Abstract
This disclosure describes techniques to measure risk trajectories in user journeys involving online search tasks performed by a user using a search engine, chatbot, or another query answering engine. Search tasks include the user interacting with (e.g., selecting and viewing) search results, chatbot generated answers, and web pages linked to those search results. Based on metadata about user-submitted queries, the user search session is divided into user visit segments that include sensitive queries by the user relating to seeking assistance (“help seeking”) or seeking potentially detrimental content (“harm seeking”). Determination of risk categories for sensitive queries are made (e.g., by a human evaluator and/or automated system) and a risk trajectory for the user is determined over a user session based on determined risk valuations. The user session is categorized based on risk trajectory to determine potential of risk for harm seeking by the user. Described techniques can measure risk trajectories that include multiple interactions of a user journey and enable improvement in providing assistance to help-seeking and harm-seeking users. The discussion in this paper is the result of exploratory studies conducted to assess risks associated with user journeys - using mental health as a particular example.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Roy, Abhishek; Jin, Ellie; and Umbach, Rebecca, "Determining User Journey Risk Trajectories in Information Seeking Sessions", Technical Disclosure Commons, (July 27, 2023)
https://www.tdcommons.org/dpubs_series/6082