Taha Hassan | Virginia Tech

Research Directions

I. Protected editorial contexts: AI trust and explainability in higher education

UMAP'21

Editorial processes are a critical guardrail for a trustworthy user experience of novel information systems. An editorial process represents a consensus, formal or informal, of domain stakeholders’ beliefs about ability, authority, utility, safety, and responsibility. In domains of higher education and religio-spiritual storytelling, these editorial processes are often protected or informally negotiated. More Info

In my work at UMAP 2021, for instance, we discover that faculty vary widely in their allocation of editorial tasks for educational recommendation. Some favor a role-preserving or conservative model, where students can view (and rate or comment on, in a subset of cases) suggested readings, but not create or remove them. Others lean towards a collaborative authoring or egalitarian model where students are actively involved in all or most authoring and feedback tasks. This distinction of editorial task allocations is linked to the trust faculty and teaching assistants place in students and recommendation agents. Role-preservers often believe in instructor prerogatives, burdens of algorithm supervision, and overall student disengagement. Collaborative authors often cite the needs for student feedback and content moderation.

Keywords: Explainable AI, editorial processes, trustworthiness in recommendation, transparency affordances

II. Platform analytics for institutional support and learner intervention

LDT'22 ITiCSE'20

Understanding the adoption and impact of learning management system (LMS) services is central to faculty, university administrators, and instructional designers in better designing and evaluating course content. However, it is challenging and expensive to (a) petition, store, combine, and analyze LMS data to identify meaningful use-contexts, and to (b) evaluate learning, design and support outcomes across departments. More Info

These, in turn, challenge IT leadership’s ability to consolidate evidence on outcomes and best practices. These also threaten faculty’s trust in new LMS tools, course redesign efforts, and professional development initiatives.

In my work at ITiCSE 2020, we address these challenges by developing “Depth-of-Use” (DOU), a first-principles measurement process for assessing user engagement with LMS tools. At Virginia Tech IT, DOU has been successfully applied towards (a) adoption analyses of Canvas LMS services, (b) course redesigns, and (c) pandemic-era allocation of institutional support.

Keywords: Learning management system, adoption, institutional support, learner intervention, instructional design, emergency remote teaching

III. Protected editorial contexts: religiospiritual storytelling in sacred spaces

CHI'24

I'm investigating how institutions of spiritual storytelling, philanthropy, and community-engagement emerge, and assert their relationships with social strata in rural South Asian communities. More specifically, I'm investigating issues of

(a) story emergence and evolution, community loss
(b) story management and editorial process negotiation (community values of ownership, narration and preservation)
(c) facilitating shrine economies: donation and infrastructure grant management More Info

Keywords: Rural computing, sociocultural HCI, postcolonial computing, editorial process, community values, sociocultural HCI

IV. Social computing: trust in digital communities

CHIPlay'22 WWW'19 WebSci'19 ASONAM'19

I examine social media signatures of how perceptions of competence, benevolence, and integrity affect technology use in communities like higher education (e.g., faculty-student trust signals on RateMyProfessors) and the outdoors (e.g., social mistrust in location-based gaming and hiking communities on Reddit). These beliefs can provide a nuanced look at how initial social relationships in online communities are preserved and disrupted by new information systems. More Info

Keywords: Social computing, online communities, higher education, location-based gaming, long-distance hiking

V. Gamified psychometric assessments for self-evaluation of cognitive biases

I’m helping develop gamified self-assessments of key information processing biases, including herd instinct, framing, mental accounting, probability neglect, and present bias. These biases are known to affect human decision-making in the investing, borrowing, and spending arenas. More Info

These assessments are part of a comprehensive mobile-friendly profiling service planned by Be-IQ for end-users. They rely on robust measurements of key cognitive biases, often tied to risk attitudes, individual propensity to trust, and trust in financial institutions. Research objectives include:

(a) recommending self-directed educational materials to individuals susceptible to one or more of these biases
(b)identifying how market conditions and global socioeconomic realities might reinforce or diminish these biases.

Keywords: cognitive bias, psychometric tests, investing, borrowing, spending