Digital Commons Network^™

Learning Cross-Modal Embeddings With Adversarial Networks For Cooking Recipes And Food Images, Hao Wang, Doyen Sahoo, Chenghao Liu, Ee-Peng Lim, Steven C. H. Hoi

Research Collection School Of Computing and Information Systems

Food computing is playing an increasingly important role in human daily life, and has found tremendous applications in guiding human behavior towards smart food consumption and healthy lifestyle. An important task under the food-computing umbrella is retrieval, which is particularly helpful for health related applications, where we are interested in retrieving important information about food (e.g., ingredients, nutrition, etc.). In this paper, we investigate an open research task of cross-modal retrieval between cooking recipes and food images, and propose a novel framework Adversarial Cross-Modal Embedding (ACME) to resolve the cross-modal retrieval task in food domains. Specifically, the goal is to …

Proud To Be Thai: The Puzzling Absence Of Ethnicity-Based Political Cleavages In Northeastern Thailand, Jacob I. Ricks

Research Collection School of Social Sciences

Underneath the veneer of a homogenous state-approved Thai ethnicity,Thailand is home to a heterogeneous population. Only about one-thirdof Thailand’s inhabitants speak the national language as their mothertongue; multiple alternate ethnolinguistic groups comprise the remainderof the population, with the Lao in the northeast, often called Isan people,being the largest at 28 percent of the population. Ethnic divisions closelyalign with areas of political party strength: the Thai Rak Thai Party and itssubsequent incarnations have enjoyed strong support from Isan people andKhammuang speakers in the north while the Democrat Party dominatesamong the Thai- and Paktay-speaking people of the central plains and thesouth. Despite …

Readability Of 10-K Reports And Stock Price Crash Risk, Chansog (Francis) Kim, Ke Wang, Liandong Zhang

Research Collection School Of Accountancy

This study shows that less readable 10‐K reports are associated with higher stock price crash risk. The results are consistent with the argument that managers can successfully hide adverse information by writing complex financial reports, which leads to stock price crashes when the hidden bad news accumulates and reaches a tipping point. Cross‐sectional analyses show that the effect of financial reporting complexity on crash risk is more pronounced for firms with persistent negative earnings news or transitory positive earnings news, greater chief executive officer stock option incentives, or lower litigation risk. Finally, accrual manipulation appears to be positively related to …

Dynamic Fusion With Intra-And Inter-Modality Attention Flow For Visual Question Answering, Peng Gao, Zhengkai Jiang, Haoxuan You, Pan Lu, Steven C. H. Hoi, Xiaogang Wang, Hongsheng Li

Research Collection School Of Computing and Information Systems

Learning effective fusion of multi-modality features is at the heart of visual question answering. We propose a novel method of dynamically fusing multi-modal features with intra- and inter-modality information flow, which alternatively pass dynamic information between and across the visual and language modalities. It can robustly capture the high-level interactions between language and vision domains, thus significantly improves the performance of visual question answering. We also show that the proposed dynamic intra-modality attention flow conditioned on the other modality can dynamically modulate the intramodality attention of the target modality, which is vital for multimodality feature fusion. Experimental evaluations on the …

Importance Sampling Of Interval Markov Chains, Cyrille Jegourel, Jingyi Wang, Jun Sun

Research Collection School Of Computing and Information Systems

In real-world systems, rare events often characterize critical situations like the probability that a system fails within some time bound and they are used to model some potentially harmful scenarios in dependability of safety-critical systems. Probabilistic Model Checking has been used to verify dependability properties in various types of systems but is limited by the state space explosion problem. An alternative is the recourse to Statistical Model Checking (SMC) that relies on Monte Carlo simulations and provides estimates within predefined error and confidence bounds. However, rare properties require a large number of simulations before occurring at least once. To tackle …

Legal Area Classification: A Comparative Study Of Text Classifiers On Singapore Supreme Court Judgments, Jerrold Soh, How Khang Lim, Ian Ernst Chai

Research Collection Yong Pung How School Of Law

This paper conducts a comparative study on the performance of various machine learning(“ML”) approaches for classifying judgments into legal areas. Using a novel dataset of 6,227 Singapore Supreme Court judgments, we investigate how state-of-the-art NLP methods compare against traditional statistical models when applied to a legal corpus that comprised few but lengthy documents. All approaches tested, including topic model, word embedding, and language model-based classifiers, performed well with as little as a few hundred judgments. However, more work needs to be done to optimize state-of-the-art methods for the legal domain.

Digital Disruption: The New York Times Asia-Pacific Writing Competition 2018, New York Times

Student Publications

This is a yearly writing competition organised by International New York Times (NYT). This year’s topic is “Digital Disruption” and SMU’s social sciences student Victoria Ivory Birrell was the winner in the University category. See her essay:

Ignore the Fluff, Technology hasn’t failed us and here’s why

Recommendations From Sec’S Plain English Handbook: Perspectives From Behavioral Research, Clarence Goh

Research Collection School Of Accountancy

The SEC published the plain English handbook in 1998 with the objective of providing preparers of financial disclosures with a guide to using well-established techniques to create clearer and more informative disclosure documents. The handbook provides various recommendations to preparers of disclosure documents. In this study, I examine and systematically synthesize the extent behavioural research literature that investigates the impact of implementing these recommendations on investors. I do so by developing a framework to systematically examine relevant studies. Overall, the analysis provides important insights from a behavioural perspective into how and why investors react to various recommendations in the SEC’s …

Concluding Remarks, Lei Meng, Ah-Hwee Tan, Donald C. Wunsch

Research Collection School Of Computing and Information Systems

This chapter summarizes the major contributions in this book and discusses their possible positions and requirements in some future scenarios. Section 8.1 follows the book structure to revisit the key contributions of this book in both theories and applications. The developed algorithms, such as the VA-ARTs for hyperparameter adaptation and the GHF-ART for multimedia representation and fusion, and the four applications, such as clustering and retrieving socially enriched multimedia data, are concentrated using one paragraph and three paragraphs, respectively. In Sect. 8.2, the roles of the proposed ART-embodied algorithms in social media clustering tasks are highlighted, and their possible evolutions …

Neural Multimodal Belief Tracker With Adaptive Attention For Dialogue Systems, Zheng Zhang, Lizi Liao, Minlie Huang, Xiaoyan Zhu, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Multimodal dialogue systems are attracting increasing attention with a more natural and informative way for human-computer interaction. As one of its core components, the belief tracker estimates the user's goal at each step of the dialogue and provides a direct way to validate the ability of dialogue understanding. However, existing studies on belief trackers are largely limited to textual modality, which cannot be easily extended to capture the rich semantics in multimodal systems such as those with product images. For example, in fashion domain, the visual appearance of clothes play a crucial role in understanding the user's intention. In this …

Dumbing Down?: Trends In The Complexity Of Political Communication, Kenneth Benoit, Kevin Munger, Arthur Spirling

Research Collection School of Social Sciences

Commentators and some political scholars claim to have observed a “dumbing down” in the level of sophistication of political language, leading to anxiety over the quality of democratic deliberation, knowledge, policy design, and implementation. This work typically focuses on the president’s State of the Union addresses. Using quantitative indicators of textual complexity, we measure trends since 1790 in that and other key political corpora, including rulings of the Supreme Court, the Congressional Record, and presidential executive orders. To draw comparative lessons, we also study political texts from the United Kingdom, in the form of party broadcasts and manifestos. Not only …

Emerging App Issue Identification From User Feedback: Experience On Wechat, Cuiyun Gao, Wujie Zheng, Yuetang Deng, David Lo, Jichuan Zeng, Michael R. Lyu, Irwin King

Research Collection School Of Computing and Information Systems

It is vital for popular mobile apps with large numbers of users to release updates with rich features while keeping stable user experience. Timely and accurately locating emerging app issues can greatly help developers to maintain and update apps. User feedback (i.e., user reviews) is a crucial channel between app developers and users, delivering a stream of information about bugs and features that concern users. Methods to identify emerging issues based on user feedback have been proposed in the literature, however, their applicability in industry has not been explored. We apply the recent method IDEA to WeChat, a popular messenger …

Detect Rumors On Twitter By Promoting Information Campaigns With Generative Adversarial Learning, Jing Ma, Wei Gao, Kam-Fai Wong

Research Collection School Of Computing and Information Systems

Rumors can cause devastating consequences to individual and/or society. Analysis shows that widespread of rumors typically results from deliberately promoted information campaigns which aim to shape collective opinions on the concerned news events. In this paper, we attempt to fight such chaos with itself to make automatic rumor detection more robust and effective. Our idea is inspired by adversarial learning method originated from Generative Adversarial Networks (GAN). We propose a GAN-style approach, where a generator is designed to produce uncertain or conflicting voices, complicating the original conversational threads in order to pressurize the discriminator to learn stronger rumor indicative representations …

Constructed Response Formats And Their Effects On Minority-Majority Differences And Validity, Filip Lievens, Paul R. Sackett, Jeffrey Dahlke, Janneke Oostrom, Britt De Soete

Research Collection Lee Kong Chian School Of Business

The inflow of immigrants challenges organizations to consider alternative selection procedures that reduce potential minority (immigrants)-majority (natives) differences, while maintaining valid predictions of performance. To deal with this challenge, this paper proposes response format as a practically and theoretically relevant factor for situational judgment tests (SJTs). We examine a range of response format categories (from traditional multiple-choice formats to more innovative constructed response formats) and conceptually link these response formats to mechanisms underlying minority-majority differences. Two field experiments are conducted with SJTs. Study 1 (274 job seekers) contrasts minority-majority differences in scores on a multiple-choice versus a written constructed response …

Neural Multimodal Belief Tracker With Adaptive Attention For Dialogue Systems, Zheng Zhang, Lizi Liao, Minlie Huang, Xiaoyan Zhu, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Multimodal dialogue systems are attracting increasing attention with a more natural and informative way for human-computer interaction. As one of its core components, the belief tracker estimates the user's goal at each step of the dialogue and provides a direct way to validate the ability of dialogue understanding. However, existing studies on belief trackers are largely limited to textual modality, which cannot be easily extended to capture the rich semantics in multimodal systems such as those with product images. For example, in fashion domain, the visual appearance of clothes play a crucial role in understanding the user's intention. In this …

The Challenges Of Creating Engaging Content: Results From A Focus Group Study Of A Popular News Media Organization, Kholoud Khalil Aldous, Jisun An, Bernard J. Jansen

Research Collection School Of Computing and Information Systems

The process of content creation for distribution via social media platforms is not a trivial one for social media editors as the goal of creating both serious and engaging content is challenging, with no clear or differing guidelines or rules across and between platforms. For creators of serious content, such as news organizations, advertisers, or educational institutions, engagement has a deeper meaning beyond likes, shares, etc. that is aimed at the audience actually processing the underlying content associated with a social media post. In this research, we report findings from a group study that aimed to understand the process and …

Multimodal Review Generation For Recommender Systems, Quoc Tuan Truong, Hady W. Lauw

Research Collection School Of Computing and Information Systems

Key to recommender systems is learning user preferences, which are expressed through various modalities. In online reviews, for instance, this manifests in numerical rating, textual content, as well as visual images. In this work, we hypothesize that modelling these modalities jointly would result in a more holistic representation of a review towards more accurate recommendations. Therefore, we propose Multimodal Review Generation (MRG), a neural approach that simultaneously models a rating prediction component and a review text generation component. We hypothesize that the shared user and item representations would augment the rating prediction with richer information from review text, while sensitizing …

Neo-Daoism In Stanford Encyclopedia Of Philosophy, Alan Kam Leung Chan

Research Collection College of Integrative Studies

The term “Neo-Daoism” (or “Neo-Taoism”) seeks to capture the focal development in early medieval Chinese philosophy, roughly from the third to the sixth century C.E. Chinese sources generally identify this development as Xuanxue, or “Learning (xue) in the Profound (xuan).”

Property In Digital Coins, J.G. Allen

Research Collection Yong Pung How School Of Law

Digital coins have burst into mainstream awareness recently, mainly as a result of high-worth ‘Initial Coin Offerings’ (‘ICOs’). The most immediate question in the legal treatment of digital coins is whether they are properly seen as digital ‘commodities’, and/or as ‘securities’, and/or as units of ‘money’. But the conceptual underpinnings of these categories are not clear, nor is it clear how these categories relate to each other; no legal system currently deals adequately with incorporeal objects as objects of property law. This category includes not only digital coins but also some forms of conventional money and securities. Establishing a satisfactory …

Deepjit: An End-To-End Deep Learning Framework For Just-In-Time Defect Prediction, Thong Hoang, Hoa Khanh Dam, Yasutaka Kamei, David Lo, Naoyasu Ubayashi

Research Collection School Of Computing and Information Systems

Software quality assurance efforts often focus on identifying defective code. To find likely defective code early, change-level defect prediction – aka. Just-In-Time (JIT) defect prediction – has been proposed. JIT defect prediction models identify likely defective changes and they are trained using machine learning techniques with the assumption that historical changes are similar to future ones. Most existing JIT defect prediction approaches make use of manually engineered features. Unlike those approaches, in this paper, we propose an end-to-end deep learning framework, named DeepJIT, that automatically extracts features from commit messages and code changes and use them to identify defects. Experiments …

Question Answering With Textual Sequence Matching, Shuohang Wang

Dissertations and Theses Collection (Open Access)

Question answering (QA) is one of the most important applications in natural language processing. With the explosive text data from the Internet, intelligently getting answers of questions will help humans more efficiently collect useful information. My research in this thesis mainly focuses on solving question answering problem with textual sequence matching model which is to build vectorized representations for pairs of text sequences to enable better reasoning. And our thesis consists of three major parts.

In Part I, we propose two general models for building vectorized representations over a pair of sentences, which can be directly used to solve the …

Automatic Short Answer Grading Using Siamese Bidirectional Lstm Based Regression, Arya Prabhudesai, Nguyen Binh Duong Ta

Research Collection School Of Computing and Information Systems

Automatic student assessment plays an important role in education - it provides instant feedback to learners, and at the same time reduces tedious grading workload for instructors. In this paper, we investigate new machine learning techniques for automatic short answer grading (ASAG). The ASAG problem mainly involves assessing short, natural language responses to given questions automatically. While current research in the field has focused either on feature engineering or deep learning, we propose a new approach which combines the advantages of both. More specifically, we propose a Siamese Bidirectional LSTM Neural Network based Regressor in conjunction with handcrafted features for …

Recruitment And Selection, Filip Lievens, Derek Chapman

Research Collection Lee Kong Chian School Of Business

Few people question that recruitment and selection are key strategic domains in HRM. At the same time, recruitment and selection also have an image problem. First, recruitment and selection are often viewed as ‘old’ ingrained HRM domains. It seems like the traditional recruitment and selection procedures have been around for decades, which is at odds with the ever changing internal and external environment of organizations. Hence, practitioners often wonder whether there are any new research-based ways for recruiting and selecting personnel. Another image problem for recruitment and selection is that a false dichotomy is often created between so-called macro HR …

Measuring And Explaining Political Sophistication Through Textual Complexity, Kenneth Benoit, Kevin Munger, Arthur Spirling

Research Collection School of Social Sciences

Political scientists lack domain-specific measures for the purpose of measuring the sophistication of political communication. We systematically review the shortcomings of existing approaches, before developing a new and better method along with software tools to apply it. We use crowdsourcing to perform thousands of pairwise comparisons of text snippets and incorporate these results into a statistical model of sophistication. This includes previously excluded features such as parts of speech and a measure of word rarity derived from dynamic term frequencies in the Google Books data set. Our technique not only shows which features are appropriate to the political domain and …

Disjunctures Of Belonging And Belief: Christian Migrants And The Bordering Of Identity In Singapore, Lily Kong, Orlando Woods

Research Collection School of Social Sciences

Migration results in people that are different from one another living in closer physicalproximity. Proximity increases the chances of encountering difference, and can lead to boththe formation of new communities, and the strengthening of old. As a religion that claims tointegrate people into a trans-ethnic, trans-territorial faith community, Christianity encouragessuch encounters, whilst Christian groups play an important role in mediating them.Disjunctures of belonging and belief are the outcomes that arise from encounters withdifference within spaces of Christianity. Drawing on 100 interviews conducted betweenAugust 2017 and February 2018, this paper unravels these disjunctures through a focus on theinterplay between migrant and …

Thinking Algorithmically: From Cold War Computer Science To The Socialist Information Culture, Ksenia Tatarchenko

Research Collection College of Integrative Studies

Cold War competition shaped the process of computerization in both East and West during the second half of the twentieth century. This article combines insights from Science and Technology Studies, which brought the analysis of Cold War technopolitics beyond the context of the nation-state, with approaches from Critical Algorithm Studies, to question the algorithm's role in the global "computer revolution." It traces the algorithm's trajectory across several geographical, political, and discursive spaces to argue that its mutable cultural valences made the algorithm a universalizing attribute for representing human-machine interactions across the ideological divide. It shows that discourses about the human …

Discursive Power In Contemporary Media Systems: A Comparative Framework, Andreas Jungherr, Oliver Posegga, Jisun An

Research Collection School Of Computing and Information Systems

Contemporary media systems are in transition. The constellation of organizations, groups, and individuals contributing information to national and international news flows has changed as a result of the digital transformation. The 'hybrid media system' has proven to be one of the most instructive concepts addressing this change. Its focus on the mutually dependent interconnections between various types of media organizations, actors, and publics has inspired prolific research. Yet the concept can tempt researchers to sidestep systematic analyses of information flows and actors’ differing degrees of influence by treating media systems as a black box. To enable large-scale, empirical comparative studies …

A Network Analysis Of The Singapore Court Of Appeal's Citations To Precedent, Jerrold Tsin Howe Soh

Research Collection Yong Pung How School Of Law

This article presents findings from an empirical network analysis of citation practices in Singapore’s highest court. A network of all 987 reported Court of Appeal judgments handed down from 2000 to 2017 is constructed. Network centrality algorithms are used to rank judgments by centrality. Judgments on contract law, particularly on contractual interpretation and terms, emerge as the most central. Based on this, this article argues that more attention can be paid to interpretation per se as a legal skill. More generally, this article establishes a framework for applying network analysis to Singapore jurisprudence on a larger scale.

Preventing Accounting Fraud, Singapore Management University

Perspectives@SMU

Technology can help identify possible fraud but it might not overcome ‘capture theory’

Social Media Mining For Journalism, Arkaitz Zubiaga, Bahareh Heravi, Jisun An, Haewoon Kwak

Research Collection School Of Computing and Information Systems

The exponential growth of social media as a central communication practice, and its agility in capturing and announcing breaking news events more rapidly than traditional media, has changed the journalistic landscape: social media has been adopted as a significant source by professional journalists, and conversely, citizens are able to use social media as a form of direct reportage. This brings along new opportunities for newsrooms and journalists by providing new means for newsgathering through access to a wealth of citizen reportage and updates about current affairs, as well as an additional showcase for news dissemination.