Topic: UX research & UX design
Microsoft Research Asia
3 Months (2012-2013)
Mentor: Furu Wei
My Role: UX Research and design
Methods: Functional flow diagrams, task analysis (user journey map), cognitive walkthrough, heuristic analysis (expert reviews)
For the next generation of search engines, question answering (QA) is a key technology. Light (it is called Engkoo Answers for now) is a QA research platform that has been created by the Natural Language Computing group since 2011 at Microsoft Research Asia.
This design process makes sure that each design decision is supported by findings from specific methods and evidence.
Different from many traditional search engines that give users a huge number of query results, Light pursues to offer users an exact answer. Answers are based on three major sources: web-based question answering (WQA), knowledgebase question answering (KBQA), and community-based question answering (CQA).
The main UX design purpose for project Light is to enable users to find the right answer efficiently with evidence back up. Beyond the purpose of suggesting an answer, throughout this UX research project I was seeking to answer how to design the QA platform in order to afford a sense of trustiness of the system.
From the user’s point of view, task analysis uncovers the process of how users approach the Light platform; where are the problems; and what are the design opportunities.
According to the findings from the function flow diagram and the user journey map, a sitemap, and a lo-fi mockup were created. The lo-fi mock-up was used for the cognitive walkthrough task.
I used the cognitive walkthrough approach to evaluate the usability of Light by taking the users’ points of view. I supposed that the user was using this platform for the first time and had a task of searching for the right answer to “who is the CEO of Microsoft?”.
Our research managers, developers, and user experience experts from IEG (Innovation Engineering Group) are involved in this review. The lo-fi mockup was used in the analysis.
The Hi-fi prototype had been made based on findings from the cognitive walkthrough and heuristic analysis.
This project was shown in the 2013 Microsoft Tech-Fest in Seattle.
Based on this work, the group hoped to build domain-specific chatbots with question answering technologies.
https://www.microsoft.com/en-us/research/group/natural-language-computing/