Now showing 1 - 6 of 6
  • Publication
    An agent for effective negotiation dialogues
    A design is presented for a negotiating agent that can construct coherent joint plans with human or artificial agents. In negotiation there is always a trade-off between plan quality and dialogue length. In dynamic conditions and with human partners, length becomes critical. The approach to efficient negotiation is to use an acquaintance model that predicts which plans will be acceptable. The negotiation dialogue then consists of exchanges to construct the acquaintance model and exchanges of plan proposals.
      71
  • Publication
    A planner for plan construction plans for multi-agent plans
    A natural language dialogue planner is described that chooses dialogue moves to revise the beliefs of an agent. In particular those beliefs that refer to the mental state of another agent are revised. In a planning problem of repeated decisions, the future decision of the other agent is better predicted, and therefore the immediate plan decision of the first agent is a better one
      60
  • Publication
    Dynamic choice of robust strategies in dialogue management
    An important tradeoff in error-prone dialogue is between the cost of using more robust dialogue strategies and the cost of recovering from failed understanding without using them. A strategy has to be quantitatively planned for each dialogue state, since too robust a strategy might not have a worthwhile effect on the failure rate. A dialogue manager is described which chooses between strategies that have differing levels of robustness with a view to maximising the efficiency of the dialogue
      48
  • Publication
    Efficient Dialogue Using a Probabilistic Nested User Model
    We describe a set of dialogue simulation experiments, in which a probabilistic nested user model is employed in deciding between speech acts for a collaborative planning task, finding that a gain in utility can be obtained by using a probabilistic rather than a logical model. Given a set of ordinary dialogue plan rules, our system generates a gametree representation of the dialogue, using chance nodes to represent uncertain preconditions in the plan. Then, the game-tree is evaluated with respect to a given user model state.
      68
  • Publication
    Machine Learning for Adaptive Spoken Control in PDA Applications
    A machine learning approach to interpreting utterances in spoken interfaces is described, where evidence from the utterance and from the dialogue context is combined to estimate a probability distribution over interpretations. The algorithm for the utterance evidence uses nearest-neighbour classification on a set of training examples, while the contextual evidence is provided by dialogue act n-grams derived from dialogue corpora. Each algorithm can adapt by recording data from the user at hand. Experimental results for the utterance interpreter show that adaptation to a particular user’s training utterances significantly improves recognition accuracy over training on utterances from the general population.
      69
  • Publication
    Planning to replan in a multi-agent environment
    A multi-agent planner is described that accounts for the replanning occurring when one agent’s action is observed by another. A nested belief model is used to generate an expectation of the other agent’s response. Using the planner’s output, a dialogue system is being developed which decides whether uncertainties in the belief model should be resolved through dialogue before execution of the domain level plan
      62