Misinformation spreading over the Internet poses a significant threat to both societies and individuals, necessitating robust and scalable fact-checking that relies on retrieving accurate and trustworthy evidence. Previous methods rely on semantic and social-contextual patterns learned from training data, which limits their generalization to new data distributions. Recently, Retrieval Augmented Generation (RAG) based methods have been proposed to utilize the reasoning capability of LLMs with retrieved grounding evidence documents. However, these methods largely rely on textual similarity for evidence retrieval and struggle to retrieve evidence that captures multi-hop semantic relations within rich document contents. These limitations lead to overlooking subtle factual correlations between the evidence and the claims to be fact-checked during evidence retrieval, thus causing inaccurate veracity predictions.

Providing appropriate answers to questions is necessary in many situations, not just in the conversational AI systems we see and develop today. Research in Natural Language Processing (NLP) and User Modeling (UM) have investigated this topic for decades, starting in the era of ''symbolic AI''. While NLP in general was needed for the whole interaction (understanding the question and answering it), Natural Language Generation (NLG) was particularly concerned with providing good and coherent answers appropriate for the information need and the intended audience, which is where UM also played a role. At that time, information to include in the answers typically came from knowledge bases or data bases. Information Retrieval (IR) then was concerned with retrieving the documents (and later websites) most relevant to a query. As the amount of data and number of documents increased, information needs from users became increasingly complex. As a result, it seemed that combining advances in both NLP and IR was required. And of course, now, research often spans these two fields.

In human–robot teams, human situational awareness is the operator’s conscious knowledge of the team’s states, actions, plans and their environment. Appropriate human situational awareness is critical to successful human–robot collaboration. In human–robot teaming, it is often assumed that the best and required level of situational awareness is knowing everything at all times. This view is problematic, because what a human needs to know for optimal team performance varies given the dynamic environmental conditions, task context, and roles and capabilities of team members. We explore this topic by interviewing 16 participants with active and repeated experience in diverse human–robot teaming applications. Based on analysis of these interviews, we derive a framework explaining the dynamic nature of required situational awareness in human–robot teaming. In addition, we identify a range of factors affecting the dynamic nature of required and actual levels of situational awareness (i.e., dynamic situational awareness), types of situational awareness inefficiencies resulting from gaps between actual and required situational awareness, and their main consequences. We also reveal various strategies, initiated by humans and robots, that assist in maintaining the required situational awareness. Our findings inform the implementation of accurate estimates of dynamic situational awareness and the design of user-adaptive human–robot interfaces. Therefore, this work contributes to the future design of more collaborative and effective human–robot teams.

In an era where misinformation spreads freely, factchecking (FC) plays a crucial role in verifying claims and promoting reliable information.While automated fact-checking (AFC) has advanced significantly, existing systems remain vulnerable to adversarial attacks that manipulate or generate claims, evidence, or claim-evidence pairs.These attacks can distort the truth, mislead decision-makers, and ultimately undermine the reliability of FC models.Despite growing research interest in adversarial attacks against AFC systems, a comprehensive, holistic overview of key challenges remains lacking.These challenges include understanding attack strategies, assessing the resilience of current models, and identifying ways to enhance robustness.This survey provides the first in-depth review of adversarial attacks targeting FC 1 , categorizing existing attack methodologies and evaluating their impact on AFC systems.Additionally, we examine recent advancements in adversary-aware defenses and highlight open research questions that require further exploration.Our findings underscore the urgent need for resilient FC frameworks capable of withstanding adversarial manipulations in pursuit of preserving high verification accuracy.Attack Target Edit Granularity Attack Technique Claim attack Evidence attack Claim-evidence pair attack Generate Manipulate Generate Manipulate

In this paper, we generate two types of explanations, Impact and Critique, of predictions made by a Goal Recognizer (GR) – a system that infers agents’ goals from observations. Impact explanations describe the top-predicted goal(s) and the main observations that led to these predictions. Critique explanations augment these explanations with evidence that challenges the GR’s predictions if so warranted. Our user study compares users’ goal-recognition accuracy for Impact and Critique explanations, and users’ views about these explanations, under three prediction-correctness conditions: correct, partially correct and incorrect. Our results show that (1) users stick with a GR’s predictions, even when a Critique explanation highlights its flaws; yet (2) Critique explanations are deemed better than Impact explanations in most respects.

The increasing complexity of decision-making in dynamic environments, particularly in high-stakes domains like cybersecurity, demands more than automated solutions—it requires effective integration of human expertise with advanced AI capabilities. While approaches like ensemble learning and Mixture of Experts (MoE) enhance automated decision-making, they struggle with handling uncertainty and novel scenarios. Techniques such as learning to defer and learning to complement mitigate this by incorporating human input, but assume that a definitive expert is always available—an assumption that often fails in real-world settings. To bridge this gap, we introduce A 2 C , a modular, multi-stage collaborative decision-making framework that enhances adaptability and decision robustness under uncertainty by seamlessly transitioning between three decision-making modes: Automated, Augmented Deferral, and Collaborative Exploration (CoEx). A key innovation of CoEX is its ability to handle cases where both AI and human experts face uncertainty, overcoming a critical limitation of traditional deferral systems. We validate A 2 C through experiments on benchmark datasets, Large Language Model (LLM) simulations of human–AI collaboration, and real-world human–AI interaction studies with cybersecurity researchers. Results show that A 2 C consistently outperforms conventional approaches that rely solely on full automation or selective human intervention, demonstrating its potential as a practical and scalable solution for expert decision-making in complex domains. For image detection on CIFAR-10, detection rates improved from 37.8% with automation alone to 64.75% with augmented deferral, and further to 92.95% with collaborative exploration. Similarly, for intrusion detection on KDDCup, rates rose from 33.43% with automation to 35.18% with augmented deferral, and finally reached 87.04% with CoEx, highlighting its effectiveness in handling uncertainty.

Large language models are able to generate code for visualisations in response to simple user requests.This is a useful application and an appealing one for NLP research because plots of data provide grounding for language.However, there are relatively few benchmarks, and those that exist may not be representative of what users do in practice.This paper investigates whether benchmarks reflect realworld use through an empirical study comparing benchmark datasets with code from public repositories.Our findings reveal a substantial gap, with evaluations not testing the same distribution of chart types, attributes, and actions as real-world examples.One dataset is representative, but requires extensive modification to become a practical end-to-end benchmark.This shows that new benchmarks are needed to support the development of systems that truly address users' visualisation needs.These observations will guide future data creation, highlighting which features hold genuine significance for users.

In this paper, we generate and compare three types of explanations of Machine Learning (ML) predictions: simple, conservative and unifying.Simple explanations are concise, conservative explanations address the surprisingness of a prediction, and unifying explanations convey the extent to which an ML model's predictions are applicable.The results of our user study show that (1) conservative and unifying explanations are liked equally and considered largely equivalent in terms of completeness, helpfulness for understanding the AI, and enticement to act, and both are deemed better than simple explanations; and (2) users' views about explanations are influenced by the (dis)agreement between the ML model's predictions and users' estimations of these predictions, and by the inclusion/omission of features users expect to see in explanations.

OBJECTIVE: Active adverse event surveillance monitors Adverse Drug Events (ADE) from different data sources, such as electronic health records, medical literature, social media and search engine logs. Over the years, many datasets have been created, and shared tasks have been organised to facilitate active adverse event surveillance. However, most - if not all - datasets or shared tasks focus on extracting ADEs from a particular type of text. Domain generalisation - the ability of a machine learning model to perform well on new, unseen domains (text types) - is under-explored. Given the rapid advancements in natural language processing, one unanswered question is how far we are from having a single ADE extraction model that is effective on various types of text, such as scientific literature and social media posts. METHODS: We contribute to answering this question by building a multi-domain benchmark for adverse drug event extraction, which we named MultiADE. The new benchmark comprises several existing datasets sampled from different text types and our newly created dataset-CADECv2, which is an extension of CADEC (Karimi et al., 2015), covering online posts regarding more diverse drugs than CADEC. Our new dataset is carefully annotated by human annotators following detailed annotation guidelines. CONCLUSION: Our benchmark results show that the generalisation of the trained models is far from perfect, making it infeasible to be deployed to process different types of text. In addition, although intermediate transfer learning is a promising approach to utilising existing resources, further investigation is needed on methods of domain adaptation, particularly cost-effective methods to select useful training instances. The newly created CADECv2 and the scripts for building the benchmark are publicly available at CSIRO's Data Portal (https://data.csiro.au/collection/csiro:62387). These resources enable the research community to further information extraction, leading to more effective active adverse drug event surveillance.

Detecting a community in a network is a matter of discerning the distinct features and connections of a group of members that are different from those in other communities. The ability to do this is of great significance in network analysis. However, beyond the classic spectral clustering and statistical inference methods, there have been significant developments with deep learning techniques for community detection in recent years-particularly when it comes to handling high-dimensional network data. Hence, a comprehensive review of the latest progress in community detection through deep learning is timely. To frame the survey, we have devised a new taxonomy covering different state-of-the-art methods, including deep learning models based on deep neural networks (DNNs), deep nonnegative matrix factorization, and deep sparse filtering. The main category, i.e., DNNs, is further divided into convolutional networks, graph attention networks, generative adversarial networks, and autoencoders. The popular benchmark datasets, evaluation metrics, and open-source implementations to address experimentation settings are also summarized. This is followed by a discussion on the practical applications of community detection in various domains. The survey concludes with suggestions of challenging topics that would make for fruitful future research directions in this fast-growing deep learning field.

Security Operations Centres (SOCs) play a pivotal role in defending organisations against evolving cyber threats. They function as central hubs for detecting, analysing, and responding promptly to cyber incidents with the primary objective of ensuring the confidentiality, integrity, and availability of digital assets. However, they struggle against the growing problem of alert fatigue, where the sheer volume of alerts overwhelms SOC analysts and raises the risk of overlooking critical threats. In recent times, there has been a growing call for human-AI teaming, wherein humans and AI collaborate with each other, leveraging their complementary strengths and compensating for their weaknesses. The rapid advances in AI and the growing integration of AI-enabled tools and technologies within SOCs give rise to a compelling argument for the implementation of human-AI teaming within the SOC environment. Therefore, in this article, we present our vision for human-AI teaming to address the problem of alert fatigue in the SOC. We propose the 𝒜 2 𝒞 Framework, which enables flexible and dynamic decision making by allowing seamless transitions between automated, augmented, and collaborative modes of operation. Our framework allows AI-powered automation for routine alerts, AI-driven augmentation for expedited expert decision making, and collaborative exploration for tackling complex, novel threats. By implementing and operationalising 𝒜 2 𝒞, SOCs can significantly reduce alert fatigue while empowering analysts to efficiently and effectively respond to security incidents.

A human supervisor’s Situational Awareness (SA) is a critical aspect for successful Human-Robot Teaming (HRT). SA has been estimated using different techniques; however, many of those are associated with various biases, including recall and overgeneralisation biases. A key SA metric is latency, the delay between the time the robotic system requires supervisor assistance and the time the supervisor identifies that need in HRT experiments. Eye movements are increasingly used to assess SA across a range of domains, enabling objective and continuous SA assessment. However, to date, only a small number of features have been evaluated for estimating different types of SA latencies. In this paper, we investigated how two types of SA latencies (perceptual and comprehending) correlate with eye movement data collected during a remote field experiment, where a human supervisor directed a team of robots in a smart farming context. We identified 39 instances of SA latencies (13 perceptual and 26 comprehending). These instances were used to identify how a human supervisor’s SA is affected by task context, and to evaluate correlations between five eye movement features and SA latencies. Two eye movement features related to fixation duration and saccade duration demonstrated very strong correlations ($r \approx - 0.8$ and $r \approx 0.85$). Our findings can be extended to estimate the real-time likelihood of the human experiencing SA latency.

A challenge for search technologies is to support scientific literature surveys that present overviews of the reported numerical values documented for specific physical properties. We present SciHarvester, a system tailored to address this problem for agronomic science. It provides an interface to search PubAg documents, allowing complex queries involving restrictions on numerical values. SciHarvester identifies relevant documents and generates overview of reported parameter values. The system allows interrogation of the results to explain the system's performance. Our evaluations demonstrate the promise of incorporating information extraction techniques with the use of neural scoring mechanisms.

Businesses face a range of cyber risks, both external threats and internal vulnerabilities that continue to evolve over time. As cyber attacks continue to increase in complexity and sophistication, more organisations will experience them. For this reason, it is important that organisations seek timely consultancy from cyber professionals so that they can respond to and recover from cyber attacks as quickly as possible. However, huge surges in cyber attacks have long left cyber professionals short of what is required to cover the security needs. This problem is getting worse when an increasing number of people choose to work from home during the pandemic because this situation usually yields extra communication cost.

Tracking the evolution of clusters in social media streams is becoming increasingly important for many applications, such as early detection and monitoring of natural disasters or pandemics. In contrast to clustering on a static set of data, streaming data clustering does not have a global view of the complete data. The local (or partial) view in a high-speed stream makes clustering a challenging task. In this paper, we propose a novel density peak based algorithm,TStream, for tracking the evolution of clusters and outliers in social media streams, via the evolutionary actions of cluster adjustment, emergence, disappearance, split, and merge.TStreamis based on a temporal decay model and text stream summarisation. The decay model captures the decreasing importance of textual documents over time. The stream summarisation compactly represents them with the help of cells (akamicro-clusters) in the memory. We also propose a novel efficient index calledshared dependency tree(akaSD-Tree) based on the ideas of density peak and shared dependency. It maintains the dynamic dependency relationships inTStreamand thereby improves the overall efficiency. We conduct extensive experiments on five real datasets.TStreamoutperforms the existing state-of-the-art solutions based onMStream,MStreamF,EDMStream,OSGM, andEStream, in terms of cluster mapping measure (CMM) by up to 17.8%, 18.6%, 6.9%, 16.4%, and 20.1%, respectively. It is also significantly more efficient thanMStream,MStreamF,OSGM, andEStream, in terms of response time and throughput.

Multi-turn response selection is a key issue in retrieval-based chatbots and has attracted considerable attention in the NLP (Natural Language processing) field. So far, researchers have developed many solutions that can select appropriate responses for multi-turn conversations. However, these works are still suffering from the semantic mismatch problem when responses and context share similar words with different meanings. In this article, we propose a novel chatbot model based on Semantic Awareness Matching, called SAM. SAM can capture both similarity and semantic features in the context by a two-layer matching network. Appropriate responses are selected according to the matching probability made through the aggregation of the two feature types. In the evaluation, we pick 4 widely used datasets and compare SAM’s performance to that of 12 other models. Experiment results show that SAM achieves substantial improvements, with up to 1.5% R 10 @1 on Ubuntu Dialogue Corpus V2, 0.5% R 10 @1 on Douban Conversation Corpus, and 1.3% R 10 @1 on E-commerce Corpus.

Machine learning (ML) techniques are increasingly important in cybersecurity, as they can quickly analyse and identify different types of threats from millions of events. In spite of the increasing number of possible applications of ML, successful adoption of ML models in cybersecurity still highly relies on the explainability of those models that are used for making predictions. Explanations that support ML model outputs are crucial in cybersecurity-oriented ML applications because people need to get more information from the model than just binary output for analysis. The explainable models help ML developers solve the “trust” problem for a security application prediction in a faithful way: validating model behaviours, diagnosing misclassifications and sometimes automatically patching errors in the target models. Therefore, explainable ML for cybersecurity has become a necessary and important research branch. In this paper, we present the topic of explainable ML in cybersecurity through two general types of explanations: (1) ante hoc explanation, and (2) post hoc explanation, with their methodologies. We systematically review and categorise the state-of-the-art research, and provide comparative studies to help researchers find the optimal solutions to specific problems. We further list open issues in this field to facilitate future studies. This survey will benefit diverse groups of readers from both academia and industries, who want to effectively use ML to solve cybersecurity challenges.

Computer users are generally faced with difficulties in making correct security decisions. While an increasingly fewer number of people are trying or willing to take formal security training, online sources including news, security blogs, and websites are continuously making security knowledge more accessible. Analysis of cybersecurity texts from this grey literature can provide insights into the trending topics and identify current security issues as well as how cyber attacks evolve over time. These in turn can support researchers and practitioners in predicting and preparing for these attacks. Comparing different sources may facilitate the learning process for normal users by creating the patterns of the security knowledge gained from different sources. Prior studies neither systematically analysed the wide range of digital sources nor provided any standardisation in analysing the trending topics from recent security texts. Moreover, existing topic modelling methods are not capable of identifying the cybersecurity concepts completely and the generated topics considerably overlap. To address this issue, we propose a semi-automated classification method to generate comprehensive security categories to analyse trending topics. We further compare the identified 16 security categories across different sources based on their popularity and impact. We have revealed several surprising findings as follows: (1) The impact reflected from cybersecurity texts strongly correlates with the monetary loss caused by cybercrimes, (2) security blogs have produced the context of cybersecurity most intensively, and (3) websites deliver security information without caring about timeliness much.

Xuelin Situ, Ingrid Zukerman, Cecile Paris, Sameen Maruf, Gholamreza Haffari. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Lifelong Learning (LL) black-box models are dynamic in that they keep learning from new tasks and constantly update their parameters.Owing to the need to utilize information from previously seen tasks, and capture commonalities in potentially diverse data, it is hard for automatic explanation methods to explain the outcomes of these models.In addition, existing explanation methods, e.g., LIME (Ribeiro et al., 2016), which are computationally expensive when explaining a static black-box model, are even more inefficient in the LL setting.In this paper, we propose a novel Lifelong Explanation (LLE) approach that continuously trains a student explainer under the supervision of a teacher -an arbitrary explanation algorithm -on different tasks undertaken in LL.We also leverage the Experience Replay (ER) mechanism to prevent catastrophic forgetting in the student explainer.Our experiments comparing LLE to three baselines on text classification tasks show that LLE can enhance the stability of the explanations for all seen tasks and maintain the same level of faithfulness to the black-box model as the teacher, while being up to 10 2 times faster at test time.Our ablation study shows that the ER mechanism in our LLE approach enhances the learning capabilities of the student explainer.Our code is available at https://github.com/situsnow/LLE.

‘Watch The Flu’ is a tool that monitors tweets posted in Australia for symptoms of influenza. The tool is a unique combination of two areas of artificial intelligence: natural language processing and time series monitoring, in order to assist public health surveillance. Using a real-time data pipeline, it deploys a web-based dashboard for visual analysis, and sends out emails to a set of users when an outbreak is detected. We expect that the tool will assist public health experts with their decision-making for disease outbreaks, by providing them insights from social media.

Unlike widely used Named Entity Recognition (NER) data sets in generic domains, biomedical NER data sets often contain mentions consisting of discontinuous spans.Conventional sequence tagging techniques encode Markov assumptions that are efficient but preclude recovery of these mentions.We propose a simple, effective transition-based model with generic neural encoding for discontinuous NER.Through extensive experiments on three biomedical data sets, we show that our model can effectively recognize discontinuous mentions without sacrificing the accuracy on continuous mentions.

There exists a gulf between the rhetoric in visualisation research about the significance of uncertainty and the inclusion of representations of uncertainty in visualisations used in practice. The graphical representation of uncertainty information has emerged as a problem of great importance in visualisation research. This contribution presents a survey of 286 uncertainty visualisation research publications. All publications are categorised with regard to publication type, publication venue, application domain, target user, and evaluation type. We present an interactive web-based browser that facilitates easy visual search and exploration of the publications included in the survey. We conclude that uncertainty visualisation is severely limited by the quality and scope of uncertainty data, by the limited confidence in the data, and by the perceptual and cognitive confusion that the graphical representation of the data can generate.

Users receive a multitude of security information in written articles, e.g., newspapers, security blogs, and training materials. However, prior research suggests that these delivery methods, including security awareness campaigns, mostly fail to increase people's knowledge about cyber threats. It seems that users find such information challenging to absorb and understand. Yet, to raise users' security awareness and understanding, it is essential to ensure the users comprehend the provided information so that they can apply the advice it contains in practice. We conducted a subjective study to measure the level of users' understanding of security texts. We find that 61% of the terms security experts used in their writings are hard for the public to understand, even for people with some IT backgrounds. We also observe that 88% of security texts have at least one such term. Moreover, we notice that existing dictionaries, including the online ones (e.g., Google Dictionary), cover no more than 35% of the terms found in security texts. To improve users' ability to understand security texts, we developed a framework to build a user-oriented security-centric dictionary from multiple sources. To evaluate the effectiveness of the dictionary, we developed a tool as a service to detect technical terms and explain their meanings to the user in pop-ups. The results of a subjective study to measure the tool's performance showed that it could increase users' ability to understand security articles by 30%.

We address the issue of having a limited number of annotations for stance classification in a new domain, by adapting out-of-domain classifiers with domain adaptation. Existing approaches often align different domains in a single, global feature space (or view), which may fail to fully capture the richness of the languages used for expressing stances, leading to reduced adaptability on stance data. In this paper, we identify two major types of stance expressions that are linguistically distinct, and we propose a tailored dual-view adaptation network (DAN) to adapt these expressions across domains. The proposed model first learns a separate view for domain transfer in each expression channel and then selects the best adapted parts of both views for optimal transfer. We find that the learned view features can be more easily aligned and more stance-discriminative in either or both views, leading to more transferable overall features after combining the views. Results from extensive experiments show that our method can enhance the state-of-the-art single-view methods in matching stance data across different domains, and that it consistently improves those methods on various adaptation tasks.

As communities represent similar opinions, similar functions, similar purposes, etc., community detection is an important and extremely useful tool in both scientific inquiry and data analytics. However, the classic methods of community detection, such as spectral clustering and statistical inference, are falling by the wayside as deep learning techniques demonstrate an increasing capacity to handle high-dimensional graph data with impressive performance. Thus, a survey of current progress in community detection through deep learning is timely. Structured into three broad research streams in this domain – deep neural networks, deep graph embedding, and graph neural networks, this article summarizes the contributions of the various frameworks, models, and algorithms in each stream along with the current challenges that remain unsolved and the future research opportunities yet to be explored.

Trust can be employed for finding reliable information in Online Social Networks (OSNs). Since users in OSNs may intentionally change their behavior over time (in some cases for deceiving other users), modeling (pair-wise) trust relations in such complex environment is a challenging task. However, most of the existing trust prediction approaches assume that trust relations are fixed over time and they fail to capture the dynamic behavior of users in OSNs. In this paper, we propose a dynamic deep trust prediction model. As the impact of incidental emotions on trust has been proven in psychology studies, in this paper, we also study this impact on our trust prediction approach. First, we propose a novel deep structure that incorporates users' emotions and their textual contents in OSNs. Second, we use embeddings to represent the users and their self-descriptions provided. Finally, considering different time windows, we dynamically predict pair-wise trust relations. To evaluate our approach, we collected a large twitter dataset. The evaluation results demonstrate the effectiveness of our approach compared to the state-of-the-art approaches.

Harnessing data from social media to monitor health events is a promising avenue for public health surveillance. A key step is the detection of reports of a disease (referred to as ‘health mention classification’) amongst tweets that mention disease words. Prior work shows that figurative usage of disease words may prove to be challenging for health mention classification. Since the experience of a disease is associated with a negative sentiment, we present a method that utilises sentiment information to improve health mention classification. Specifically, our classifier for health mention classification combines pre-trained contextual word representations with sentiment distributions of words in the tweet. For our experiments, we extend a benchmark dataset of tweets for health mention classification, adding over 14k manually annotated tweets across diseases. We also additionally annotate each tweet with a label that indicates if the disease words are used in a figurative sense. Our classifier outperforms current SOTA approaches in detecting both health-related and figurative tweets that mention disease words. We also show that tweets containing disease words are mentioned figuratively more often than in a health-related context, proving to be challenging for classifiers targeting health-related tweets.

BACKGROUND: Finding relevant literature is crucial for many biomedical research activities and in the practice of evidence-based medicine. Search engines such as PubMed provide a means to search and retrieve published literature, given a query. However, they are limited in how users can control the processing of queries and articles-or as we call them documents-by the search engine. To give this control to both biomedical researchers and computer scientists working in biomedical information retrieval, we introduce a public online tool for searching over biomedical literature. Our setup is guided by the NIST setup of the relevant TREC evaluation tasks in genomics, clinical decision support, and precision medicine. RESULTS: To provide benchmark results for some of the most common biomedical information retrieval strategies, such as querying MeSH subject headings with a specific weight or querying over the title of the articles only, we present our evaluations on public datasets. Our experiments report well-known information retrieval metrics such as precision at a cutoff of ranked documents. CONCLUSIONS: We introduce the A2A search and benchmarking tool which is publicly available for the researchers who want to explore different search strategies over published biomedical literature. We outline several query formulation strategies and present their evaluations with known human judgements for a large pool of topics, from genomics to precision medicine.

Benefiting from advances in machine vision and natural language processing techniques, current image captioning systems are able to generate detailed visual descriptions. For the most part, these descriptions represent an objective characterisation of the image, although some models do incorporate subjective aspects related to the observer’s view of the image, such as sentiment; current models, however, usually do not consider the emotional content of images during the caption generation process. This paper addresses this issue by proposing novel image captioning models which use facial expression features to generate image captions. The models generate image captions using long short-term memory networks applying facial features in addition to other visual features at different time steps. We compare a comprehensive collection of image captioning models with and without facial features using all standard evaluation metrics. The evaluation metrics indicate that applying facial features with an attention mechanism achieves the best performance, showing more expressive and more correlated image captions, on an image caption dataset extracted from the standard Flickr 30K dataset, consisting of around 11K images containing faces. An analysis of the generated captions finds that, perhaps unexpectedly, the improvement in caption quality appears to come not from the addition of adjectives linked to emotional aspects of the images, but from more variety in the actions described in the captions.

The overwhelming amount of information continuously flowing through the Twitter environment makes topic derivation essential. It indeed plays a valuable role in a variety of Twitter-based applications, including content recommendations, news summarization, market analysis, etc. Topic derivation methods are typically based on semantic features of tweet contents. Because tweets are short by nature, such methods suffer from data sparsity. To alleviate this problem, this paper proposes a topic derivation method that incorporates tweet text similarity and interactions measures. Besides the tweet contents, the approach takes into account several types of interactions amongst tweets: Tweets which mention the same people, replies and retweets. Topic derivation is done through a two-step matrix factorization process. We conducted a number of experiments on several Twitter datasets to reveal both the individual and integrated effects of the various features being considered. Our experimental results against TREC2014 and our self collected tweetMarch datasets demonstrate that the proposed method is able to provide more than 30 percent improvement compared to other advanced topic derivation methods.

Personal health mention detection deals with predicting whether or not a given sentence is a report of a health condition. Past work mentions errors in this prediction when symptom words, i.e., names of symptoms of interest, are used in a figurative sense. Therefore, we combine a state-of-the-art figurative usage detection with CNN-based personal health mention detection. To do so, we present two methods: a pipeline-based approach and a feature augmentation-based approach. The introduction of figurative usage detection results in an average improvement of 2.21% F-score of personal health mention detection, in the case of the feature augmentation-based approach. This paper demonstrates the promise of using figurative usage detection to improve personal health mention detection.

Named entity recognition (NER) is widely used in natural language processing applications and downstream tasks.However, most NER tools target flat annotation from popular datasets, eschewing the semantic information available in nested entity mentions.We describe NNE-a fine-grained, nested named entity dataset over the full Wall Street Journal portion of the Penn Treebank (PTB).Our annotation comprises 279,795 mentions of 114 entity types with up to 6 layers of nesting.We hope the public release of this large dataset for English newswire will encourage development of new techniques for nested NER.

We identify agreement and disagreement between utterances that express stances towards a topic of discussion.Existing methods focus mainly on conversational settings, where dialogic features are used for (dis)agreement inference.We extend this scope and seek to detect stance (dis)agreement in a broader setting, where independent stance-bearing utterances, which prevail in many stance corpora and real-world scenarios, are compared.To cope with such non-dialogic utterances, we find that the reasons uttered to back up a specific stance can help predict stance (dis)agreements.We propose a reason comparing network (RCN) to leverage reason information for stance comparison.Empirical results on a well-known stance corpus show that our method can discover useful reason information, enabling it to outperform several baselines in stance (dis)agreement detection.

With the increasing popularity of e-commerce, the number of product-related queries generated by customers is growing. Answering these queries manually in real time is infeasible, and so automatic question-answering systems can be immensely helpful. Product queries are, however, very different from open-domain questions: they tend to be product-specific and the answers they demand can be very subjective. Previous research suggests that reviews are a valuable resource for answering product queries, but a key challenge is the language mismatch between user queries and reviews. To address this, we propose two neural models that discover relevant reviews for answering product queries. We demonstrate that our best model produces strong performance, outperforming state-of-the-art systems by consistently finding the most relevant reviews for product queries.

Customer reviews are now increasingly available on Online Social Networks (OSNs) for a wide range of products and services. Trust in the review's author is a crucial basis for believing in the reliability of reviews generated on such networks. In this context, the main challenge is to predict the unknown trust relationship between two users. Existing trust prediction approaches fail to incorporate textual footprint of users. To address this challenge, we present a deep learning-based graph analytics model to predict trust relations in OSNs. We leverage and extend GraphSAGE, a method for computing node representations in an inductive manner, to develop a deep classifier. We present our experiment with datasets from review websites to train classifiers that predict trust relations between pairs of users, and highlight how our approach significantly improves the quality of predicted trust relations compared to the state-of-the-art approaches.

Android users are increasingly concerned with the privacy of their data and security of their devices. To improve the security awareness of users, recent automatic techniques produce security-centric descriptions by performing program analysis. However, the generated text does not always address users’ concerns as they are generally too technical to be understood by ordinary users. Moreover, different users have varied linguistic preferences that do not match the text. Motivated by this challenge, we develop an innovative scheme to help users avoid malware and privacy-breaching apps by generating security descriptions that explain the privacy and security related aspects of an Android app in clear and understandable terms. We implement a prototype system, PERSCRIPTION, to generate personalised security-centric descriptions that automatically learn users’ security concerns and linguistic preferences to produce user-oriented descriptions. We evaluate our scheme through experiments and user studies. The results clearly demonstrate the improvement on readability and users’ security awareness of PERSCRIPTION’s descriptions compared to existing description generators.

With the ubiquitous presence of smart phones and the availability of easy-to-use applications, there is an increase in the number of online services. A growing number of people now search for information and interact online. They expect to see services available and accessible online. To meet citizens’ expectations, governments have also increased their online presence. However, information and services are not the only reasons people go online. People also build their social circle online, seeking support and empathy, looking for someone with whom they can talk and who can understand their situation and worries. Online communities (and social networks in general) have been shown to have the potential to provide social and emotional peer-support. Our work aimed at determining whether online communities could be deployed in the public administration domain, in particular to support people receiving welfare payments, with similar benefits. We hypothesized that an online community could provide such support to disadvantaged citizens. Toward testing this hypothesis, after a user requirements analysis and some preparatory work, we designed and developed an online community for a specific target group of welfare recipients, as a collaboration between CSIRO and the Australian Department of Human Services. The community was deployed for one year. In this paper, we briefly explain our aims and the work that went into preparing for the community. We introduce the portal and the support it offered. We then report our observations and findings about both the informational and emotional support participants received, through an analysis of the comments posted in the community, and whether this support was perceived as welcome and useful.

Interactions among tweets, i.e., mentions, retweets, replies, are important factors contributing to the quality of topic derivation on Twitter. If applied correctly, the incorporation of tweet interactions can significantly improve the quality of topic derivation in comparison with approaches that are mainly based on the content similarity analysis. However, how interactions can be measured and integrated with content similarity for topic derivation remains a challenge. In previous work, the strength of tweet-to-tweet relationship has been computed by simply adding measures for content similarity, mentions, and reply-retweets. This simple linear addition does not accurately reflect the various impacts these factors have on tweet relationships. In order to address this issue, we propose a joint probability model that can effectively integrate the effects of the content similarity, mentions, and reply-retweets to measure the tweet relationship for the purpose of topic derivation. The proposed method is based on matrix factorization techniques, which enables a flexible implementation on a distributed system in an incremental manner. Experimental results show that the proposed model results in a significant improvement in the quality of topic derivation over existing methods.

BACKGROUND: Evidence-based medicine practice requires medical practitioners to rely on the best available evidence, in addition to their expertise, when making clinical decisions. The medical domain boasts a large amount of published medical research data, indexed in various medical databases such as MEDLINE. As the size of this data grows, practitioners increasingly face the problem of information overload, and past research has established the time-associated obstacles faced by evidence-based medicine practitioners. In this paper, we focus on the problem of automatic text summarisation to help practitioners quickly find query-focused information from relevant documents. METHODS: We utilise an annotated corpus that is specialised for the task of evidence-based summarisation of text. In contrast to past summarisation approaches, which mostly rely on surface level features to identify salient pieces of texts that form the summaries, our approach focuses on the use of corpus-based statistics, and domain-specific lexical knowledge for the identification of summary contents. We also apply a target-sentence-specific summarisation technique that reduces the problem of underfitting that persists in generic summarisation models. RESULTS: In automatic evaluations run over a large number of annotated summaries, our extractive summarisation technique statistically outperforms various baseline and benchmark summarisation models with a percentile rank of 96.8%. A manual evaluation shows that our extractive summarisation approach is capable of selecting content with high recall and precision, and may thus be used to generate bottom-line answers to practitioners' queries. CONCLUSIONS: Our research shows that the incorporation of specialised data and domain-specific knowledge can significantly improve text summarisation performance in the medical domain. Due to the vast amounts of medical text available, and the high growth of this form of data, we suspect that such summarisation techniques will address the time-related obstacles associated with evidence-based medicine.

Abstract Increasing interactions and engagements in social networks through monetary and material incentives is not always feasible. Some social networks, specifically those that are built on the basis of fairness, cannot incentivize members using tangible things and thus require an intangible way to do so. In such networks, a personalized recommender could provide an incentive for members to interact with other members in the community. Behavior‐based trust models that generally compute social trust values using the interactions of a member with other members in the community have proven to be good for this. These models, however, largely ignore the interactions of those members with whom a member has interacted, referred to as “friendship effects.” Results from social studies and behavioral science show that friends have a significant influence on the behavior of the members in the community. Following the famous Spanish proverb on friendship “Tell Me Your Friends and I Will Tell You Who You Are,” we extend our behavior‐based trust model by incorporating the “friendship effect” with the aim of improving the accuracy of the recommender system. In this article, we describe a trust propagation model based on associations that combines the behavior of both individual members and their friends. The propagation of trust in our model depends on three key factors: the density of interactions, the degree of separation, and the decay of friendship effect. We evaluate our model using a real data set and make observations on what happens in a social network with and without trust propagation to understand the expected impact of trust propagation on the ranking of the members in the recommended list. We present the model and the results of its evaluation. This work is in the context of moderated networks for which participation is by invitation only and in which members are anonymous and do not know each other outside the community. Copyright © 2014 John Wiley & Sons, Ltd.

Research data on predisposition to mental health problems, and the fluctuations and regulation of emotions, thoughts, and behaviors are traditionally collected through surveys, which cannot provide a real-time insight into the emotional state of individuals or communities. Large datasets such as World Health Organization (WHO) statistics are collected less than once per year, whereas social network platforms, such as Twitter, offer the opportunity for real-time analysis of expressed mood. Such patterns are valuable to the mental health research community, to help understand the periods and locations of greatest demand and unmet need. We describe the "We Feel" system for analyzing global and regional variations in emotional expression, and report the results of validation against known patterns of variation in mood. 2.73 ×10(9) emotional tweets were collected over a 12-week period, and automatically annotated for emotion, geographic location, and gender. Principal component analysis (PCA) of the data illustrated a dominant in-phase pattern across all emotions, modulated by antiphase patterns for "positive" and "negative" emotions. The first three principal components accounted for over 90% of the variation in the data. PCA was also used to remove the dominant diurnal and weekly variations allowing identification of significant events within the data, with z-scores showing expression of emotions over 80 standard deviations from the mean. We also correlate emotional expression with WHO data at a national level and although no correlations were observed for the burden of depression, the burden of anxiety and suicide rates appeared to correlate with expression of particular emotions.

A key challenge in online communities is that of keeping a community active and alive. All online communities work hard to keep their members through various initiatives, such as personalisation and recommendation technologies. In online communities aimed at supporting behavioural change, that is, in domains such as diet, lifestyle, or the environment, the main reason for participation is not to connect with real-world friends for sharing and communicating, but to meet and gain support from like-minded people in an online environment. Introducing personalisation and recommendation features in these networks is challenging, as traditional approaches leverage the densely populated friendship relations found in typical social networks, and these are not present in these new community types. We address this challenge by looking beyond the articulated friendships of a community for evidence of relationships. In particular, we look at the interactions of members of an online community with other members and resources. In this article, we present a social behaviour model and apply it to two types of recommendation systems, a people recommender and a content recommender system. We evaluate our systems using the interaction logs of an online diet and lifestyle community in which 5,000 Australians participated in a 12-week programme. Our results show that our social behaviour-based recommendation algorithms outperform baselines, friendship-based, and link-prediction algorithms.

Social media is an invaluable source of feedback not just about consumer products and services but also about the effectiveness of government services. Our aim is to help analysts identify how government services can be improved based on citizen-contributed feedback found in publicly available social media. We present ongoing research for a social media monitoring interactive prototype with federated search and text analysis functionality. The prototype, developed to fit the workflow of social media monitors in the government sector, collects, analyses, and provides overviews of social media content. It facilitates relevance judgements on specific social media posts to decide whether or not to engage online. Our user log analysis validates the original design requirements and indicates ongoing utility to our federated search approach.

Gamification, the idea of inserting game dynamics into portals or social networks, has recently evolved as an approach to encourage active participation in online communities. For an online community to start and proceed on to a sustainable operation, it is important that members are encouraged to contribute positively and frequently. We decided to introduce gamification in an online community that we designed and developed with the Australian Government's Department of Human Services to support welfare recipients transitioning from one payment to another. We first defined a formal model of gamification and a gamification design process. In instantiating our model to the online community, we realised that our context applied a number of constraints on the gamification elements that could be introduced. In this paper, we outline the design and implementation of a gamification model for online communities and its instantiation into our context, with its specific requirements. While we cannot comment on the success of gamification to drive user engagement in our context (for lack of the possibility of a controlled experiment), we found our implementation of badges-based gamification a helpful way to provide a useful abstraction on the life of the community, providing feedback enabling us to monitor and analyze the community. We thus show how feedback provided by such gamification data has a potential to be useful to community providers to better understand the community needs and addressing them appropriately to maintain a level of engagement in the community.

The articles in this special issue focus on the use of cloud computing for social computing applications. Currently, two complimentary Internet based research areas are emerging: social computing and cloud computing. On the one hand, social computing empowers individual users with relatively low technological sophistication to use the web to engage in social interactions, contribute their expertise and share their content, experiences and opinions. On the other hand, cloud computing offers everyone sophisticated computing infrastructures and resources as utilities, so that individual users with relatively low computing knowledge can have at their disposal a high performing computing infrastructure with little investment. Together, these two complementary technological advances form the backbone of our digitized world, when coupled with the rise of sensors,mobile devices and the internet of things. Of course, they also face significant challenges. These articles look at each area and describe how they complement each other. The Social Web has become an important means of communication for everyone: individuals, organizations, and governments all use it to disseminate and share information, offer opinions and engage in discussions.

Recommendations to connect like-minded people can result in increased engagement amongst members of online communities, thus playing an important role in their sustainability. We have developed a suite of algorithms for friend recommendations using a social trust model called STrust. In STrust, the

Online support groups have proven to be very effective, in the health domain in particular, to provide their members with social, emotional and moral support. Drawing from this success, we are exploring the use of an online community to build a support group for parents currently in receipt of income support but needing to transition to find a job. Complementing the financial and informational support offered through existing welfare transition programs, our work aims to help people improve their prospect of finding a job to become financially self-sufficient and develop a support network to help them in their transition back to work. In this paper, we present the design of a journey of reflection and collaboration to encourage reflective thinking and foster relationships between community members.

We present an approach for extracting 3-sentence evidence-based summaries relevant to clinical questions. We approach this task as one of query-focused, extractive, single-document summarisation using sentencespecific statistics for each target sentence. We incorporate simple statistics and domain knowledge and show that such an approach is effective for identifying informative sentences from medical abstracts. Our system is evaluated automatically using ROUGE, and we compare our results with several baselines. The ROUGE-L F-scores of our system outperform all baselines. In addition, our approach is computationally efficient, and, on a percentile rank measure, our system achieves a percentile rank of 97.3%.

Gamification has recently evolved as an approach to engage and encourage active participation of members in online communities. For an online community to start and proceed on to a sustainable operation, it is important that members are encouraged to contribute positively and frequently. This pa

Engagement serves as an important metrics in judging the success of any online social network. It is thus important to identify how specific mechanisms can help build engagement. In this work, we investigate how rewarding members for their honesty can offer better engagement and raise the level of cooperativeness in the society. We define a write-rate game to simulate activities in an online forum like setting and conduct experiments with it. Our results show that providing right reward for honest actions promotes engagement and cooperation in the society.

Citizen-focused documents in Public Administration devote considerable effort to the expression of conditions. These conditions are commonly expressed as statements of eligibility requirements for the programs being described, but they manifest themselves in other places as well, such as in feedback to readers in tailored informational brochures and as input fields on program application forms. This paper discusses how administrative conditions can be represented in a manner that supports both the eligibility reasoning required for the generation of citizen-tailored documents and also the automated generation of condition expressions in a variety of forms. The paper pays particular attention to the question of how a generation mechanism can allow authors to override the default forms of automated expression when necessary. The discussion is based on a prototype tailored delivery application whose knowledge base is implemented in OWL DL and whose output is constructed using Myriad, a platform for tailored document planning and formatting.

We propose a trust model for social networks with the aim of building trust communities that inspire members to share their experiences, feelings and opinions in an open and honest way without the fear of being judged. The unique feature of our model is that the trust value is derived from the social capital built in the social networks over a period of time. First, we introduce a framework for building trust communities using social trust. We then define an underlying social trust model, called STrust. Finally, we report the current state of the development and the analysis of the proposed trust model.

E-Health is defined by the World Health Organisation as 'the combined use of electronic communication and information technology in the health sector' (WHO http:// www.who.int/topics/ehealth

In the process of designing tailored information delivery systems based on natural language generation technology, linguistic resources need to be developed. One of such resources is a set of discourse strategies, the abstract structures that specify how to construct a coherent text to achieve a specific communicative goal. Discourse strategies, however, prove to be difficult to develop, in part due to their abstractness. In this paper, we describe a graphical tool which allows users to annotate documents to indicate its various elements and how they relate to each other. The annotations can then be used to generate a first draft of the discourse strategies automatically. The tool is presented in the context of the Myriad delivery platform which aims at facilitating the rapid development of tailored information delivery systems.

With the continuous growth of information and its high connectivity, it is hard to follow only the relevant links and remain focused. This is true on the web, which has become a major source of information to learn about a topic, and it is also true in a research environment, where the amount of scientific material available electronically is forever increasing, making reading the published literature a difficult task. While reading a document, people often encounter a promising link (or citation), which they decide to follow, only to discover after a quick browse that the document is not relevant to their current needs. Often, people open the linked document into a new tabbed window to which they return later. In such cases, users usually have many tabs opened. When they finally get to the linked documents to read them, they sometimes wonder why they opened these documents in the first place.

Abstract To work effectively in information-rich environments, knowledge workers must be able to distil the most appropriate information from the deluge of information available to them. This is difficult to do manually. Natural language engineers can support these workers by developing information delivery tools, but because of the wide variety of contexts in which information is acquired and delivered, these tools have tended to be domain-specific, ad hoc solutions that are hard to generalise. This paper discusses Myriad, a platform that generalises the integration of sets of resources to a variety of information delivery contexts. Myriad provides resources from natural language generation for discourse planning as well as a service-based architecture for data access. The nature of Myriad's resources is driven by engineering concerns. It focuses on resources that reason about and generate from coarse-grained units of information, likely to be provided by existing information sources, and it supports the integration of pipe-lined planning and template mechanisms. The platform is illustrated in the context of three information delivery applications and is evaluated with respect to its utility.

Engagement with computer games causes children and adolescent users to spend a substantial amount of time at sedentary game playing activity. We hypothesise that this engagement can be leveraged to motivate users to increase their amount of physical activity. In this paper, we present a novel approach for designing computer games, according to which the users' physical activity reinforces their game character. This way the users are seamlessly motivated to perform physical activity while maintaining their enjoyment of playing the game.

We are looking at new effective ways to engage families in weight management. Going beyond targeting individuals out of their family context with a single intervention, we investigate ways of delivering tailored health information as an information service, using the family as a major contextual factor. In particular, we explore interaction modes and processes of engagement with families that would be most effective to provide an information service that users will perceive as a real benefit to them.

In the context of innovative Airborne Early Warning and Control (AEW&C) platform capabilities, we are building an environment that can support the generation of information tailored to operators' tasks. The challenging issues here are to improve the methods for managing information delivery to the operators, and thus provide them with high-value information on their display whilst avoiding noise and clutter. To this end, we enhance the operator's graphical interface with information delivery mechanisms that support maintenance of situation awareness and improving efficiency. We do this by proactively delivering task-relevant information.

The challenge of automatically summarising Web pages and sites is a great one.However, currently there is no solution which offers an easy way to produce unbiased, coherent , and contentfull summaries of Web sites.In this work we suggest a new approach, which relies on the structure of hypertext and the way people describe information in it.As a proof-of-concept, we applied our approach to the problem of tailoring coherent snippets for search results.In this paper we describe the approach as it is implemented in our system, InCommonSense, and present results from a large scale evaluation of the snippets produced.We conclude by suggesting other applications that could make use of this technique.

This paper presents the results of an analysis we conducted to this end on a corpus of software instructions in French. 2 Methodology

A question‐answering program that provides access to a large amount of data will be most useful if it can tailor its answers to each individual user. In particular, a user's level of knowledge about the domain of discourse is an important factor in this tailoring if the answer provided is to be both informative and understandable to the user. In this research, we address the issue of how the user's domain knowledge, or the level of expertise, might affect an answer. We present TAILOR, a flexible computer system that takes into account this knowledge to provide an answer that is appropriate for users with varying levels of expertise (including novices and experts), without requiring an a priori set of user types. Un programme qui donne accès à un système de connaissances à de nombreux et divers utilisateurs serait plus utile s'il pouvait adapter ses réponses pour chaque utilisateur. En particulier, le savoir de l'utilisateur est un facteur important pour que la réponse soit compréhensible et informative. Dans cet article, nous étudions le rǒle de la connaissance de l'utilisateur pour former un texte qui soit approprié pour la personne qui le lira. Nous présentons TAILOR, un système de génération flexible qui est sensitif au niveau de connaissance de l'utilisateur. TAILOR peut générer des textes appropries pour des utilisateurs aux niveaux de connaissances très variés, sans avoir besoin ?un groupe de stéréotypes définis a priori.

Explanation is an interactive process requiring a dialogue between advice-giver and advice-seeker. In this paper, we argue that in order to participate in a dialogue with its users, a generation system must be capable of reasoning about its own utterances and therefore must maintain a rich representation of the responses it produces. We present a text planner that constructs a detailed text plan, containing the intentional, attentional, and rhetorical structures of the text it generates.

In this paper, we show that one benefit of FUG, the ability to state global constraints on choice separately from syntactic rules, is difficult in generation systems based on augmented context free grammars (e.g., Definite Clause Grammars). They require that such constraints be expressed locally as part of syntactic rules and therefore, duplicated in the grammar. Finally, we discuss a reimplementation of FUG that achieves the similar levels of efficiency as Rubinoff's adaptation of MUMBLE, a deterministic language generator.

It is widely recognized that a question-answerlng system should be able to tailor its answers to the user.One of the dimensions Mong which thus tailoring can occur is with respect to the level of knowledge of a user about a domain.In particular, responses should be different depending on whether they are addressed to ns/ve or expert users.To understand what those differences should be, we a~alyzed texts from adult and iunior encyclopedias.We found that two different strategies were used in describing complex physical obiects to juniors and adults.We show how these strategies have been implemented on a test database.