Abstract
Proposed herein is an Artificial Intelligence (AI)-powered system that can automatically calculate personal relevance scores for action items using an inverse diffusion algorithm that scores tasks contrastively. The algorithm may compute how well each action item matches a contextual profile of an authenticated user, such as a profile built from conversation history, completed tasks, role, department, technical expertise, or combinations thereof, relative to how well the action item matches contextual profiles of other participants, with weighting based on participant similarity through diffusion. By constructing contextual profiles for participants using internal conversation signals, external enterprise data, or both, and by applying diffusion-based crowd matching that accounts for team structure and role overlap, the system can identify tasks that are more uniquely relevant to a particular user rather than tasks that are ambiguously applicable to multiple team members. Unlike systems that rely primarily on explicit assignment detection, this contrastive ranking approach can provide context-aware prioritization that surfaces relevant tasks for appropriate users, including in group conversations involving participants with overlapping expertise.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Balow, Ed; Ergin, Hieu; Lanov, Dennis; and Schenkel, Ray, "PROBABILISTIC RELEVANCE SCORING OF COLLABORATIVE ACTION ITEMS USING MULTI-FACTOR CONVERSATIONAL CONTEXT, ROOM-TYPE-AWARE BASELINE ADJUSTMENT, AND TRANSPARENT REASONING", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10342