Digital Commons Network^™

Automated Theme Search In Ico Whitepapers, Chuanjie Fu, Andrew Koh, Paul Griffin

Research Collection School Of Computing and Information Systems

The authors explore how topic modeling can be used to automate the categorization of initial coin offerings (ICOs) into different topics (e.g., finance, media, information, professional services, health and social, natural resources) based solely on the content within the whitepapers. This tool has been developed by fitting a latent Dirichlet allocation (LDA) model to the text extracted from the ICO whitepapers. After evaluating the automated categorization of whitepapers using statistical and human judgment methods, it is determined that there is enough evidence to conclude that the LDA model appropriately categorizes the ICO whitepapers. The results from a two-population proportion test …

A Deep Learning Approach For Tweet Classification And Rescue Scheduling For Effective Disaster Management, Md. Yasin Kabir, Sanjay Kumar Madria

Computer Science Faculty Research & Creative Works

Every activity in disaster management demands accurate and up-todate information to allow a quick, easy, and cost-efective response to reduce the possible loss of lives and properties. It is a challenging and complex task to acquire information from diferent regions of a disaster-afected area in a timely fashion. The extensive spread and reach of social media and networks such as Twitter allow people to share information in real-time. However, gathering of valuable information requires a series of operations such as (1) processing each tweet for the text classiication, (2) possible location determination of people needing help based on tweets, and …

Aspect And Opinion Aware Abstractive Review Summarization With Reinforced Hard Typed Decoder, Yufei Tian, Jianfei Yu, Jing Jiang

Research Collection School Of Computing and Information Systems

In this paper, we study abstractive review summarization. Observing that review summaries often consist of aspect words, opinion words and context words, we propose a two-stage reinforcement learning approach, which first predicts the output word type from the three types, and then leverages the predicted word type to generate the final word distribution. Experimental results on two Amazon product review datasets demonstrate that our method can consistently outperform several strong baseline approaches based on ROUGE scores.

Semi-Supervised Entity Alignment Via Joint Knowledge Embedding Model And Cross-Graph Model, Chengjiang Li, Yixin Cao, Lei Hou, Jiaxin Shi, Juanzi Li, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Entity alignment aims at integrating complementary knowledge graphs (KGs) from different sources or languages, which may benefit many knowledge-driven applications. It is challenging due to the heterogeneity of KGs and limited seed alignments. In this paper, we propose a semi-supervised entity alignment method by joint Knowledge Embedding model and Cross-Graph model (KECG). It can make better use of seed alignments to propagate over the entire graphs with KG-based constraints. Specifically, as for the knowledge embedding model, we utilize TransE to implicitly complete two KGs towards consistency and learn relational constraints between entities. As for the cross-graph model, we extend Graph …

Collective Representation Learning On Spatiotemporal Heterogeneous Information Networks, Dakshak Keerthi Chandra, Pengyang Wang, Jennifer Leopold, Yanjie Fu

Computer Science Faculty Research & Creative Works

Representation learning is a technique that is used to capture the underlying latent features of complex data. Representation learning on networks has been widely implemented for learning network structure and embedding it in a low dimensional vector space. In recent years, network embedding using representation learning has attracted increasing attention, and many deep architectures have been widely proposed. However, existing network embedding techniques ignore the multi-class spatial and temporal relationships that crucially reflect the complex nature among vertices and links in spatiotemporal heterogeneous information networks(SHINs).

To address this problem, in this paper, we present two types of collective representation learning …

Assessing The Generalizability Of Code2vec Token Embeddings, Hong Jin Kang, Tegawende F. Bissyande, David Lo

Research Collection School Of Computing and Information Systems

Many Natural Language Processing (NLP) tasks, such as sentiment analysis or syntactic parsing, have benefited from the development of word embedding models. In particular, regardless of the training algorithms, the learned embeddings have often been shown to be generalizable to different NLP tasks. In contrast, despite recent momentum on word embeddings for source code, the literature lacks evidence of their generalizability beyond the example task they have been trained for. In this experience paper, we identify 3 potential downstream tasks, namely code comments generation, code authorship identification, and code clones detection, that source code token embedding models can be applied …

Predicting Audience Engagement Across Social Media Platforms In The News Domain, Kholoud Khalil Aldous, Jisun An, Bernard J. Jansen

Research Collection School Of Computing and Information Systems

We analyze cross-platform factors for posts on both single and multiple social media platforms for numerous news outlets to better predict audience engagement, precisely the number of likes and comments. We collect 676,779 social media posts from 53 news outlets during eight months on four social media platforms (Facebook, Instagram, Twitter, and YouTube), along with the associated comments (more than 31 million) and the number of likes (more than 840 million). We develop a framework for predicting the audience engagement based on both linguistic features of the post and social media platform factors. Among other findings, results show that content …

Vireo-Eurecom @ Trecvid 2019: Ad-Hoc Video Search (Avs), Phuong Anh Nguyen, Jiaxin Wu, Chong-Wah Ngo, Francis Danny, Benoit Huet

Research Collection School Of Computing and Information Systems

In this paper, we describe the systems developed for Ad-hoc Video Search (AVS) task at TRECVID 2019[1] and the achieved results.

Low-Resource Name Tagging Learned With Weakly Labeled Data, Yixin Cao, Zikun Hu, Tat-Seng Chua, Zhiyuan Liu, Heng Ji

Research Collection School Of Computing and Information Systems

Name tagging in low-resource languages or domains suffers from inadequate training data. Existing work heavily relies on additional information, while leaving those noisy annotations unexplored that extensively exist on the web. In this paper, we propose a novel neural model for name tagging solely based on weakly labeled (WL) data, so that it can be applied in any low-resource settings. To take the best advantage of all WL sentences, we split them into high-quality and noisy portions for two modules, respectively: (1) a classification module focusing on the large portion of noisy data can efficiently and robustly pretrain the tag …

Estimating Glycemic Impact Of Cooking Recipes Via Online Crowdsourcing And Machine Learning, Helena Lee, Palakorn Achananuparp, Yue Liu, Ee-Peng Lim, Lav R. Varshney

Research Collection School Of Computing and Information Systems

Consumption of diets with low glycemic impact is highly recommended for diabetics and pre-diabetics as it helps maintain their blood glucose levels. However, laboratory analysis of dietary glycemic potency is time-consuming and expensive. In this paper, we explore a data-driven approach utilizing online crowdsourcing and machine learning to estimate the glycemic impact of cooking recipes. We show that a commonly used healthiness metric may not always be effective in determining recipes suitable for diabetics, thus emphasizing the importance of the glycemic-impact estimation task. Our best classification model, trained on nutritional and crowdsourced data obtained from Amazon Mechanical Turk (AMT), can …

Classifying Fiction And Non-Fiction Works Using Machine Learning, Rachna Gupta '21

Student Publications & Research

The objective of this project was to create a program that can determine whether an unknown text is a work of fiction or non-fiction using machine learning. Various datasets of speeches, ebooks, poems, scientific papers, and texts from Project Gutenberg and the Wolfram Example Data were utilized to train and test a Markov Chain machine learning model. A microsite was deployed with the final product that returns a probability of fictionality based on input from the user with 95% accuracy.

Deep Neural Network Based Iterative Self-Taught Learning On Text Mining, Xiangwen Liu

Theses and Dissertations

[Motivation] Machine learning and deep learning models is the main methodology for discovering hidden knowledge of data. Data plays a significant role in the training process. However, lack of data especially labeled data is precious due to accessibility and high cost. How to use the abundant unlabeled data to enhance the downstream learning process on labeled data is an urgent task in machine learning and pattern recognition. [Methods] An STL (Self-taught learning) model is proposed in this dissertation, which can successfully learn features from downloaded unlabeled text and enhance the downstream tasks on labeled text. In addition, DLI-IT (Drug labeling …

Collaborating On Machine Reading: Training Algorithms To Read Complex Collections, Carrie M. Pirmann, Brian R. King, Bhagawat Acharya, Katherine M. Faull

Bucknell University Digital Scholarship Conference

Interdisciplinary collaboration between two faculty members in the humanities and computer science, a research librarian, and an undergraduate student has led to remarkable results in an ongoing international DH research project that has at its core 18th century manuscripts. The corpus stems from a vast collection of archival materials held by the Moravian Church in the UK, Germany, and the US. The number of pages to be transcribed, differences in handwriting styles, paper quality, and original language pose enormous problems for the feasibility of human transcription. This presentation will review the hypothesis, process, and findings of a summer research project …

Update Frequency And Background Corpus Selection In Dynamic Tf-Idf Models For First Story Detection, Fei Wang, Robert J. Ross, John D. Kelleher

Conference papers

First Story Detection (FSD) requires a system to detect the very first story that mentions an event from a stream of stories. Nearest neighbour-based models, using the traditional term vector document representations like TF-IDF, currently achieve the state of the art in FSD. Because of its online nature, a dynamic term vector model that is incrementally updated during the detection process is usually adopted for FSD instead of a static model. However, very little research has investigated the selection of hyper-parameters and the background corpora for a dynamic model. In this paper, we analyse how a dynamic term vector model …

Who, Where, And What To Wear?: Extracting Fashion Knowledge From Social Media, Yunshan Ma, Xun Yang, Lizi Liao, Yixin Cao, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Fashion knowledge helps people to dress properly and addresses not only physiological needs of users, but also the demands of social activities and conventions. It usually involves three mutually related aspects of: occasion, person and clothing. However, there are few works focusing on extracting such knowledge, which will greatly benefit many downstream applications, such as fashion recommendation. In this paper, we propose a novel method to automatically harvest fashion knowledge from social media. We unify three tasks of occasion, person and clothing discovery from multiple modalities of images, texts and metadata. For person detection and analysis, we use the off-the-shelf …

Comprehending Test Code: An Empirical Study, Chak Shun Yu, Christoph Treude, Maurício Aniche

Research Collection School Of Computing and Information Systems

Developers spend a large portion of their time and effort on comprehending source code. While many studies have investigated how developers approach these comprehension tasks and what factors influence their success, less is known about how developers comprehend test code specifically, despite the undisputed importance of testing. In this paper, we report on the results of an empirical study with 44 developers to understand which factors influence developers when comprehending Java test code. We measured three dependent variables: the total time spent reading a test suite, the ability to identify the overall purpose of a test suite, and the ability …

Topicsummary: A Tool For Analyzing Class Discussion Forums Using Topic Based Summarizations, Swapna Gottipati, Venky Shankararaman, Renjini Ramesh

Research Collection School Of Computing and Information Systems

This Innovative Practice full paper, describes the application of text mining techniques for extracting insights from a course based online discussion forum through generation of topic based summaries. Discussions, either in classroom or online provide opportunity for collaborative learning through exchange of ideas that leads to enhanced learning through active participation. Online discussions offer a number of benefits namely providing additional time to reflect and synthesize information before writing, providing a natural platform for students to voice their ideas without any one student dominating the conversation, and providing a record of the student’s thoughts. An online discussion forum provides a …

Sieve: Helping Developers Sift Wheat From Chaff Via Cross-Platform Analysis, Agus Sulistya, Gede A. A. P. Prana, David Lo, Christoph Treude

Research Collection School Of Computing and Information Systems

Software developers have benefited from various sources of knowledge such as forums, question-and-answer sites, and social media platforms to help them in various tasks. Extracting software-related knowledge from different platforms involves many challenges. In this paper, we propose an approach to improve the effectiveness of knowledge extraction tasks by performing cross-platform analysis. Our approach is based on transfer representation learning and word embedding, leveraging information extracted from a source platform which contains rich domain-related content. The information extracted is then used to solve tasks in another platform (considered as target platform) with less domain-related content. We first build a word …

Who, Where, And What To Wear?: Extracting Fashion Knowledge From Social Media, Yunshan Ma, Xun Yang, Lizi Liao, Yixin Cao, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Fashion knowledge helps people to dress properly and addresses not only physiological needs of users, but also the demands of social activities and conventions. It usually involves three mutually related aspects of: occasion, person and clothing. However, there are few works focusing on extracting such knowledge, which will greatly benefit many downstream applications, such as fashion recommendation. In this paper, we propose a novel method to automatically harvest fashion knowledge from social media. We unify three tasks of occasion, person and clothing discovery from multiple modalities of images, texts and metadata. For person detection and analysis, we use the off-the-shelf …

Fusion Of Multimodal Embeddings For Ad-Hoc Video Search, Danny Francis, Phuong Anh Nguyen, Benoit Huet, Chong-Wah Ngo

Research Collection School Of Computing and Information Systems

The challenge of Ad-Hoc Video Search (AVS) originates from free-form (i.e., no pre-defined vocabulary) and freestyle (i.e., natural language) query description. Bridging the semantic gap between AVS queries and videos becomes highly difficult as evidenced from the low retrieval accuracy of AVS benchmarking in TRECVID. In this paper, we study a new method to fuse multimodal embeddings which have been derived based on completely disjoint datasets. This method is tested on two datasets for two distinct tasks: on MSR-VTT for unique video retrieval and on V3C1 for multiple videos retrieval.

Designing Women: Essentializing Femininity In Ai Linguistics, Ellianie S. Vega

Student Publications

Since the eighties, feminists have considered technology a force capable of subverting sexism because of technology’s ability to produce unbiased logic. Most famously, Donna Haraway’s “A Cyborg Manifesto” posits that the cyborg has the inherent capability to transcend gender because of its removal from social construct and lack of loyalty to the natural world. But while humanoids and artificial intelligence have been imagined as inherently subversive to gender, current artificial intelligence perpetuates gender divides in labor and language as their programmers imbue them with traits considered “feminine.” A majority of 21st century AI and humanoids are programmed to fit female …

Who, Where, And What To Wear?: Extracting Fashion Knowledge From Social Media, Yunshan Ma, Xun Yang, Lizi Liao, Yixin Cao, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Fashion knowledge helps people to dress properly and addresses not only physiological needs of users, but also the demands of social activities and conventions. It usually involves three mutually related aspects of: occasion, person and clothing. However, there are few works focusing on extracting such knowledge, which will greatly benefit many downstream applications, such as fashion recommendation. In this paper, we propose a novel method to automatically harvest fashion knowledge from social media. We unify three tasks of occasion, person and clothing discovery from multiple modalities of images, texts and metadata. For person detection and analysis, we use the off-the-shelf …

Smartembed: A Tool For Clone And Bug Detection In Smart Contracts Through Structural Code Embedding, Zhipeng Gao, Magalle Hewa Vinoj Yasanga Jayasundara, Lingxiao Jiang, Xin Xia, David Lo, John C. Grundy

Research Collection School Of Computing and Information Systems

Ethereum has become a widely used platform to enable secure, Blockchain-based financial and business transactions. However, a major concern in Ethereum is the security of its smart contracts. Many identified bugs and vulnerabilities in smart contracts not only present challenges to maintenance of blockchain, but also lead to serious financial loses. There is a significant need to better assist developers in checking smart contracts and ensuring their reliability. In this paper, we propose a web service tool, named SMARTEMBED, which can help Solidity developers to find repetitive contract code and clone-related bugs in smart contracts. Our tool is based on …

Fractals As Basis For Design And Critique, John Charles Driscoll

Dissertations and Theses

The design profession is responding to the complex systems represented by architecture and planning by increasingly incorporating the power of computer technology into the design process. This represents a paradigm shift, and requires that designers rise to the challenge of both embracing modern technologies to perform increasingly sophisticated tasks without compromising their objective to create meaningful and environmentally sensitive architecture. This dissertation investigated computer-based fractal tools applied within a traditional architectural charette towards a design process with the potential to address the complex issues architects and planners face today. We developed and presented an algorithm that draws heavily from fractal …

Machine Learning-Based Models For Assessing Impacts Before, During And After Hurricane Events, Julie L. Harvey

Electronic Theses and Dissertations

Social media provides an abundant amount of real-time information that can be used before, during, and after extreme weather events. Government officials, emergency managers, and other decision makers can use social media data for decision-making, preparation, and assistance. Machine learning-based models can be used to analyze data collected from social media. Social media data and cloud cover temperature as physical sensor data was analyzed in this study using machine learning techniques. Data was collected from Twitter regarding Hurricane Florence from September 11, 2018 through September 20, 2018 and Hurricane Michael from October 1, 2018 through October 18, 2018. Natural language …

Why Does This Entity Matter? Support Passage Retrieval For Entity Retrieval, Shubham Chatterjee, Laura Dietz

Computer Science Faculty Research & Creative Works

Our goal is to complement an entity ranking with human-readable explanations of how those retrieved entities are connected to the information need. While related to the problem of support passage retrieval, in this paper, we explore two underutilized indicators of relevance: contextual entities and entity salience. The effectiveness of the indicators is studied within a supervised learning-to-rank framework on a dataset from TREC Complex Answer Retrieval. We find that salience is a useful indicator, but it is often not applicable. In contrast, although performance improvements are obtained by using contextual entities, using contextual words still outperforms contextual entities.

Capturing Dialogue State Variable Dependencies With An Energy-Based Neural Dialogue State Tracker, Anh Duong Trinh, Robert J. Ross, John D. Kelleher

Conference papers

Dialogue state tracking requires the population and maintenance of a multi-slot frame representation of the dialogue state. Frequently, dialogue state tracking systems assume independence between slot values within a frame. In this paper we argue that treating the prediction of each slot value as an independent prediction task may ignore important associations between the slot values, and, consequently, we argue that treating dialogue state tracking as a structured prediction problem can help to improve dialogue state tracking performance. To support this argument, the research presented in this paper is structured into three stages: (i) analyzing variable dependencies in dialogue data; …

A Test Of Structural Model For Fear Of Crime In Social Networking Sites, Seong-Sik Lee, Kyung-Shick Choi, Sinyong Choi, Elizabeth Englander

International Journal of Cybersecurity Intelligence & Cybercrime

This study constructed a structural model which consists of social demographic factors, experience of victimization, opportunity factors, and social context factors to explain the public’s fear of crime on social networking sites (SNS). The model is based on the risk interpretation model, which predicts that these factors influence users’ fear of crime victimization. Using data from 486 university students in South Korea, an empirically-tested model suggests that sex and age have direct and significant effects on fear of victimization, supporting the vulnerability hypothesis. Among opportunity factors, the level of personal information and the number of offending peers have significant effects …

Bigger Versus Similar: Selecting A Background Corpus For First Story Detection Based On Distributional Similarity, Fei Wang, Robert J. Ross, John D. Kelleher

Conference papers

The current state of the art for First Story Detection (FSD) are nearest neighbour-based models with traditional term vector representations; however, one challenge faced by FSD models is that the document representation is usually defined by the vocabulary and term frequency from a background corpus. Consequently, the ideal background corpus should arguably be both large-scale to ensure adequate term coverage, and similar to the target domain in terms of the language distribution. However, given these two factors cannot always be mutually satisfied, in this paper we examine whether the distributional similarity of common terms is more important than the scale …

Amazon Alexa + Linked Open Data: Theorizing Concerning Relationships Between (Surveillant) Smart-Home Voice Assistants And Linked Open Data, Michelle Nitto

Publications and Research

No abstract provided.