Vol.14 No.5&6 November 1, 2015

Editorial (pp361-362)
        Sven Casteleyn, Gustavo Rossi, and Marco Winckler
Patterns in Eyetracking Scanpaths and the Affecting Factors (pp363-385)
        Sukru Eraslan and Yeliz Yesilada
Web pages are typically decorated with different kinds of visual elements that help sighted people complete their tasks. Unfortunately, people accessing web pages in constrained environments, such as visually disabled and small screen device users, cannot benefit from them. In our previous work, we show that tracking the eye movements of sighted users provide good understanding of how people use these visual elements. We also show that reengineering web pages by using these visual elements can improve people's experience in constrainted environments. However, in order to reengineer web pages based on eyetracking, we first need to aggregate, analyse and understand how a group of people's eyetracking data can be combined to create a common scanpath (namely, eye movement sequence) in terms of visual elements. This paper presents an algorithm that aims to achieve this. This algorithm was developed iteratively and experimentally evaluated with an eyetracking study. This study shows that the proposed algorithm is able to identify patterns in eyetracking scanpaths and it can work well with different number of participants. We then extended our experiments to investigate the effects of the task, gender and familiarity factors on common scanpaths. The results suggest that these factors can cause some differences in common scanpaths. This study also suggests that this algorithm can be improved by considering different techniques for pre-processing the data, by addressing the drawbacks of using the hierarchical structure and by taking into account the underlying cognitive processes.

From TMR to Turtle: Predicting Result Relevance from Mouse Cursor Interactions in Web Search (pp386-413)
        Maximilian Speicher, Sebastian Nuck, Lars Wesemann, Andreas Both, and Martin Gaedke 
The prime aspect of quality for search-driven web applications is to provide users with the best possible results for a given query. Thus, it is necessary to predict the relevance of results a priori. Current solutions mostly engage clicks on results for respective predictions, but research has shown that it is highly beneficial to also consider additional features of user interaction. Nowadays, such interactions are produced in steadily growing amounts by internet users. Processing these amounts calls for streaming-based approaches and incrementally updatable relevance models. We present StreamMyRelevance! --- a novel streaming-based system for ensuring quality of ranking in search engines. Our approach provides a complete pipeline from collecting interactions in real-time to processing them incrementally on the server side. We conducted a large-scale evaluation with real-world data from the hotel search domain. Results show that our system yields predictions as good as those of competing state-of-the-art systems, but by design of the underlying framework at higher efficiency, robustness, and scalability. Additionally, our system has been transferred into a real-world industry context. A modified solution called Turtle has been integrated into a new search engine for general web search. To obtain high-quality judgments for learning relevance models, it has been augmented with a novel crowdsourcing tool.

Identifying Web Performance Degradations through Synthetic and Real-User Monitoring (pp414-442)
        Jurgen Cito, Devan Gotowka, Philipp Leitner, Ryan Pelette, Dritan Suljoti, and Schahram Dustdar
The large scale of the Internet has offered unique economic opportunities, that in turn introduce overwhelming challenges for development and operations to provide reliable and fast services in order to meet the high demands on the performance of online services. In this paper, we investigate how performance engineers can identify three different classes of externally-visible performance problems (global delays, partial delays, periodic delays) from concrete traces. We develop a simulation model based on a taxonomy of root causes in server performance degradation. Within an experimental setup, we obtain results through synthetic monitoring of a target Web service, and observe changes in Web performance over time through exploratory visual analysis and changepoint detection. We extend our analysis and apply our methods to real-user monitoring (RUM) data. In a use case study, we discuss how our underlying model can be applied to real performance data gathered from a multinational, high-traffic website in the financial sector. Finally, we interpret our findings and discuss various challenges and pitfalls.

Designing Complex Crowdsourcing Applications Covering Multiple Platforms and Tasks (pp443-473)
        Alessandro Bozzon, Marco Brambilla, Stefano Ceri, Andrea Mauri, and Riccardo Volonterio

A number of emerging crowd-based applications cover very different scenarios, including opinion mining, multimedia data annotation, localised information gathering, marketing campaigns, expert response gathering, and so on. In most of these scenarios, applications can be decomposed into tasks that collectively produce their results; tasks interactions give rise to arbitrarily complex workflows. In this paper we propose methods and tools for designing crowd-based workflows as interacting tasks. We describe the modelling concepts that are useful in this framework, including typical workflow patterns, whose function is to decompose a cognitively complex task into simple interacting tasks for cooperative solving. We then discuss how workflows and patterns are managed by CrowdSearcher, a system for designing, deploying and monitoring applications on top of crowd-based systems, including social networks and crowdsourcing platforms. Tasks performed by humans consist of simple operations which apply to homogeneous objects; the complexity of aggregating and interpreting task results is embodied within the framework. We show our approach at work on a validation scenario and we report quantitative findings, which highlight the effect of workflow design on the final results.

Bayesian Based Type Discrimination of Web Events (pp525-544)
        Qichen Ma, Xiangfeng Luo, Junyu Xuan, and Huimin Liu
There are a large number of web events emerging on the web and attracting people’s attention every day, and it is of great interest and significance to distinguish the different types of these web events in practice. For example, the distinguished emergent web events should be paid more attentions by the departments of the government to save lives and damages or by news websites to increase their hit-rates using limited resources. However, how to efficiently distinguish the types of web events remains a challenge issue due to the seldom efforts paid to this issue in the community. In this paper, we conduct a thorough consideration on this problem and then propose an innovative Bayesian-based model to distinguish the different types of web events. To be specific, all web events are firstly assumed within three types whose formal definitions are given by considering their properties. Aiming to sufficiently describe and distinguish three types web events, a set of specially designed features are then extracted from the volume and the content of web events. Finally, a Bayesian-based model is proposed based on the designed features. The experimental results demonstrate the capability of the proposed model to distinguish types of web events, and the comparisons with other state-of-the-art classifiers also show the efficiency of the proposed model.

Back to JWE Online Front Page