Abstract

Despite the fact that advertisements (ads) often include strongly emotional content, very little work has been devoted to affect recognition (AR) from ads. This work explicitly compares content-centric and user-centric ad AR methodologies, and evaluates the impact of enhanced AR on computational advertising via a user study. Specifically, we (1) compile an affective ad dataset capable of evoking coherent emotions across users; (2) explore the efficacy of content-centric convolutional neural network (CNN) features for encoding emotions, and show that CNN features outperform low-level emotion descriptors; (3) examine user-centered ad AR by analyzing Electroencephalogram (EEG) responses acquired from eleven viewers, and find that EEG signals encode emotional information better than content descriptors; (4) investigate the relationship between objective AR and subjective viewer experience while watching an ad-embedded online video stream based on a study involving 12 users. To our knowledge, this is the first work to (a) expressly compare user vs content-centered AR for ads, and (b) study the relationship between modeling of ad emotions and its impact on a real-life advertising application.

Abstract

Quality control of responses enhances sustainability and adoption of crowdsourcing. Expert and peer reviews, majority voting, machine learning, game theory, etc. are some of the practices for quality control in crowdsourcing. However, quality of crowdsourced responses is still a concern. We propose a quality control approach drawing inspiration from Requirements Engineering quality attributes - Completeness, Consistency and Correctness (3Cs). The 3Cs of a response are assessed and displayed as a CSQuaRE score based on coverage with reference to knowledge base (ontology), cohesiveness and contributor credibility in the domain. The knowledge base would evolve with continued extraction of instances of information and thus responses would be re-calibrated for relevance. The suggested approach would be demonstrated for Information Security related Question and Answers on a crowdsourcing platform. The evaluation of the approach would be based on comparison with existing quality control techniques and feedback from security experts.

Abstract

Recent work on Acoustic Unit Discovery (AUD) has led to the development of a non-parametric Bayesian phone-loop model where the prior over the probability of the phone-like units is assumed to be sampled from a Dirichlet Process (DP). In this work, we propose to improve this model by incorporating a Hierarchical Pitman-Yor based bigram Language Model on top of the units' transitions. This new model makes use of the phonotactic context information but assumes a fixed number of units. To remedy this limitation we first train a DP phone-loop model to infer the number of units, then, the bigram phone-loop is initialized from the DP phone-loop and trained until convergence of its parameters. Results show an absolute improvement of 1–2%on the Normalized Mutual Information (NMI) metric. Furthermore, we show that, combined with Multilingual Bottleneck (MBN) features the model yields a same or higher NMI as an English phone recogniser trained on TIMIT.

Abstract

With immense amount of data growing on web in Hindi, a text summariser would be helpful in summarising Government data, medical reports, news, and research articles. Hindi is the fourth most-spoken first language in the world. Hindi written in the Devanagari script is the official language of the Government of India. There is no public dataset for extractive summarisation available in Hindi and thus a dataset of 24253 News articles was extracted and the extractive summaries results were evaluated on various parameters with manual gold summaries of exactly 60 words each.

Abstract

Data structures is an important area of knowledge in computer science that is often difficult to understand for high school students. There has been a lot of research in computer science education around the subject of programming. Hence, it becomes imperative to explore the domain of data structures as well. With this work, we have looked into an active pedagogical approach for facilitating learning of data structures in school. This approach has been implemented by means of an interactive game. The game aims to teach arrays and linked lists through activities based on the principle of compare and contrast. The innovation in the game is the use of musical notes to teach operations in arrays and linked lists. The game was evaluated with 27 students across 8th, 9th and 10th grades using a questionnaire. The results of the evaluation showed that the group of students who learnt about arrays and linked lists through the game scored an average of 81.5% of the total marks while the other group scored an average of 72% of the total marks.

Abstract

Mobile devices are gaining prominence for a variety of activities. From casual browsing to complex tasks like online shopping and banking, people have started using mobiles for almost everything previously done on a desktop. As more and more tasks are shifting towards mobiles, it has become crucial to redesign desktop elements to fit and function consistently on mobile devices as well. One such element is data visualization. The process of designing visualizations for desktops is fairly matured; however, when it comes to mobile devices there are still a number of challenges faced. Most of these challenges revolve around the limited screen space and fewer conventional mediums of interactions offered by these devices. This paper presents a set of design patterns that help overcome some of the challenges faced when designing data visualizations. These patterns are designed for line charts on mobile devices. However, some of them might apply to other kinds of charts as well. There are four patterns presented in this paper, namely details slider, legend filter, selection brush, and responsive data grid. These patterns can be used either individually or together to help make visualizations more convenient to use on mobile devices.

Abstract

News reporting is increasingly becoming an exercise in attention catching headlines1, and raising expectations which trigger sharing on social media. The proposed paper examines how conventional journalism can stay viable despite the transformations brought by rapidly changing media forms. We examine the role of human curiosity as a major contributing factor in online content navigating mechanisms such as clickbaits, listicles, and sharing behaviour on social media. As news reporting shifts online, the navigation on the internet resembles the exploration of primitive humans beyond their immediate needs. The clickbaity headlines are becoming popular with editors and readers putting pressure on news reports to ’dumbdown’. This paper will explore how evolutionary anthropology, psychology, neuroscience and sociology provide insights into the working of clickbaits: ‘how exactly do they attract viewers and why?’ It is based on a study of news reports in three mainstream Indian newspapers over one year (2014) during which an important general election was held. It studies the headlines that garner attention and analyse their morphology. We hope to use our learnings on clickbaits to see how journalism which stays true to its mandate of informing citizens and providing news can attract views and readership and thus remain viable.

Abstract

Observational evidence in clinical practice is critical in healthcare and policy making. Researchers spend a lot of time searching for relevant published articles to write a systematic review of a topic. In this paper, we present our participation as the team of IIIT Hyderabad at Task2 Technologically Assisted Reviews in Empirical Medicine as an effort to automate this task and deliver relevant information in medical literature. We base our approach on query expansion according to relevance feedback. Query expansion is a standard technique in information retrieval tasks with growing use in medical literature [1, 2]. Articles returned from pubmed query performed during a systematic review are first indexed using lucene’s inverted index. The query is porcessed for term boosting, fuzzy search and used for scoring documents according to TF-IDF similarity. Relevance feedback is used to update the query and become more pragmatic.

Abstract

This paper analyzes, compares and looks at tradeoffs in different load distribution schemes for consistent hashing in DHTs(distributed hash tables). Different traffic patterns, including an adversarial pattern, were constructed to test the load distribution. A simulator was made for each load distribution scheme and the load on each node was recorded for each traffic pattern. It was shown that increasing the number of consistent hash rings doesn't change the load distribution characteristics of the DHT in simple traffic scenarios, but in the case of adversarial traffic, increasing the number of hash rings leads to an improvement of around 15% in the load distribution statistics

Abstract

Millimeter wave (mmWave) communication systems employ hybrid RF/baseband transceivers that divide the spatial signal processing into radio frequency and baseband domains in order to reduce the hardware complexity resulting from the use of large number of antennas elements. In this paper, we consider two designs for hybrid multi-user MIMO downlink that minimizes the sum mean square error (SMSE), based on perfect channel knowledge (non-robust design) and transceiver design considering the channel imperfections (robust design). In the robust design, the channel state information (CSI) is assumed to be perturbed by estimation error and follows Gaussian distribution with known error variance. In both the designs, reduced hardware complexity is achieved by analog-digital hybrid processing by using orthogonal matching pursuit (OMP)-based sparse signal processing. We evaluate the performance of both the proposed schemes based on various parameters and also compare it with conventional fully-digital system. We present numerical results over various dictionaries. The comparison results show that robust design is resilient to the presence of CSI errors. Furthermore, we also demonstrate the convergence of both the proposed algorithms to a limit even though global convergence is hard to prove due to non convex nature of overall optimization problem.