Catálogo de publicaciones - libros

The aim of the work described in this paper is to extend the EPFL dialogue platform with multimodal capabilities. Based on our experience with the EPFL Rapid Dialogue Prototyping Methodology (RDPM), we formulate precise design principles that provide the necessary frame to use the RDPM to rapidly create an efficient multimodal interface for a given application. We analyze the consequences of the proposed design principles on the generic GUI and architecture required for the system.

As language is fundamental to human activities, proficiency in other languages becomes important. Besides for developing abilities for communication, the knowledge is also a tool for a survival. With the introduction of computerized mobile devices, i.e. PDAs, new opportunities for communicating in other language arose. This paper describes a new communication paradigm that is language independent using icon language on a PDA. Users can create iconic messages as realization of their concepts or ideas in mind. The proof of concept tool is able to interpret and convert the messages to (natural language) text and speech in different languages. To provide faster interactions in next icon selection, the tool has icon prediction. Our user test results confirmed that using provided icons our target users could express their concepts and ideas solely using a spatial arrangement of icons.

We have used text, graphics and non-speech audio to tutor new users in a spoken dialogue system. The guidance is given by a software tutor, a software component that interactively tutors the user. Four different variations of tutoring were implemented and experiences were collected from user tests in order to gain insights into these tutoring concepts. Real-time visualization of speech interaction with comic book style balloons and structured guidance were received well while various other methods received mixed acceptance.

Questions have been analyzed in Estonian information dialogues with the purpose of finding out the linguistic features that can be used in automatic recognition of various types of questions. Information questions (i.e. the questions that are used for requesting information) and questions that initiate solving communication problems are considered. The study shows which types of questions (wh-question, open and closed yes/no question, question that offers answer, alternative question) are preferred in both cases. The results can be implemented in a dialogue system which performs the role of information provider and interacts with a user in Estonian.

In a previous paper we proved that Named Entity Recognition plays an important role to improve Question Answering by both increasing the quality of the data and by reducing its quantity. Here we present a more in-depth discussion, studying several ways in which NER can be applied in order to produce a maximum data reduction. We achieve a 60% reduction without significant data loss and a 92.5% with a reasonable implication in data quality.

An enriched corpus annotation lies at the core of this paper. We will argue for an annotation that builds upon the tagging of moves and is thus meant as addendum to existing annotations. To this end, we will propose some way of marking up the fulfilment and violation of constraints by an -annotation which thereby allows to predict where rhetorical relations might occur: only if all constraints are fulfilled between two discourse units a rhetorical relation can obtain between them. As a starting point, two constraints will be explained and annotated which play an important role in discourse, namely Polanyi’s and a constraint we will term . Our approach acknowledges and supports Asher’s and Lascarides’ principle [1].

In this paper we present a new method to improve the coverage of Passage Retrieval (PR) systems when these systems are employed for the Question Answering (QA) tasks. The ranking of passages obtained by the PR system is rearranged to emphasize those passages with more probability to contain the answer. The new ranking is based on finding the n-gram structures of the question that are presented in the passage, and the weight of the passages increases when they contain longer n-grams structures of the question. The results we present show that the application of this method improves notably the coverage of the classical PR system based on the Space Vectorial Model.

We are interested in the area of Dialogue Act () tagging. Identifying the dialogue acts of utterances is recognised as an important step towards understanding the content and nature of what speakers say. We have built a simple dialogue act classifier based on purely features – principally word n-gram cue phrases. Although such a classifier performs surprisingly well, rivalling scores obtained using far more sophisticated language modelling techniques for the corpus we address, we want to understand further the issues raised by this approach. We have performed an error analysis of the output of our classifier, with a view to casting light both on the system’s performance, and on the classification scheme itself.