Image: Reasearch Focus Banner

Constructive exchanges

Image: Lu XiaoLu Xiao

Lu Xiao is an Associate Professor at the University of Western Ontario, London, Canada. She holds appointments in the programmes of Library & Information Science and Computer Science.

Her most recent work on rationales was published in the Journal of the Association for Information Science and Technology (JASIST). She also works on projects related to participatory design, informal learning, community informatics, and digital humanities.

Articles by Lu Xiao

Academic opinions of Wikipedia and Open Access publishing

Wikipedia for academic publishing: advantages and challenges

From Online Information Review

View Online Information Review Table of Contents.

Find out more about this subject in our Library Studies eJournals collection.

Exclusive blogs

The Importance of Collaboration - Howard Thomas

The importance of customer loyalty, trust and advocacy - John Bowen

Exploring the issues of gender and stereotyping in marketing - Victoria Crittenden

The evolving role of the librarian - David Shumaker

The future of teaching cases in the classroom - Gina Vega

Dr Lu Xiao investigates the roles of reflection and rationale in both online and face-to-face group interactions, designing information technology aids that assist debate, learning and ideation.

Could you describe the main aims of your research at the Human-Information Interaction Research Lab?

We are interested in designing and evaluating social-technical support to help people solve problems and/or learn. This is the objective of my research too.

What inspires your work?

My graduate study exposed me to various fields, including information science, computer science, education, psychology and sociology, and I was lucky to be involved in several research projects related to helping people to solve problems or learn.

I became very interested in rationale sharing in group activities in particular. Where they are articulating rationales, members need to engage in reflective thinking; so I am very interested in this process and how to promote it in group activities. Donald Schön's work on reflection, reflection in action, reflection on action, is also very influential for me.

Can you explain why 'rationale', in the context of your research, is key to understanding group ideation processes?

In my research, rationale refers to someone's justification of his or her ideas, suggestions, viewpoints or perspectives. If we look at Stephen Toulmin's argumentation structure, I would say some of the arguments in the argumentation structure are rationales. There is also research that looks at explanation in expert systems, and researchers in this field have considered mainly three types of explanation: how, what and why explanations. For me, rationale is the 'why', or justification, explanation. My research is also related to justificatory reasoning.

Does idea generation differ between small groups and much larger settings?

In small online group activities, I found that the sharing of rationales seemed to promote or influence other people's reflection skills, though sharing rationales has its up and downsides. It facilitates engagement in the associative thinking process. The downside is the 'group think' problem.

I think there are probably more similarities than differences when we compare idea generation tasks in small versus large groups. For example, in work I did on Amazon Mechanical Turk, I found that sharing rationales in large-scale settings improved the average idea quality rating. Sharing rationales may allow people to think more deeply about their own ideas or engage in associative thinking processes – but I don't have data to determine which, or maybe it's both. However, sharing rationales didn't produce the best quality ideas. That indicates the problem of 'group think' for thinking 'out of the box'. I also noticed this in small group activities: people complained that sharing rationales made it difficult for them to think of new ideas.

Why did you decide to analyse Wikipedia's Article for Deletion (AfD) Discussions?

Our Lab decided to analyse AfD Discussions because they are an example of online and easily-accessible deliberation. They also require people to provide opinions as well as supporting rationales, in accordance with Wikipedia policy.

We looked at the types of situations that may affect deliberation outcome (whether the article will be deleted or not), such as the type of article or deliberation condition (whether it is unanimous or not). We also looked at computational approaches to help us detect important rationales from this platform.

In addition, we developed algorithms to help us identify imperative rationales. I then saw that we could find ways to apply them to other deliberation contexts, to help detect imperative statements and also to identify similarities and differences of opinion in arguments.

I found it very interesting that, in AfD Discussions, people resolve conflicts effectively and efficiently and function well in terms of reaching consensus, despite using pseudonyms in an online environment. The environment really encourages rational and democratic deliberation – other deliberation contexts can learn from this. Also, if there are ways we can identify or detect comments on a specific rationale point, this might also help improve the quality of deliberation. In other words, what I found here was the importance of rationale.

Your more recent project focuses on human rights violation data. What does this entail?

We are working with a group of oral history researchers to develop tools to navigate audio or video interview recordings about human rights issues and then analyse them. We have created a prototype that utilizes machine learning and visualization techniques to help researchers interact with interview recordings in an oral history database.

Using the prototype, researchers can identify similar terms or targets shared among interviews. They can also annotate specific interviews and share the annotations with other researchers.

Image: Collective intelligenceCollective intelligence

The Human-Information Interaction Research Lab at the University of Western Ontario in Canada explores means of improving group learning and collaborative problem solving using computational approaches.

Classically, the quality of group efforts to produce new ideas, or achieve a breakthrough in solving a problem, is entirely dependent on the availability of information to all participants. However, according to Dr Lu Xiao of the University of Western Ontario, the primary factor lies in how each individual's reasoning is conveyed to all members of the group.

In 2010, Xiao created the Human-Information Interaction Research Lab (HIIRL), which concentrates on the design of computer-based analytical and visualization techniques for helping people in groups to solve problems or learn. In this context, the HIIRL recently established that sharing rationales in deliberations – giving an explanation of a decision, a proposition or an assertion – directly improves the quality of collaboration outcomes, however small or large the group. So Xiao's main focus is on computational tools that promote the articulation, sharing and management of rationales among groups; she also explores rationale-based approaches for knowledge management in collective activities and enhanced curatorship of disparate datasets.

Whether it is for ideation or deliberation, Xiao has found that a dedicated rationale space, maybe virtual, maybe not, improves group members' awareness of others' knowledge and intellectual contributions, helps idea generation by improving associative thinking and, when the other group members are suitably supportive, improves each member's reasoning skills. The requirement to articulate and share rationales also enables continuous quality monitoring and control: "We need to provide our rationales more explicitly," Xiao explains. "We want to multitask, be fast and efficient, but we need to stop and reflect. My work has shown that we should promote reflection in collective activities, even though this means pushing people to take their time." 

Informal learning

Most learning happens during reflection and outside of formal educational systems. Xiao has therefore investigated provision of structures to support informal learning within schools and also how community/researcher interactions affect the conduct of joint projects.

She is now looking into how 21st Century children may be better helped to learn mathematics through informal learning programmes. A pilot study has been carried out in Nanjing, China, and London, Canada, with faculty members from information science, education and educational technology from the University of Western Ontario and Nanjing Normal University. The goal is to establish a partnership between the school, parents and the community via a community workshop model, to teach mathematical concepts and techniques in a more enjoyable, accessible and relevant way. Children and parents are taught together, and are encouraged to engage in mathematics through hands-on exploration and using tools. The project also aims to lead to parent- and child-friendly computer technologies for mediating learning, and make use of social networking systems for collective and reflective informal learning activities, while adding to knowledge about the role of culture, such as international differences in parenting styles that affect parent-child interactions.

Wikipedia publishing

Quality is often raised as a concern about open-access publishing in sites such as Wikipedia and journals such as PLoS. How Wikipedia manages quality compared to open-access journals was a recent topic for Xiao's team that aimed at improving understanding of processes and outcomes.

As a matter of Wikipedia policy, formulating, articulating and sharing justifications is essential in debates between multiple contributors on whether an article is fit for purpose or should be deleted. Such deliberations are termed Article for Deletion (AfD) Discussions and are conducted online. The AfD process relies on the analysis of the validity and strength of rationales based on such factors as information sources and inclusion criteria. Observing more than 200 discussions that took place over three days, Xiao focused on article topic, type of rationale, degree of unanimity and final result, in the context of Wikipedia's peer review and conflict resolution mechanisms: "We found that Wikipedia's publishing model has distinct characteristics and advantages," says Xiao. "Wikipedia's democratic policy has a huge impact, and could be mirrored in other deliberation contexts."

The HIIRL developed a natural language processing algorithm to quickly extract imperative rationales from 84 days' worth of AfD Discussions; they also developed an algorithm based on text similarity and sentiment analysis that can distinguish rationales from different opinions within a single discussion.

Xiao then turned to the perspectives of academics and researchers on open-access publishing. The HIIRL administered questionnaires to 120 academics and researchers to explore their knowledge and perceptions of such services. More than 40 per cent of the respondents were over 45 years old; only 28 per cent were under 35. The majority were male, research-oriented, tenured academics of a rank higher than assistant professor. 77 per cent thought that research publications were important in performance appraisals, and most had little or no experience of Wikipedia.

The study found that gender was a factor. While most had little experience of open-access publishing in general, males were more likely to have edited, or have written an article, as a registered user on Wikipedia, whereas females were unlikely to have had any active involvement. While all acknowledged the benefit of publishing in Wikipedia of a large user base and readership, they queried the suitability of content for average readers. They were also concerned about how Wikipedia qualifies a user as a reviewer and that there was a conflict between original research and current Wikipedia policy. The majority did not feel comfortable about other researchers editing a paper in progress, but none complained about the Wikipedia peer review model, including the pull style of obtaining reviewers through volunteering, nor its flexible communication flow between reviewers and authors.

Wikipedia was generally identified as offering advantages over open-access journals: in cost reduction, timely review, post-publication corrections, and making articles available before validation. Major disadvantages included questionable stability, absence of integration with libraries and scholarly search engines, lower quality, lower credibility, lower academic acceptance and lower impact on academia. Interestingly, however, the study found that, while academic milieu and status within it had some influence on perceptions, familiarity was key: "People who have more experience with Wikipedia tend to have more positive opinions of it," notes Xiao.

Helping survivor communities

A further strand of Xiao's research aims to enhance extraction of information from audio and visual recordings of witness testimony to human rights violations. In the Curating Testimony Project, Xiao is integrating machine learning techniques and natural language processing algorithms with interactive visualization, to create tools for curatorship and analysis of these recordings. These tools will then enable researchers, survivor communities and other users, such as teachers, to better browse, probe and use the datasets.

The work is a component of the wider 'Digging into Human Rights Violations: Anaphora Resolution and Emergent Witnesses' project, and is being carried out in collaboration with the Centre of Oral History and Digital Storytelling at Concordia University, the Page Rwanda association of parents and friends affected by the 1994 genocide of Tutsi people in Rwanda, and the Latin American organization for fostering digital literacy and civic engagement, HablaCentro. Xiao's team has so far developed a prototype system, the Clock-based Keyphrase Map that can identify and navigate through similar topics among different testimonies and supports annotation and sharing: "My vision is to further develop tools to facilitate the data sharing, data analysis, and data management processes in oral history research," Xiao concludes.

Human-information interaction research lab

Research Interests

  • Large-scale ideation and deliberation
  • Rationale sharing and articulation
  • Rationale-based approaches for knowledge management in collective activities
  • Reflective thinking
  • Informal learning

Key collaborators

  • Dr Robert Mercer, University of Western Ontario, Canada
  • Dr Immaculate Namukasa, University of Western Ontario, Canada
  • Dr Steven High, Centre of Oral History and Digital Storytelling, Concordia University, Canada


  • InfoTrellis, Canada
  • Xerox Research Center Europe, France