In this paper we describe the challenges involved in automating collaborative negotiations and propose solution idea that can reduce computational cost of deal identification. We consider negotiations over a matter that has multiple issues and each issue can take any one of the multiple possible values for that issue. Two challenges are discussed in detail. The first one is the difficulty of assigning utility value ( i.e. evaluating) each possible contract in large contract spaces. The second is the computational complexity of locating the optimal contract. Previous work proposed grouping contracts to avoid evaluating each contract and a mediated bidding based deal identification algorithm to reduce the search for the optimal contract. However, the computational cost of exhaustively matching bids from agents grows exponentially with the number of agents and bids. We propose a rule that can be used during grouping of contracts (defining constraints) that reduces the number of possible bids from agents and hence increases the number of agents that could participate in the negotiation. We assume that each constraint corresponds to one evaluation criterion. The rule states that when evaluating contracts by a criterion, only contracts that satisfied previous contracts are considered.