Papers by Christopher Mitchell
Learning and improving natural turn-taking behaviors for dialogue systems is a topic of growing i... more Learning and improving natural turn-taking behaviors for dialogue systems is a topic of growing importance. In task-oriented dialogue where the user can engage in task actions in parallel with dialogue, unrestricted turn taking may be particularly important for dialogue success. This paper presents a novel Markov Decision Process (MDP) representation of dialogue with unrestricted turn taking and a parallel task stream in order to automatically learn effective turn-taking policies for a tutorial dialogue system from a corpus. It also presents and evaluates an approach to automatically selecting features for an MDP state representation of this dialogue. The results suggest that the MDP formulation and the feature selection framework hold promise for learning effective turn-taking policies in taskoriented dialogue systems.
Designing dialogue systems that engage in rich tutorial dialogue has long been a goal of the inte... more Designing dialogue systems that engage in rich tutorial dialogue has long been a goal of the intelligent tutoring systems community. A key challenge for these systems is determining when to intervene during student problem solving. Although intervention strategies have historically been hand-authored, utilizing machine learning to automatically acquire corpus-based intervention policies that maximize student learning holds great promise. To this end, this paper presents a Markov Decision Process (MDP) framework to learn an intervention policy capturing the most effective tutor turn-taking behaviors in a task-oriented learning environment with textual dialogue. The model and its learned policy highlight important design considerations, including maintaining tutor engagement during student problem solving and avoiding multiple consecutive interventions.
Convergence is thought to be an important phenomenon in dialogue through which interlocutors adap... more Convergence is thought to be an important phenomenon in dialogue through which interlocutors adapt to each other. Yet, its mechanisms and relationship to dialogue outcomes are not fully understood. This paper explores convergence in textual task-oriented dialogue during a longitudinal study. The results suggest that over time, convergence between interlocutors increases with successive dialogues. Additionally, for the tutorial dialogue domain at hand, convergence metrics were found to be significant predictors of dialogue outcomes such as learning, mental effort, and emotional states including frustration, boredom, and confusion. The results suggest ways in which dialogue systems may leverage convergence to enhance their interactions with users.
One-on-one tutoring is significantly more effective than traditional classroom instruction. In re... more One-on-one tutoring is significantly more effective than traditional classroom instruction. In recent years, automated tutoring systems are approaching that level of effectiveness by engaging students in rich natural language dialogue that contributes to learning. A promising approach for further improving the effectiveness of tutorial dialogue systems is to model the differential effectiveness of tutorial strategies, identifying which dialogue moves or combinations of dialogue moves are associated with learning. It is also important to model the ways in which experienced tutors adapt to learner characteristics. This paper takes a corpusbased approach to these modeling tasks, presenting the results of a study in which task-oriented, textual tutorial dialogue was collected from remote one-on-one human tutoring sessions. The data reveal patterns of dialogue moves that are correlated with learning, and can directly inform the design of student-adaptive tutorial dialogue management systems. 2 450
Designing dialogue systems that engage in rich tutorial dialogue has long been a goal of the inte... more Designing dialogue systems that engage in rich tutorial dialogue has long been a goal of the intelligent tutoring systems community. A key challenge for these systems is determining when to intervene during student problem solving. Although intervention strategies have historically been hand-authored, utilizing machine learning to automatically acquire corpus-based intervention policies that maximize student learning holds great promise. To this end, this paper presents a Markov Decision Process (MDP) framework to learn when to intervene, capturing the most effective tutor turn-taking behaviors in a taskoriented learning environment with textual dialogue. This framework is developed as a part of the JavaTutor tutorial dialogue project and will contribute to data-driven development of a tutorial dialogue system for introductory computer science education.
Co-author by Christopher Mitchell
The task of narrative visualization has been the subject of increasing interest in recent years. ... more The task of narrative visualization has been the subject of increasing interest in recent years. Much like data visualization, narrative visualization offers users an informative and aesthetically pleasing perspective on "story data." Automatically creating visual representations of narratives poses significant computational challenges due to the complex affective and causal elements, among other things, that must be realized in visualizations. In addition, narratives that are composed by novice writers pose additional challenges due to the disfluencies stemming from ungrammatical text. In this paper, we introduce the NARRATIVE THEATRE, a narrative visualization system under development in our laboratory that generates narrative visualizations from middle school writers' text. The NARRATIVE THEATRE consists of a rich writing interface, a robust natural language processor, a narrative reasoner, and a storyboard generator. We discuss design issues bearing on narrative visualization, introduce the NARRATIVE THEATRE, and describe narrative corpora that have been collected to study narrative visualization. We conclude with a discussion of a narrative visualization research agenda.
Dialogue act modeling in task-oriented dialogue poses significant challenges. It is particularly ... more Dialogue act modeling in task-oriented dialogue poses significant challenges. It is particularly challenging for corpora consisting of two interleaved communication streams: a dialogue stream and a task stream. In such corpora, information can be conveyed implicitly by the task stream, yielding a dialogue stream with seemingly missing information. A promising approach leverages rich resources from both the dialog and the task streams, combining verbal and non-verbal features. This paper presents work on dialogue act modeling that leverages body posture, which may be indicative of particular dialogue acts. Combining three information sources (dialogue exchanges, task context, and users' posture), three types of machine learning frameworks were compared. The results indicate that some models better preserve the structure of task-oriented dialogue than others, and that automatically recognized postural features may help to disambiguate user dialogue moves.
Learning dialogue management models poses significant challenges. In a complex taskoriented domai... more Learning dialogue management models poses significant challenges. In a complex taskoriented domain in which information is exchanged via parallel, interleaved dialogue and task streams, effective dialogue management models should be able to make dialogue moves based on both the dialogue and the task context. This paper presents a data-driven approach to learning dialogue management models that determine when to make dialogue moves to assist users' task completion activities, as well as the type of dialogue move that should be selected for a given user interaction context. Combining features automatically extracted from the dialogue and the task, we compare two alternate modeling approaches. The results of an evaluation indicate the learned models are effective in predicting both the timing and the type of system dialogue moves.
Uploads
Papers by Christopher Mitchell
Co-author by Christopher Mitchell