1,721,002 research outputs found
Annotazione dei file raccolti nel progetto SpeechDatCar/VODIS-II
No full textQuesto documento è stato redatto per fornire un riferimento in fase di annotazione del materiale vocale raccolto nell`ambito dei progetti SpeechDatCar e VODIS-II. Il seguente documento è il frutto della standardizzazione delle convenzioni di annotazione decise durante il corso dell`anno 1999. Si vuole innanzitutto descrivere la procedura di elaborazione dei dati, a partire da quelli memorizzati nelle piattaforme PltM (mobile) e PltF (fissa) utilizzate durante la raccolta del database, fino ad arrivare a quelli risultanti dall`annotazione di ciascun file. Si riscontra inoltre una notevole eterogeneità nelle informazioni da includere nei file di documentazione del database risultante. Tali informazioni dovrebbero essere inserite seguendo delle indicazioni fornite attraverso deliverable prodotti nell`ambito dei suddetti progetti; tutavia, tali indicazioni risultano spesso ambigue. Nel seguito del documento si vuole ridurre per quanto possibile tali ambiguità, rendendo in questo modo più uniforme la documentazione finale risultante
Object-Oriented Design Patterns Recovery
No full textObject-Oriented design patterns are an emergent technology: they are reusable micro-architectures, high level building blocks. A system which has been designed using well-known, documented and accepted design patterns is also likely to exhibit good properties such as modularity, separation of concerns and maintainability. While for forward engineering the benefits of using design patterns are clear, using reverse engineering technologies to discover instances of patterns in a software artifact (e.g., design or code) may help in several key areas, among which program understanding, design to code traceability and quality assessment. This paper describes a conservative approach and experimental results, based on a multi-stage reduction stragegy using OO software metrics and structural properties to extract structural design patterns from OO design or C++ code. To assess the effectiveness of the pattern recovery process a portable environment written in Java, remotely accessible by means of any WEB browser, has been developed. The developed system and experimental results on 8 industrial sofware (design and code) and 300.000 line of public domain C++ code are presented
DICIT: Evaluation of a Distant-talking Speech Interface for Television
No full textThe EC-funded project DICIT developed distant-talking interfaces for interactive TV. The final DICIT prototype system processes multimodal user input by speech and remote control. It was designed to understand both natural language and command-and-control-style speech input. We conducted an evaluation campaign to examine the usability and performance of the prototype. The task-oriented evaluation involved na ̈ve test persons and consisted of a subjective part with a usability questionnaire and an objective part. We used thee groups of objective metrics to assess the system: one group related to speech component performance, one related to interface design and user awareness, and a final group related to task-based effectiveness and usability. These metrics were acquired with a dedicated
transcription and annotation tool. The evaluation revealed a quite positive subjective assessments of the system and reasonable objective
results. We report how the objective metrics helped us to determine problems in specific areas and to distinguish design-related issues
from technical problems. The metrics computed over modality-specific groups also show that speech input gives a usability advantage over remote control for certain types of tasks
Design Pattern Recovery in Object-Oriented Software
No full textAn approach to recover object oriented design patterns from C++ code is presented. The pattern recovery process is based on a multi-stage filtering strategy to avoid combinatorial explosion on large software systems. To maintain independence from the language and the case tools adopted in developing software, both design and code are mapped into an intermediate representation. The multi-stage searching strategy allows to safely determine patters candidates. To assess the effectiveness of the pattern recovery process a portable environment written in Java has been developed. Based on this environment, experimental results on public domain and industrial software were obtained and are discussed in the paper. Evidence is shown that, by exploiting information about method calls as a further constraint beyond the structural ones, the number of false alarm is reduced
Using Metrics to identify Design Patterns in Object-Oriented Software
No full textObject-Oriented design patterns are an emergent technology. They are reusable micro-architectures, high level building blocks. This paper presents a conservative approach, based on a multi-stage reduction strategy using OO software metrics and structural properties to extract structural design patterns from OO design or code. Code and design are mapped into an intermediate representation, called Abstract Object Language, to maintain independence from the programming language and the adopted CASE tools. To assess the effectiveness of the pattern recovery process a portable environment written in Java, remotely accessible by means of any WEB browser, has been developed. Based on this environment, experimental results obtained on public domain and industrial software are discussed in this paper
Use of parallel recognizers for robust in-car speech interaction
No full textThis paper refers to an activity under way at the speech recognition technology level for the development of a hands-free dialogue interaction system in the car environment. The use of a set of HMM recognizers, running in parallel, is being investigated in order to ensure low complexity, modularity, fast response, and to allow a real-time reconfiguration of the language models and grammars according to the policy indicated by natural language understanding and dialogue manager modules. A corpus of spontaneous speech interactions was collected using the Wizard-of-Oz method in a real driving situation with a microphone placed far from the driver. The use of parallel recognition units, each specialized on a given geographical domain, was explored using the resulting real corpus. Experiments show the advantage of selecting the recognized sentence according to the maximum likelihood among the active units when compared to the use of a single language model based on a very large vocabulary
- …
