Search

Graph encoding methods have been proven exceptionally useful in many classification tasks — from molecule toxicity prediction to social network recommendations. However, most of the existing methods are designed to work in a centralized environment that requires the whole graph to be kept in memory. Moreover, scaling them on very large networks remains a challenge. In this work, we propose a distributed and permutation invariant graph embedding method denoted as Distributed Graph Statistical Distance (DGSD) that extracts graph representation on independently distributed machines. DGSD finds nodes’ local proximity by considering only nodes’ degree, common neighbors and direct connectivity that allows it to run in the distributed environment. On the other hand, the linear space complexity of DGSD makes it suitable for processing large graphs. We show the scalability of DGSD on sufficiently large random and real-world networks and evaluate its performance on various bioinformatics and social networks with the implementation in a distributed computing environment.

Is it possible to identify crime suspects by their mobile phone call records? Can the spatial-temporal movements of individuals linked to convicted criminals help to identify those who facilitate crime? Might we leverage the usage of mobile phones, such as incoming and outgoing call numbers, coordinates, call duration and frequency of calls, in a specific time window on either side of a crime to provide a focus for the location and period under investigation? Might the call data records of convicted criminals' social networks serve to distinguish criminals from non-criminals? To address these questions, we used heterogeneous call data records dataset by tapping into the power of social network analysis and the advancements in graph convolutional networks. In collaboration with the Punjab Police and Punjab Information Technology Board, these techniques were useful in identifying convicted individuals. The approaches employed are useful in identifying crime suspects and facilitators to support smart policing in the fight against the country's increasing crime rates. Last but not least, the applied methods are highly desirable to complement high-cost video-based smart city surveillance platforms in developing countries.

The advancements of search engines for traditional text documents have enabled the effective retrieval of massive textual information in a resource-efficient manner. However, such conventional search methodologies often suffer from poor retrieval accuracy especially when documents exhibit unique properties that behoove specialized and deeper semantic extraction. Recently, AlgorithmSeer, a search engine for algorithms has been proposed, that extracts pseudo-codes and shallow textual metadata from scientific publications and treats them as traditional documents so that the conventional search engine methodology could be applied. However, such a system fails to facilitate user search queries that seek to identify algorithm-specific information, such as the datasets on which algorithms operate, the performance of algorithms, and runtime complexity, etc. In this paper, a set of enhancements to the previously proposed algorithm search engine are presented. Specifically, we propose a set of methods to automatically identify and extract algorithmic pseudo-codes and the sentences that convey related algorithmic metadata using a set of machine-learning techniques. In an experiment with over 93,000 text lines, we introduce 60 novel features, comprising content-based, font style based and structure-based feature groups, to extract algorithmic pseudo-codes. Our proposed pseudo-code extraction method achieves 93.32% F1-score, outperforming the state-of-the-art techniques by 28%. Additionally, we propose a method to extract algorithmic-related sentences using deep neural networks and achieve an accuracy of 78.5%, outperforming a Rule-based model and a support vector machine model by 28% and 16%, respectively.

Although over 64 million people worldwide speak Urdu language and are well aware of its Roman script, limited research and efforts have been made to carry out sentiment analysis and build language resources for the Roman Urdu language. This article proposes a deep learning model to mine the emotions and attitudes of people expressed in Roman Urdu - consisting of 10,021 sentences from 566 online threads belonging to the following genres: Sports; Software; Food & Recipes; Drama; and Politics. The objectives of this research are twofold: (1) to develop a human-annotated benchmark corpus for the under-resourced Roman Urdu language for the sentiment analysis; and (2) to evaluate sentiment analysis techniques using the Rule-based, N-gram, and Recurrent Convolutional Neural Network (RCNN) models. Using Corpus, annotated by three experts to be positive, negative, and neutral with 0.557 Cohen's Kappa score, we run two sets of tests, i.e., binary classification (positive and negative) and tertiary classification (positive, negative and neutral). Finally, the results of the RCNN model are analyzed by comparing it with the outcome of the Rule-based and N-gram models. We show that the RCNN model outperforms baseline models in terms of accuracy of 0.652 for binary classification and 0.572 for tertiary classification.

In the current socio-economic environment, to face challenges such as the emergence of new technologies, globalisation and increasing demands from their clients it is inevitable that enterprises will collaborate with others and progressively shift their boundaries. In this context, interoperability has become a prerequisite in the jigsaw of such collaboration. By definition, it is entities’ ability to work together as an organisation. This ability spans a wide range of aspects, embracing both technical and business issues. Over the past decade, both the concept and the context of interoperability have been extended from a largely IT-focused domain to a business-focused domain and the evaluation of interoperability has become a rising concern. An increasing number of studies have concentrated on not just digital but business aspects of human behaviour in the social environment. In general, the wider application domain is the assessment of the interoperability of information systems and processes in any organisation (especially medium and large) that needs multiple processes to interact effectively. To deal with such concerns and pave the way to achievement of more effective collaborative goals in business, the concept of interoperability has been adopted to measure the efficiency and productivity of information systems’ integration. More than twenty approaches have so far been adopted to evaluate this interoperability, however most are unable to assess it at the higher levels, such as at the pragmatic, process and social levels. Hence, we have conducted a three-phase study. Phase 1 reviewed existing interoperability evaluation approaches. To prove the concept, phase 2 proposed the concept of semiotic interoperability and its application to healthcare information systems. This article reports on the third phase of the study, a proposed framework with a group of metrics to measure interoperability from a new perspective – a semiotics perspective. The framework is named

Foreign Exchange or Forex is the sale purchase market point of foreign currency pairs. Due to the high volatility in the forex market, it is difficult to predict the future price of any currency pair. This study shows that a significant enhancement in the prediction of forex price can be achieved by incorporating domain knowledge in the process of training machine learning models. The proposed system integrates the Forex Loss Function (FLF) into a Long Short-Term Memory model called FLF-LSTM — that minimizes the difference between the actual and predictive average of Forex candles. Using the data of 10,078 four-hour candles of EURUSD pair, it is found that compared to the classic LSTM model, the proposed FLF-LSTM system shows a decrease in overall mean absolute error rate by 10.96%. It is also reported that the error in forecasting the high and low prices is reduced by 10% and 9%, respectively. The proposed model, in comparison to the Recurrent Neural Network-based prediction system, shows an overall reduction of 73.57% in mean absolute error, by exhibiting up to 68.71% and 72.31% error reduction in high and low prices, respectively. In comparison to Auto-Regressive Integrated Moving Average, our proposed model shows a 13% reduced error. Specifically, in the open, high, and low prices, the error is reduced by 28.5%, 14.2%, 9.3%, respectively. Finally, we compare our model with another well-known time series forecasting model, i.e., FB Prophet — where FLF-LSTM demonstrates 31.8%, 47.7%, 23.6%, 47.7% error reduction in open, high, low, and close prices, respectively. The data and the code used in this study can be accessed at the following URL: https://github.com/slab-itu/forex_flf_lstm.

Text simplification and text summarisation are related, but different sub-tasks in Natural Language Generation. Whereas summarisation attempts to reduce the length of a document, whilst keeping the original meaning, simplification attempts to reduce the complexity of a document. In this work, we combine both tasks of summarisation and simplification using a novel hybrid architecture of abstractive and extractive summarisation called HTSS. We extend the well-known pointer generator model for the combined task of summarisation and simplification. We have collected our parallel corpus from the simplified summaries written by domain experts published on the science news website EurekaAlert (www.eurekalert.org). Our results show that our proposed HTSS model outperforms neural text simplification (NTS) on SARI score and abstractive text summarisation (ATS) on the ROUGE score. We further introduce a new metric (CSS1) which combines SARI and Rouge and demonstrates that our proposed HTSS model outperforms NTS and ATS on the joint task of simplification and summarisation by 38.94% and 53.40%, respectively. We provide all code, models and corpora to the scientific community for future research at the following URL: https://github.com/slab-itu/HTSS/.

Traditional bibliometric techniques gauge the impact of research through quantitative indices based on the citations data. However, due to the lag time involved in the citation-based indices, it may take years to comprehend the full impact of an article. This paper seeks to measure the early impact of research articles through the sentiments expressed in tweets about them. We claim that cited articles in either positive or neutral tweets have a more significant impact than those not cited at all or cited in negative tweets. We used the SentiStrength tool and improved it by incorporating new opinion-bearing words into its sentiment lexicon pertaining to scientific domains. Then, we classified the sentiment of 6,482,260 tweets linked to 1,083,535 publications covered by Altmetric.com. Using positive and negative tweets as an independent variable, and the citation count as the dependent variable, linear regression analysis showed a weak positive prediction of high citation counts across 16 broad disciplines in Scopus. Introducing an additional indicator to the regression model, i.e. ‘number of unique Twitter users’, improved the adjusted R-squared value of regression analysis in several disciplines. Overall, an encouraging positive correlation between tweet sentiments and citation counts showed that Twitter-based opinion may be exploited as a complementary predictor of literature’s early impact.

The abundance of accessible educational data, supported by the technology-enhanced learning platforms, provides opportunities to mine learning behavior of students, addressing their issues, optimizing the educational environment, and enabling data-driven decision making. Virtual learning environments complement the learning analytics paradigm by effectively providing datasets for analysing and reporting the learning process of students and its reflection and contribution in their respective performances. This study deploys a deep artificial neural network on a set of unique handcrafted features, extracted from the virtual learning environments clickstream data, to predict at-risk students providing measures for early intervention of such cases. The results show the proposed model to achieve a classification accuracy of 84%-93%. We show that a deep artificial neural network outperforms the baseline logistic regression and support vector machine models. While logistic regression achieves an accuracy of 79.82% - 85.60%, the support vector machine achieves 79.95% - 89.14%. Aligned with the existing studies - our findings demonstrate the inclusion of legacy data and assessment-related data to impact the model significantly. Students interested in accessing the content of the previous lectures are observed to demonstrate better performance. The study intends to assist institutes in formulating a necessary framework for pedagogical support, facilitating higher education decision-making process towards sustainable education.

Is it possible to identify crime suspects by their mobile phone call records? Can the spatial-temporal movements of individuals linked to convicted criminals help to identify those who facilitate crime? Might we leverage the usage of mobile phones, such as incoming and outgoing call numbers, coordinates, call duration and frequency of calls, in a specific time window on either side of a crime to provide a focus for the location and period under investigation? Might the call data records of convicted criminals' social networks serve to distinguish criminals from non-criminals? To address these questions, we used heterogeneous call data records dataset by tapping into the power of social network analysis and the advancements in graph convolutional networks. In collaboration with the Punjab Police and Punjab Information Technology Board, these techniques were useful in identifying convicted individuals. The approaches employed are useful in identifying crime suspects and facilitators to support smart policing in the fight against the country's increasing crime rates. Last but not least, the applied methods are highly desirable to complement high-cost video-based smart city surveillance platforms in developing countries.

Traditional bibliometric techniques gauge the impact of research through quantitative indices based on the citations data. However, due to the lag time involved in the citation-based indices, it may take years to comprehend the full impact of an article. This paper seeks to measure the early impact of research articles through the sentiments expressed in tweets about them. We claim that cited articles in either positive or neutral tweets have a more significant impact than those not cited at all or cited in negative tweets. We used the SentiStrength tool and improved it by incorporating new opinion-bearing words into its sentiment lexicon pertaining to scientific domains. Then, we classified the sentiment of 6,482,260 tweets linked to 1,083,535 publications covered by Altmetric.com. Using positive and negative tweets as an independent variable, and the citation count as the dependent variable, linear regression analysis showed a weak positive prediction of high citation counts across 16 broad disciplines in Scopus. Introducing an additional indicator to the regression model, i.e. ‘number of unique Twitter users’, improved the adjusted R-squared value of regression analysis in several disciplines. Overall, an encouraging positive correlation between tweet sentiments and citation counts showed that Twitter-based opinion may be exploited as a complementary predictor of literature’s early impact.

The abundance of accessible educational data, supported by the technology-enhanced learning platforms, provides opportunities to mine learning behavior of students, addressing their issues, optimizing the educational environment, and enabling data-driven decision making. Virtual learning environments complement the learning analytics paradigm by effectively providing datasets for analysing and reporting the learning process of students and its reflection and contribution in their respective performances. This study deploys a deep artificial neural network on a set of unique handcrafted features, extracted from the virtual learning environments clickstream data, to predict at-risk students providing measures for early intervention of such cases. The results show the proposed model to achieve a classification accuracy of 84%-93%. We show that a deep artificial neural network outperforms the baseline logistic regression and support vector machine models. While logistic regression achieves an accuracy of 79.82% - 85.60%, the support vector machine achieves 79.95% - 89.14%. Aligned with the existing studies - our findings demonstrate the inclusion of legacy data and assessment-related data to impact the model significantly. Students interested in accessing the content of the previous lectures are observed to demonstrate better performance. The study intends to assist institutes in formulating a necessary framework for pedagogical support, facilitating higher education decision-making process towards sustainable education.