Auction-based markets of divisible resources have proliferated over recent years. One fundamental problem facing every market mediator is how to achieve market efficiency for optimal social welfare, especially when a limited number of agents forms a monopolistic or oligopolistic market, because each agent’s selfish strategic behavior may lead to serious degradation in efficiency. In general, it is difficult for a market mediator to achieve efficiency since agents’ preferences are hidden information that they are unwilling to reveal due to security and privacy concerns. Therefore, the design of auction mechanisms should align the selfish behavior of agents with the altruistic objective of social welfare and allow the mediator to elicit necessary private information during the auction process.In this paper, we consider a market of divisible resource consisting of agents on both sides of demand and supply. We design an adaptive auction framework for a market mediator to achieve efficient resource allocation and acquisition. Our novel design generalizes demand/supply function bidding mechanisms by introducing price differentiation via tunable parameters. We design algorithms that enable the mediator and agents to jointly run the market in an adaptive fashion: the mediator sends market signals to agents; each agent submits her bid based on the signals in a distributed manner; the mediator adjusts tunable parameters based on bids and update market signals. We also design an adaptive algorithm to dynamically determine the optimal amount of resource that needs to be transacted so as to maximize social welfare, if not known a priori. By utilizing our market mechanisms, the market mediator will be able to reach an efficient market outcome under Nash equilibrium.