Setting Up User Action Probabilities in User Simulations for Dialog System Development
Ai, Hua and Litman, Diane

Article Structure

Abstract

User simulations are shown to be useful in spoken dialog system development.

Introduction

User simulations are widely used in spoken dialog system development.

Related Work

Most current simulation models are probabilistic models in which the models simulate user actions based on dialog context features (Schatzmann et al., 2006).

System and User Simulations

In this section, we describe the dialog system, the human user corpus we collected with the system, and the user simulation we used.

Evaluation Measures

In this section, we introduce the evaluation measures for comparing the simulated corpora generated by different simulation models to the human user corpus.

Results

We let all user simulations interact with our dialog system, where each simulates 250 low learners and 250 high learners.

Conclusion and Future Work

Setting up user action probabilities in user simulation is a nontrivial task, especially when no training data or only a small amount of data is available.

Topics

probability models

Appears in 7 sentences as: probabilistic model (1) probabilistic models (2) probability model (1) probability models (3)
In Setting Up User Action Probabilities in User Simulations for Dialog System Development
  1. Since most current user simulations deploy probability models to mimic human user behaviors, how to set up user action probabilities in these models is a key problem to solve.
    Page 1, “Abstract”
  2. most of these current user simulation techniques use probabilistic models to generate user actions, how to set up the probabilities in the simulations is another important problem to solve.
    Page 1, “Introduction”
  3. For the trained user simulations, we examine two sets of probabilities trained from user corpora of different sizes, since the amount of training data will impact the quality of the trained probability models .
    Page 2, “Introduction”
  4. Most current simulation models are probabilistic models in which the models simulate user actions based on dialog context features (Schatzmann et al., 2006).
    Page 2, “Related Work”
  5. They first cluster dialog contexts based on selected features and then build conditional probability models for each cluster.
    Page 2, “Related Work”
  6. In our study, we build a conditional probability model which will be described in detail in Section 3.2.1.
    Page 2, “Related Work”
  7. However, since our simulation model is a probabilistic model , the model will take an action stochastically after the same tutor turn.
    Page 6, “Evaluation Measures”

See all papers in Proc. ACL 2009 that mention probability models.

See all papers in Proc. ACL that mention probability models.

Back to top.