Maxine Eskenazi

This paper introduces the Ninth Dialog System Technology Challenge (DSTC-9). This edition of the DSTC focuses on applying end-to-end dialog technologies for four distinct tasks in dialog systems, namely, 1. Task-oriented dialog Modeling with Unstructured Knowledge Access, 2. Multi-domain task-oriented dialog, 3. Interactive evaluation of dialog and 4. Situated interactive multimodal dialog. This paper describes the task definition, provided datasets, baselines, and evaluation setup for each track. We also summarize the results of the submitted systems to highlight the general trends of the state-of-the-art technologies for the tasks.

Language models have steadily increased in size over the past few years. They achieve a high level of performance on various natural language processing (NLP) tasks such as question answering and summarization. Large language models (LLMs) have been used for generation and can now output human-like text. Due to this, there are other downstream tasks in the realm of dialog that can now harness the LLMs' language understanding capabilities. Dialog evaluation is one task that this paper will explore. It concentrates on prompting with LLMs: BLOOM, OPT, GPT-3, Flan-T5, InstructDial and TNLGv2. The paper shows that the choice of datasets used for training a model contributes to how well it performs on a task as well as on how the prompt should be structured. Specifically, the more diverse and relevant the group of datasets that a model is trained on, the better dialog evaluation performs. This paper also investigates how the number of examples in the prompt and the type of example selection used affect the model's performance.

The DialPort project http://dialport.org/, funded by the National Science Foundation (NSF), covers a group of tools and services that aim at fulfilling the needs of the dialog research community. Over the course of six years, several offerings have been created, including the DialPort Portal and DialCrowd. This paper describes these contributions, which will be demoed at SIGDIAL, including implementation, prior studies, corresponding discoveries, and the locations at which the tools will remain freely available to the community going forward.

Jessica Huynh, Shikib Mehri, Cathy Jiao, Maxine Eskenazi. Proceedings of the 23rd Annual Meeting of the Special Interest Group on Discourse and Dialogue. 2022.

Shikib Mehri, Yasemin Altun, Maxine Eskenazi. Proceedings of the 23rd Annual Meeting of the Special Interest Group on Discourse and Dialogue. 2022.

In order to develop computer tools for speech therapy that reliably classify speech productions, there is a need for speech production corpora that characterize the target population in terms of age, gender, and native language. Apart from including correct speech productions, in order to characterize the target population, the corpora should also include samples from people with speech sound disorders. In addition, the annotation of the data should include information on the correctness of the speech productions. Following these criteria, we collected a corpus that can be used to develop computer tools for speech and language therapy of Portuguese children with sigmatism. The proposed corpus contains European Portuguese children’s word productions in which the words have sibilant consonants. The corpus has productions from 356 children from 5 to 9 years of age. Some important characteristics of this corpus, that are relevant to speech and language therapy and computer science research, are that (1) the corpus includes data from children with speech sound disorders; and (2) the productions were annotated according to the criteria of speech and language pathologists, and have information about the speech production errors. These are relevant features for the development and assessment of speech processing tools for speech therapy of Portuguese children. In addition, as an illustration on how to use the corpus, we present three speech therapy games that use a convolutional neural network sibilants classifier trained with data from this corpus and a word recognition module trained on additional children data and calibrated and evaluated with the collected corpus.

Pedro Mota, Maxine Eskenazi, Luísa Coheur. Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL). 2019.

Research on topic segmentation has recently focused on segmenting documents by taking advantage of documents covering the same topics. In order to properly evaluate such approaches, a dataset of related documents is needed. However, existing datasets are limited in the number of related documents per domain. In addition, most of the available datasets do not consider documents from different media sources (PowerPoints, videos, etc.), which pose specific challenges to segmentation. We fill this gap with the MUltimedia SEgmentation Dataset (MUSED), a collection of documents manually segmented, from different media sources, in seven different domains, with an average of twenty related documents per domain. In this paper, we describe the process of building MUSED. A multi-annotator study is carried out to determine if it is possible to observe agreement among human judges and characterize their disagreement patterns. In addition, we use MUSED to compare the state-of-the-art topic segmentation techniques, including the ones that take advantage of related documents. Moreover, we study the impact of having documents from different media sources in the dataset. To the best of our knowledge, MUSED is the first dataset that allows a straightforward evaluation of both single- and multiple-documents topic segmentation techniques, as well as to study how these behave in the presence of documents from different media sources. Results show that some techniques are, indeed, sensitive to different media sources, and also that current multi-document segmentation models do not outperform previous models, pointing to a research line that needs to be boosted.

In this paper we propose a graph-community detection approach to identify cross-document relationships at the topic segment level. Given a set of related documents, we automatically find these relationships by clustering segments with similar content (topics). In this context, we study how different weighting mechanisms influence the discovery of word communities that relate to the different topics found in the documents. Finally, we test different mapping functions to assign topic segments to word communities, determining which topic segments are considered equivalent. By performing this task it is possible to enable efficient multi-document browsing, since when a user finds relevant content in one document we can provide access to similar topics in other documents. We deploy our approach in two different scenarios. One is an educational scenario where equivalence relationships between learning materials need to be found. The other consists of a series of dialogs in a social context where students discuss commonplace topics. Results show that our proposed approach better discovered equivalence relationships in learning material documents and obtained close results in the social speech domain, where the best performing approach was a clustering technique.

The CMU Let's Go Spoken Dialogue System has been used daily for about three years to answer calls to the Pittsburgh Port Authority for bus information in the evening and on weekends. This has resulted in a database of over 50 000 spoken dialogues as of January 2008, one of the largest publicly available sets of this type of data. While retraining the system with part of this data, it became apparent that there are times of the day, of the week and of the year when the average number of successful calls is significantly higher. We will present evidence, using these three measures of time (hour, day of week, month of year) and criteria such as signal‐to‐noise ratio, estimated success rate, number of turns per dialogue, number of non‐understandings per dialogue, and barge‐in rate to detect the regular, predictable appearance of high and low success rates and to suggest methods for palliating this effect in order to increase overall dialogue success rates.