289 research outputs found
Ontology Learning from Text: An Overview
No full textBuitelaar P, Cimiano P, Magnini B. Ontology Learning from Text: An Overview. In: Buitelaar P, Cimiano P, Magnini B, eds. Ontology Learning from Text: Methods, Evaluation and Applications. Frontiers in Artificial Intelligence and Applications. Vol 123. Amsterdam: IOS Press; 2005: 3-12
Learning Taxonomic Relations from Heterogeneous Sources of Evidence
No full textCimiano P, Pivk A, Schmidt-Thieme L, Staab S. Learning Taxonomic Relations from Heterogeneous Sources of Evidence. In: Buitelaar P, Cimiano P, Magnini B, eds. Ontology Learning from Text: Methods, Evaluation and Applications. Frontiers in Artificial Intelligence; 123. Amsterdam: IOS Press; 2005: 59-73
Exploiting lexical expansions and Boolean compositions for web querying
No full textThis paper describes an experiment aiming at evaluating the role of NLP based optimizations (i.e. morphological derivation and synonymy expansion) in web search strategies. Keywords and their expansions are composed in two different Boolean expressions (i.e. expansion insertion and Cartesian combination) and then compared with a keyword conjunctive composition, considered as the baseline. Results confirm the hypothesis that linguistic optimizations significantly improve the search engine performances
Opposition Relations among Verb Frames
No full textIn this paper we propose a scheme for annotating opposition relations among verb frames in lexical resources. The scheme is tested on the T-PAS resource, an inventory of typed predicate argument structures for Italian, conceived for both linguistic research and computational tasks. After discussing opposition relations from a linguistic point of view and listing the tags we decided to use, we report the results of the experiment we performed to test the annotation scheme, in terms of interannotation agreement and linguistic analysis of annotated data
Open Domain Question/Answering on the WEB
No full textThis paper presents a question/answering system for searching the web. The system accepts natural language questions in Italian and returns as answer a ranked list of document passages together with the URL of the whole document. Three crucial aspects related to the web scenario have been investigated: the linguistic expansion of the query, the optimization of the search boolean expression, the evaluation of the results
Dynamic Task-Oriented Dialogue: A Comparative Study of Llama-2 and BERT in Slot Value Generation
No full textRecent advancements in instruction-based language models have demonstrated exceptional performance across various natural language processing tasks. We present a comprehensive analysis of the performance of two open-source language models, BERT and Llama-2, in the context of dynamic task-oriented dialogues. Focusing on the Restaurant domain and utilizing the MultiWOZ 2.4 dataset, our investigation centers on the models’ ability to generate predictions for masked slot values within text. The dynamic aspect is introduced through simulated domain changes, mirroring real-world scenarios where new slot values are incrementally added to a domain over time.This study contributes to the understanding of instruction-based models’ effectiveness in dynamic natural language understanding tasks when compared to traditional language models and emphasizes the significance of open-source, reproducible models in advancing research within the academic community
Recent Advancements in Human Language Technology in Italy
No full textThis paper presents significant advancements in Human Language Technology fostered by Italian researchers in the last years. We report recent results in the following research directions: Distributional Semantics (i.e. the COMPOSES project), Multilingual Word Sense Disambiguation (the MultiJEDI project), Textual Semantic Inferences (the EXCITEMENT project), and Computer Assisted Translation (the MATECAT project). Key aspects that are common to such research initiatives include the fact that they are funded by the European Commission, under different grants, after having passed a very selective international competition; that they are led by Italian researchers, this way showing both the prominent role and the maturity of the Italian community on the international scene; and that they are currently running, indicating that the research topics of the projects are in the mainstream of Computational Linguistics
Numerical investigation of the influence of leading and sequential bubbles on slug flow boiling within a microchannel
No full textMultiphase CFD simulations are presently employed to investigate the flow boiling of multiple sequential elongated bubbles in a horizontal microchannel. Most of the computational studies published so far explored the features of boiling flows within microchannels by simulating the fluid-dynamics of a single evaporating bubble, but the present work shows that multiple bubble simulations are necessary to capture the essential features of the heat transfer process of a slug flow. In particular, it is shown that leading and sequential bubbles interact thermally and hydrodynamically due to the evaporation process, thus possessing different growth rates, velocities and thicknesses of the thin liquid films trapped between the bubbles interfaces and the channel wall. The evaporation of this thin liquid film is the dominant heat transfer mechanism in the vapor bubble region and the transit of trailing bubbles strongly enhances the time-averaged heat transfer coefficient of the bubble-liquid slug unit, by as much as 60% higher relative to the leading bubble under the operating conditions presently set. Furthermore, the presence of a recirculating vortex just after the tail of the bubble in the liquid slug trapped between the bubbles was found in the simulations, significantly improving the heat transfer between the wall and the bulk liquid, thus maintaining the heat transfer coefficient much higher than otherwise expected in the liquid slug region as well. Finally, a new multiple bubble heat transfer model is proposed to predict the local variation of the heat transfer coefficient, which might prove to be useful to improve the current boiling heat transfer methods, such as the three-zone model of Thome et al. [1,2]. The numerical framework employed to perform this study was the commercial CFD solver ANSYS Fluent 12 with a Volume Of Fluid interface capturing method, which was improved here by implementing external functions, in particular a Height Function method to better estimate the surface tension force and an evaporation model to compute the phase change. (C) 2013 Elsevier Masson SAS. All rights reserved.LTC
Height function interface reconstruction algorithm for the simulation of boiling flows
No full textIn this paper we present our results on numerical study of vapor bubbles growing in quiescent superheated liquid, as effect of liquid evaporation at the interface. Height Function interface reconstruction algorithm is coupled with an evaporation model based on continuum field representation of source terms. The flow solver is a finitevolume CFD code. Interface is tracked within a Volume-Of-Fluid framework. Continuum-Surface-Force method accounts for surface tension effects. Vapor bubble heat-transfer-controlled growth is simulated for three different working fluids: water, HFE-7100 and R134a. Accuracy of interface reconstruction algorithm is of maximum importance. Unbalance between pressure gradients and surface tension forces at interface leads to the growth of an unphysical velocity field which switches original only diffusive heat transfer mechanism to combined diffusive-convective one. Height Function algorithm reduces the magnitude of this unreal velocity field. Standard test cases are considered to assess the performances of implemented version, through comparison with the widely used Youngs algorithm. Keywords: volume of fluid, height function, evaporation, surface tension, bubbles. 1 Introduction Modern numerical implementation of surface tension driven flows, with phase change, begins with the work of Juric and Tryggvason [1]. Earlier methods used a Lagrangian approach with a moving mesh following the interface. The limit of these methods was that they could not manage large interface deformation. Juric and Tryggvason developed a Front Tracking algorithm where Lagrangian evolution of interface is tracked on a fixed grid. Welch and Wilson [2] and So
- …
