On considering the functional nature of bots in Online Knowledge Communities (OKCs), we propose a theoretical framework to explain the formation mechanism of human-bot collaborative relationships. Drawing on Social Exchange Theory (SET), we identify a critical gap between its assumptions and the contemporary dynamics of human-bot collaboration in OKCs. Through a two-way fixed effects model, we identify that the resource acquisition expectation of human members suppresses the formation of human-assisted collaborative relationships through knowledge resource complementarity patterns; conversely, the resource acquisition expectation of bots promotes the formation of both types of collaborative relationships: human-assisted and bot-assisted. Furthermore, we find that the collaborative behavioral pattern of knowledge homogeneity serves as a significant mediating role in the formation of both types of human-bot collaborative relationships, whereas skill homogeneity exerts its mediating effect specifically on bot-assisted collaborations. Our findings highlight the differences between the formation of OKC human-bot relationships and interpersonal relationships from the perspective of the functional nature of bots, which enriches previous research and offers practical insights for OKC management.