����ֱ��

Your phone or laptop screen is flickering, and your browser’s fonts are all wonky. The results of your search for a cheap flight to San Francisco at the right time or a funky room at a bargain rate are taking what seems like forever. It’s annoying, frustrating, time consuming. And it interferes with your ability to do what you need or want to do online, whether it’s booking a hotel or a flight, comparing options to buy a new car, or getting the vital stats that you need for a design or a sales presentation at work.

Software engineer Ying (Jenny) Zou is an expert in triaging and fixing the bugs that drive hundreds of millions of users crazy each moment of every day when using web browsers like Mozilla Firefox and Google Chrome.

“All these web services have very large user bases. We help developers fix bugs faster by using data mining and machine learning to analyze the history of the previously reported bugs by developers and users,” says Dr. Zou, a professor in the Department of Electrical and Computer Engineering (ECE) and a Canada Research Chair in Software Evolution.

Dr. Zou and her research team have developed new triaging techniques for analyzing and prioritizing crash types by frequency, severity, and repair difficulty. These clustering techniques help Mozilla Firefox developers make better decisions about which crash types should be fixed and when, speeding up and improving the process of repairing bugs in existing and new releases of complex software systems.

“Firefox receives millions of crash reports every day. The Mozilla developers and maintainers can leverage our work to speed up the triaging and eventual repair of bugs impacting millions of users worldwide,” says Dr. Zou, who is cross-appointed to the School of Computing at Queen’s.

Her current research with her PhD student, Mariam El Mezouar, analyzes social media data to identify which bugs users complain about the most.

“We helped Mozilla identify bugs that are more likely to be severe, if they are reported by a large number of tweets,” says Dr. Zou.

More broadly, Dr. Zou’s research program supports the evolution of service-oriented architecture (SOA) applications – including finance, e-commerce, and healthcare applications – by ensuring the delivery of reliable services with enhanced user experiences.

“When SOA applications fail, the repercussions are huge, with major impacts on our daily lives and on the success of companies that use the applications. We provide techniques and tools to ensure the SOA systems remain healthy and agile in response to the large user base and rapidly changing requirements,” says Dr. Zou, who is also a visiting scientist at the IBM Centers for Advanced Studies.

Dr. Zou has twice won IBM Faculty Awards, which are international prize that recognize the quality of one’s research program and its importance to industry, and was chosen as the 2014 IBM CAS Research Faculty Fellow of the Year. Her research at Queen’s in collaboration with IBM focuses on web services integration. The goal is to enhance online user experiences such as shopping. To buy a pair of boots, for example, a user may browse various e-commerce sites to compare products and consult user reviews. Essentially, users need to compose various services when buying boots, and face many options when choosing services. They also have to search for different kinds of information, such as price and user rating, which can affect their decisions about selecting the most appropriate services to reach their goals.

“It becomes tedious and cumbersome for users to discover and compose services that achieve their overall shopping goals,” says Dr. Zou.

Dr. Zou and her research team developed approaches to overcome several key challenges with existing web services. These challenges include limited and rigid options for users to specify their personalized preferences; conflicting preferences, like wanting a lower price for boots but a high user rating; and no prioritization of various preferences. Her solutions involved automatically learning user preferences to personalize how the user chooses services and provide tailored recommendations to help users select the most appropriate web services to meet their shopping goals.

“We worked with IBM to learn user preferences and rank different options for users using machine learning. This reduces the overloading of information for users and helps them come up with better choices faster,” she explains.

Dr. Zou is now collaborating with IBM on the development of future products that personalize and customize the integration of web services for the consumer.

“In the past, all service integrations were done by developers, giving users limited choices,” she says.

Her work aims at giving users more control and more decisions about the services that they need rather than letting developers give them just a few choices.

“We want to build a personal assistant to alert users to opportunities. Future products would personalize the daily activities for users based on their history and preferences,” says Dr. Zou.

For more information, visit .

This article was first published on the .