| Popis: |
Popular messaging applications such as Slack, Discord, and Microsoft Teams have given rise to thousands of chatbots and in-app integrations to facilitate collaborations. We utilized this design framework to study how searchbots (i.e., chatbots that perform specific searches) can facilitate collaborative search. More specifically, we investigated a design space for searchbot's that can engage in a mixed-initiative interaction. In this dissertation, we present a Wizard of Oz (WoZ) study to investigate the implications of envisioning a conversational search system capable of engaging in mixed-initiative interactions to support collaborative search. The Wizard plays the role of a conversational search system that can search for information, send relevant web results, and message users. We investigated three Wizard conditions: bot\_info, bot\_dialog, and bot\_task, which differ in how the Wizard can intervene in a conversation. The intervention modes follow the mixed-initiative framework by ~\cite{chu1998evidential}, originally developed based on human conversations. Broadly, we report on three investigations: (1) participants perceptions of the searchbot across the different levels of inititive; (2) the Wizards' motivations to take the initiative; and (3) the Wizards' characterization of the appropriateness of their interventions. Our results suggest that participants' collaborations can enhance when the searchbot can take limited initiative and align with the participants' search strategy. Additionally, in the characterization of motivations and timings, the Wizard presented a wide array of themes to provide search assistance and promote collaborations. Finally, while the participants did not prefer the advanced capabilities of the searchbot, our characterization of their motivations and timing helps us understand the complex activities the searchbot can cater to support collaborations. |
