Decision Making Strategies for Finite-State Bi-automaton in Dialog Management

Ghigi, Fabrizio; Torres, M. Inés

doi:10.1007/978-3-319-19291-8_20

Cited by 5 publications

(9 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previous to the online learning phase, the relationship between the path length and the user-awareness rate is clear for the three policies: long and user-aware paths perform worse. This conclusion validates the hypothesis of [4] that path based policies perform worse than local policies overall due to user uncertainty. In addition, it is clear that the initial models are not fitted to the user, as the TC rate gets worse when the path length is increased.…”

Section: Path Based Policy Behavior Before User Adaptationsupporting

confidence: 87%

“…This augmented model is also known as Attributed PFSBA or A-PFSBA. The A-PFSBA paradigm separates the structural learning of the dialogue interaction and its exploitation, rendering flexibility when it comes to decision making [4]. An initial exploration of the potential and flexibility of A-PFSBA was done in [19], where the inclusion of attributes in the PFSBA structure showed to improve performance.…”

Section: Introductionmentioning

confidence: 99%

“…Although promising results were obtained, a local Maximum Likelihood exploitation policy which did not take user behavior into account was used and path based exploitation policies were not explored. Some policies that take user behavior into account were presented in [4], but they achieved lower results than local user agnostic policies. The uncertainty of user behavior was hypothesized as the main reason for these results.…”

Section: Introductionmentioning

confidence: 99%

“…In addition, the lack of evaluation after user adaptation renders it untested. Extending the online learning approach of the A-PFSBA framework presented in [19] to research in decision making strategies started in [4], this paper presents the following theoretical and experimental contributions:…”

Section: Introductionmentioning

confidence: 99%

“…Each policy is evaluated performing a grid search over the path length and the user-awareness ratio before and after the online learning phase. This experimentation is carried out in order to evaluate the hypothesis that policies that take user behavior into account perform worse due to user uncertainty [4].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

User-aware dialogue management policies over attributed bi-automata

Serras

Torres

Pozo

2018

Pattern Anal Applic

View full text Add to dashboard Cite

Designing dialogue policies that take user behavior into account is complicated due to user variability and behavioral uncertainty. Attributed Probabilistic Finite State Bi-Automata (A-PFSBA) have proven to be a promising framework to develop dialogue managers that capture the users' actions in its structure and adapt to them online, yet developing policies robust to high user uncertainty is still challenging. In this paper, the theoretical A-PFSBA dialogue management framework is augmented by formally defining the notation of exploitation policies over its structure. Under such definition, multiple path based policies are implemented, those that take into account external information and those which do not. These policies are evaluated on the Let's Go corpus, before and after an online learning process whose goal is to update the initial model through the interaction with end-users. In these experiments the impact of user uncertainty and the model structural learning is thoroughly analyzed.

show abstract

Section: Path Based Policy Behavior Before User Adaptationsupporting

confidence: 87%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations