Karl Schultz

A unified approach for schema matching, coreference and canonicalization

The automatic consolidation of database records from many heterogeneous sources into a single repository requires solving several information integration tasks. Although tasks such as coreference, schema matching, and canonicalization are closely related, they are most commonly studied in isolation. Systems that do tackle multiple integration problems traditionally solve each independently, allowing errors to propagate from one task to another. In this paper, we describe a discriminatively-trained model that reasons about schema matching, coreference, and canonicalization jointly. We evaluate our model on a real-world data set of people and demonstrate that simultaneously solving these tasks reduces errors over a cascaded or isolated approach. Our experiments show that a joint model is able to improve substantially over systems that either solve each task in isolation or with the conventional cascade. We demonstrate nearly a 50% error reduction for coreference and a 40% error reduction for schema matching.

Bi-directional Joint Inference for Entity Resolution and Segmentation Using Imperatively-Defined Factor Graphs

There has been growing interest in using joint inference across multiple subtasks as a mechanism for avoiding the cascading accumulation of errors in traditional pipelines. Several recent papers demonstrate joint inference between the segmentation of entity mentions and their de-duplication, however, they have various weaknesses: inference information flows only in one direction, the number of uncertain hypotheses is severely limited, or the subtasks are only loosely coupled. This paper presents a highly-coupled, bi-directional approach to joint inference based on efficient Markov chain Monte Carlo sampling in a relational conditional random field. The model is specified with our new probabilistic programming language that leverages imperative constructs to define factor graph structure and operation. Experimental results show that our approach provides a dramatic reduction in error while also running faster than the previous state-of-the-art system.

Advantages of Spoken Language Interaction in Dialogue-Based Intelligent Tutoring Systems

The ability to lead collaborative discussions and appropriately scaffold learning has been identified as one of the central advantages of human tutorial interaction [6]. In order to reproduce the effectiveness of human tutors, many developers of tutorial dialogue systems have taken the approach of identifying human tutorial tactics and then incorporating them into their systems. Equally important as understanding the tactics themselves is understanding how human tutors decide which tactics to use. We argue that these decisions are made based not only on student actions and the content of student utterances, but also on the meta-communicative information conveyed through spoken utterances (e.g. pauses, disfluencies, intonation). Since this information is less frequent or unavailable in typed input, tutorial dialogue systems with speech interfaces have the potential to be more effective than those without. This paper gives an overview of the Spoken Conversational Tutor (SCoT) that we have built and describes how we are beginning to make use of spoken language information in SCoT.

Contextualizing learning in a reflective conversational tutor

Contextualizing learning in an intelligent tutoring system is difficult for many reasons. Goals such as presenting material in an understandable manner, minimizing confusion and frustration, and helping the student reason about their actions all need to be balanced. Previous research has shown reflective discussions (with human tutors) occurring after problem-solving to be effective in helping students reason about their own actions (S. Katz et. al, 2003). However, leading a reflective discussion makes it difficult to present information in an understandable manner, and without contextualization, it is easy to create student confusion and frustration. This raises the question: how can intelligent tutoring systems effectively contextualize learning in a reflective discussion? In this paper, we describe the tutorial architecture of SCoT, a spoken conversational tutor that uses flexible, adaptive planning and multi-modal task modeling to support the contextualization of learning in reflective dialogues.

Spoken dialogue for simulation control and conversational tutoring

This demonstration shows a flexible tutoring system for studying the effects of different tutoring strategies enhanced by a spoken language interface. The hypothesis is that spoken language increases the effectiveness of automated tutoring. The domain is Navy damage control.