In the information era, both the tourists and destinations are greatly affected by the information over the internet. Referring to tourists, they usually record their needs, hobbies, experiences as well as social relationships through internet; besides, a tourist would make a decision to travel to a particular place largely due to someone else's microblog or online travel note. Referring to destinations, information platforms, such as online social media, are gradually becoming important channels for destination image projection and communication; on the other hand, based on the reception and utilization of multi-source information on the internet, the managers and practitioners of destinations would make or change their decisions referring to the operation and management of the destination, and accordingly change the status of destination in the real world. To a great extent, the arrival of the information age has endowed human-environment relationship with new connotations, characteristics and forms. A new pattern of interaction between tourists and destinations with the intervention of network information has been formed. To understand this new relationship, finding the ways to promote the integration and sustainable development of this relationship based on it, is gradually becoming an important and urgent new proposition given by tourism geography research. In view of this, this study proposes a novel framework of human-environment relationship in "tri-space". Tri-space consists of three different spaces called geographic space, humanistic space and information pace (or cyberspace). Among them, geographic space is the physical space carrying tourism resources, facilities and activities within a specific region; humanistic space is the space constructed by tourists themselves referring to a specific region, which can be viewed as an explicit or implicit space; information space is a virtual space constructed by mass network information related to the tourism of a specific region. The concept of tri-space combines human, place and information together into an organic whole, with each sub-space interacting with each other. Machine learning techniques, the core techniques of artificial intelligence (AI), can be used to construct the metrics and measures of tri-space, discover the interaction mechanism in tri-space, and further build the co-reaction patterns of tri-space. Tri-space and its interaction view extend the traditional 'flat' human-environment relationship into a 'cubic' one, and further disclose a link between virtual/actual tourists and destinations, which provides a novel, systematic view of research on human-environment interaction.