 The detection of super spreading events and the characteristic analysis of super spreaders are crucial for public health emergency response. Super spreaders are high-risk core groups in the transmission of infectious diseases. One super spreader or super spreading event can ignite a new epidemic outbreak. Potential super spreaders should be identified to support decision-making for preventing epidemic outbreaks. Homogeneous mathematical models are widely used to study the spread of infectious diseases. However those models are hard to depict the heterogeneity of individuals. Here heterogeneous and stochastic agent-based models are proposed to explore the mechanism of super spreading events. We built an agent society including the models of severe acute respiratory syndrome epidemic progress, human contact patterns, weighted scale-free networks, and infection probabilities. Through computational experiments on our models, we found that some factors led an infectious agent to infect many others, including long delayed admission time, active contact patterns, and high pathogen load and shedding rates. Furthermore, super spreading events and large scale epidemic outbreaks were more likely to emerge if an imported case infected many other agents. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.29 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.27 The domain adaptation problem arises often in the field of sentiment classification. In the problem of domain adaptation, there are two distinct needs, namely labeling adaptation and instance adaptation. However, most of current research focuses on the former attribute, while neglects the latter one. We propose a comprehensive approach, named feature ensemble plus sample selection (SS-FE), which takes both types of adaptation into account. A feature ensemble (FE) model is first proposed to learn a new labeling function in a feature re-weighting manner. Furthermore, a PCA-based sample selection (PCA-SS) method is proposed as an aid to FE. Experimental results show that the proposed SS-FE approach could gain significant improvements, compared to FE and PCA-SS, due to its comprehensive consideration of both labeling adaptation and instance adaptation. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.27 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.25 Sentiment analysis has become a hot topic in recent years, and sentiment dictionaries are essential for research in this field. We propose a two-step method combining iterative regression and random walk with in-link normalization to build a concept-level sentiment dictionary with accurate values and a large vocabulary. We use ConceptNet as a framework to propagate sentiment values based on the assumption that semantically related concepts share common sentiment. Instead of mean error, we propose using polarity accuracy, Kendall τ distance, and average-maximum ratio to evaluate sentiment dictionaries. We collect our evaluation data using Amazon Mechanical Turk. Our proposed two-step method achieves better results than tested single-step methods. Moreover, it also outperforms the state-of-the-art sentiment dictionary in terms of both polarity accuracy and Kendall τ distance. In particular, Kendall τ distance decreases 22% relatively. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.25 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.10 A new generation of observational science instruments is dramatically increasing collected data volumes in a range of fields. These instruments include the Square Kilometre Array (SKA), Large Synoptic Survey Telescope (LSST), terrestrial sensor networks, and NASA satellites participating in "decadal survey" missions. Their unprecedented coverage and sensitivity will likely reveal wholly new categories of unexpected and transient events. Commensal methods passively analyze these data streams, recognizing anomalous events of scientific interest and reacting in real time. We report on a case example: V-FASTR, an ongoing commensal experiment at the Very Long Baseline Array (VLBA) that uses online adaptive pattern recognition to search for anomalous fast radio transients. V-FASTR triages a millisecond-resolution stream of data and promotes candidate anomalies for further offline analysis. It tunes detection parameters in real time, injecting synthetic events to continually retrain itself for optimum performance. This self-tuning approach retains sensitivity to weak signals while adapting to changing instrument configurations and noise conditions. The system has operated since July 2011, making it the longest-running real time commensal radio transient experiment to date. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.10 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.4 Recent studies show that concept-based approaches to opinion mining perform better than more canonical methods based on keyword spotting or word co-occurrence frequencies. SenticNet 1.0 is one of the most widely used publicly available resources for concept-based opinion mining. It gives polarity scores for a large number of single- and multi-word common sense concepts. However, developing high-quality opinion mining and sentiment analysis systems also requires affective information associated with the concepts. In this work, we present a methodology for enriching SenticNet concepts with affective information by assigning to them an emotion label. The created resource is freely available for academic use. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.4 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.3 The paper provides an analysis of a new type of mobile sensing data, called Auto-GPS, for the purpose of discerning human mobility and behavior during a large-scale crisis. More than nine billion Auto-GPS traces were collected nationwide from August 2010 to July 2011 in Japan. After the Great Japan earthquake on March 11, we analyzed the shifts in activity among populations at the heart of the most impacted areas. Our results reveal useful information on how humans react in disaster scenarios and how the evacuation process can be monitored in near real-time by decision-makers. This research suggests a number of promising new directions for crisis response and management. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.3 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.2 Demand response programs are an important resource which can significantly increase the efficiency of future smart grids. However, preliminary experiences evidence some difficulties in making the use of these programs successful, and also increased difficulties in extending them to include relevant participation of small consumers. This paper proposes a distributed agent-based intelligent system to model and simulate a smart grid with a diversity of players, each one with his own specific configuration and goals. This system accommodates the use of physical players, e.g., real electrical installations, as well as computationally simulated agents. This paper presents the capacities of the proposed system to simulate the use of demand response programs. The system allows assessing the impact of these programs to the involved consumers, to the other players and to the whole system. In this way, the alternative demand response program structure and parameterization can be simulated and evaluated. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.2 http://doi.ieeecomputersociety.org/10.1109/MIS.2013.1 To improve the performance of sentiment classification for word of mouth (WOM), this research re-evaluates objective sentiment words in a sentiment lexicon, SentiWordNet. SentiWordNet provides each synonymous set with three sentiment values regarding positivity, negativity, and objectivity. As the evaluation of sentiments of words in WOM is useful for sentiment classification, SentiWordNet has become a public and popular lexicon resource. This sentiment lexicon includes 117,659 entries but 93.75% of them have a stronger objective sentiment tendency than their sentimental counterparts. A sentiment lexicon such as SentiWordNet with so many objective words may suffer from noise in the application of sentiment classification. This research revises sentiment value and tendency for objective words in SentiWordNet based on assessment of the co-relevance of each objective word and its associated sentiment sentences. Experiments show that this proposed approach is significantly better than the traditional non-revised approach as evaluated by the classification accuracy criterion. http://doi.ieeecomputersociety.org/10.1109/MIS.2013.1 http://doi.ieeecomputersociety.org/10.1109/MIS.2012.119 Ambient intelligence (AmI) implies intelligence that is all around us. It is to make our everyday environments sensitive, adaptive, and responsive to the presence of people in a transparent manner. There exist several challenges to build an efficient infrastructure for AmI, e.g. interoperation of heterogeneous systems; intelligence for anticipatory user assistance; adaptability to dynamic environments for good user experience; and scalability to more users and spaces. This article proposes a smart SOA (Service Oriented Architecture, an architectural style whose goal is to achieve loose coupling among interacting software agents) framework for AmI systems, called Pervasive Service Bus (PSB), where all the computing activities are modeled as unified pervasive services. We present an on-line planning algorithm actively adapt service flows to both dynamic contexts and user tasks. With a proposed sub-bus-based layout, PSB enables effective access and interaction of distributed services. PSB exploits the scheme of direct link and distributed execution to keep efficient under large-scale service interactions. We have evaluated PSB’s performance with a real smart home. http://doi.ieeecomputersociety.org/10.1109/MIS.2012.119 http://doi.ieeecomputersociety.org/10.1109/MIS.2012.118 The ability to understand natural language text is far from being emulated in machines. One of the main hurdles to overcome is that computers lack both the common and common-sense knowledge humans normally acquire during the formative years of their lives. In order to really understand natural language, a machine should be able to grasp such kind of knowledge, rather than merely relying on the valence of keywords and word co-occurrence frequencies. In this work, the largest existing taxonomy of common knowledge is blended with a natural-language-based semantic network of common-sense knowledge, and multi-dimensional scaling is applied on the resulting knowledge base for open-domain opinion mining and sentiment analysis. http://doi.ieeecomputersociety.org/10.1109/MIS.2012.118 http://doi.ieeecomputersociety.org/10.1109/MIS.2012.106 This paper investigates the collaboration patterns and research interests in the field of Intelligence and Security Informatics (ISI) over the past decade.With the methodology of social network analysis,We constructed a co-authorship network to examine the key researchers and their collaboration patterns, and a keyword co-occurrence network to study the ISI research topics. Our studies aim to provide significant insights for researchers to better understand and evaluate the key impact factors and quickly grasp the emerging direction in the field of ISI. http://doi.ieeecomputersociety.org/10.1109/MIS.2012.106 http://doi.ieeecomputersociety.org/10.1109/MIS.2012.85 In today's business environment, competition within the industries is becoming more and more intense. In order to survive in this fast paced competitive environment, it is important to know what the competitive technology is and how the technologies can be grouped. Therefore, this study focuses on discovering the core technology and clustering technologies simultaneously using patent citation network where the core technology is represented as an influential node and the technology group as a cluster of nodes. However, existing methods have discovered the influential nodes and cluster nodes separately, especially in a citation network. This article develops the method to detect the influential nodes and clusters simultaneously in a patent citation network. It can allow an important patent in each patent group to be discovered easily and the distribution of the similar patents around the important patent to be recognized. For the study of detecting important patents and their groups simultaneously, kernel k-means clustering method with graph kernel is introduced. A graph kernel helps to compute implicitly similarities between patents in a high-dimensional feature space, even if the similarities between structured patents cannot be explicitly represented. The proposed approaches are compared to the widely used centrality measures using US patents data in the area of information and security. http://doi.ieeecomputersociety.org/10.1109/MIS.2012.85 http://doi.ieeecomputersociety.org/10.1109/MIS.2012.63 The transformation of the IT governmental services (T-Gov) has become an emergent challenge, requiring the adoption of a whole new perception in terms of service delivery to citizens. To achieve T-Gov, Citizen-to-Government interaction should be explored under a new perspective, enhancing citizen active involvement. Such a perspective should take into account all types of Government interaction and support the notion of intermediates acting on citizen’s behalf, while third-party application development should be embraced. To this end, a citizen-centric interaction model is proposed for the provision of e-Government services, utilizing the Web 2.0 paradigm. It facilitates citizens to better control their private data through profiles and enables them to combine services provided by the Government or third-party entities to perform cross-organizational tasks. A cross-organizational service example is used to illustrate its potential. The OpenSocialGov framework, a social network platform implemented for the realisation of the proposed interaction model, is also discussed. http://doi.ieeecomputersociety.org/10.1109/MIS.2012.63 http://doi.ieeecomputersociety.org/10.1109/MIS.2012.24 Using Graham's rules on picking stocks has been proven to generate profits for value investors. We propose using 3D subspace clustering to generate rules to pick potential undervalued stocks; 3D subspace clustering is effective in handling high dimensional financial data, is adaptive to new data, and its results are not influenced by human's biases and emotions, and are easily interpretable. We conducted extensive experimentation in the stock market over a period of 28 years (from 1980 to 2007). We found that using rules generated by two 3D subspace clustering algorithms, CATSeeker and MIC, results in 60% more profits than using Graham's rules alone. http://doi.ieeecomputersociety.org/10.1109/MIS.2012.24 http://doi.ieeecomputersociety.org/10.1109/MIS.2012.7 Testing is essential part of the development of human-agent-robot team (HART) applications. Individual algorithms in such applications cannot be tested in isolation as their performance depends significantly on complex interactions among distributed software code, humans, hardware and the target environment. Any testing involving robots and human actors is, however, time-consuming and costly. We therefore propose an incremental development framework employing mixed-reality testbeds, which can reduce testing cost by replacing parts of the application and surrounding reality with synthetic computational models. The proposed framework introduces the concept of testbed fidelity and proposes how test reliability and cost should be managed to maximize the effectiveness of the development process. The framework is illustrated on two example applications in the domain of multi-UAV tracking and anti-maritime piracy operations. http://doi.ieeecomputersociety.org/10.1109/MIS.2012.7 http://doi.ieeecomputersociety.org/10.1109/MIS.2011.93 This paper studies how people reveal private information in strategic settings in which participants need to negotiate over resources, but are uncertain about each other's objectives. The study compared two negotiation protocols which differed in whether they allowed participants to disclose their objectives in a repeated negotiation setting of incomplete information. Results show that most people agree to reveal their goals when asked, and this leads participants to more beneficial agreements. Machine learning was used to model the likelihood that people reveal their goals in negotiation, and this model was used to make goal request decisions in the game. In simulation, use of this model is shown to outperform people making the same type of decisions. These results demonstrate the benefit of this approach towards designing agents to negotiate with people under incomplete information. http://doi.ieeecomputersociety.org/10.1109/MIS.2011.93 http://doi.ieeecomputersociety.org/10.1109/MIS.2011.41 Intelligent System (IS) techniques have been continuously attracting attentions for a variety of electric power engineering applications. A recent article has summarized, from a selection of the state-of-the-art research works, the key areas in which power systems and energy markets can benefit most from IS techniques. Along with the notable capability in solving problems like electricity market simulation, market risk management, power grid planning and voltage control, the potential of IS in facilitating power system real-time stability assessment for blackout prevention has also been clearly shown in recent years. This article systematically describes the incentive, benefit, and process of constructing IS for power system real-time stability assessment, and discusses some most critical issues in the development and implementation stages. Also, on the basis of our own research output, some possible solutions to the problems are provided. Hopefully, more attention and research efforts can be drawn, to this promising while challenging topic. http://doi.ieeecomputersociety.org/10.1109/MIS.2011.41 http://doi.ieeecomputersociety.org/10.1109/MIS.2011.36 The aim of educational systems is to assemble learning objects on a set of topics tailored to the goals and individual students' styles. Given the amount of available learning objects, the challenge of e-learning is to select the proper objects, define their relationships, and adapt their sequencing (i.e. course composition) to the specific needs, objectives and background of the student. This paper describes the general requirements for this course adaptation, the full potential of applying planning techniques on the construction of personalized e-learning routes, and how to accommodate the temporal and resource constraints to make the course applicable in a real scenario. http://doi.ieeecomputersociety.org/10.1109/MIS.2011.36 http://doi.ieeecomputersociety.org/10.1109/MIS.2011.35 Demand response (DR) is not a new concept but it is gaining a growing focus of attention in nowadays electric power systems operation and planning, with several advantages for the reliable power system functioning and for electricity prices. In this paper, price-based DR is applied to electricity consumers through the management of electricity prices. This management is based on demand elasticity and consumers are expected to react enabling to accomplish the required load reduction. The methodology is implemented in a developed DR simulator – DemSi - that uses PSCAD® for technical validation of solutions and Particle Swarm Optimization (PSO) for solution optimization. The performance of PSO is evaluated in terms of running time and obtained solutions in comparison with the Non-Linear Programming (NLP) solutions obtained in GAMS™. Case studies involving 32 and 320 consumers are used to illustrate the proposed methodology and to discuss its performance. http://doi.ieeecomputersociety.org/10.1109/MIS.2011.35 http://www.computer.org/portal/site/intelligent/ IEEE Intelligent Systems http://www.computer.org/portal/site/intelligent/