André Machado

Statistically-Driven Alignment-Based Multiword Expression Identification for Technical Domains

Multiword Expressions (MWEs) are one of the stumbling blocks for more precise Natural Language Processing (NLP) systems. Particularly, the lack of coverage of MWEs in resources can impact negatively on the performance of tasks and applications, and can lead to loss of information or communication errors. This is especially problematic in technical domains, where a significant portion of the vocabulary is composed of MWEs. This paper investigates the use of a statistically-driven alignment-based approach to the identification of MWEs in technical corpora. We look at the use of several sources of data, including parallel corpora, using English and Portuguese data from a corpus of Pediatrics, and examining how a second language can provide relevant cues for this tasks. We report results obtained by a combination of statistical measures and linguistic information, and compare these to the reported in the literature. Such an approach to the (semi-)automatic identification of MWEs can considerably speed up lexicographic work, providing a more targeted list of MWE candidates.

A hybrid approach for multiword expression identification

Considerable attention has been given to the problem of Multiword Expression (MWE) identification and treatment, for NLP tasks like parsing and generation, to improve the quality of results. Statistical methods have been often employed for MWE identification, as an inexpensive and language independent way of finding co-occurrence patterns. On the other hand, more linguistically motivated methods for identification, which employ information such as POS filters and lexical alignment between languages, can produce more targeted candidate lists. In this paper we propose a hybrid approach that combines the strenghts of different sources of information using a machine learning algorithm to produce more robust and precise results. Automatic evaluation on gold standards shows that the performance of our hybrid method is superior to the individual results of statistical and alignment-based MWE extraction approaches for Portuguese and for English. This method can be used to aid lexicographic work by providing a more targeted MWE candidate list.

Identification of Multiword Expressions in Technical Domains: Investigating Statistical and Alignment-Based Approaches

Multiword Expressions (MWEs) are one of the stumbling blocks for more precise Natural Language Processing (NLP) systems. The lack of coverage of MWEs in resources can impact negatively on the performance of tasks and applications, and can lead to loss of information or communication errors; especially in technical domains where MWE are frequent. This paper investigates some approaches to the identification of MWEs in technical corpora based on: association measures, part-of-speech and lexical alignment information. We examine the influence of some factors on their performance such as sources of information for identification and evaluation. While the association measures emphasize recall, the alignment method focuses on precision.

O Simulador da zona de combustão do alto-forno do Laboratório de Siderurgia da UFRGS - uma nova ferramenta na seleção de carvões para injeção

Blast furnace (BF) is the main route for pig iron production. One of the biggest challenges in BF steel industry is the reduction of coke consumption. Cokemaking process employs high cost coals and accounts for the most of polluting gas emissions from integrated steel mills. A successful strategy for saving coke is the injection of auxiliary fuels in tuyere region of BF through PCI (Pulverized Coal Injection) process. Fuel combustion generates energy and reducing gases for iron ore. When injected, coal is subjected to drastic conditions in the raceway of blast furnace: temperatures around 2400 K, heating rates in order of 10 5 to 10 6 K/s and residence times of 10 to 40 ms. Therefore total combustion of coal, especially at high injection rates, is unlikely. It is fundamental to select coals for PCI with good combustibility, thus minimizing the generation of residual solid (char), which can damage the BF operation. There are several types of reactors for the evaluation of coal combustion, such as thermobalances, drop tube furnaces and PCI test rigs. Thermobalances are fixed-bed reactors operating at low heating rates and drop tube furnaces are more suitable for boiler simulation and evaluation of single coals. For operational, strategic and logistical reasons steel mills use coal blends of different characteristics for PCI. Nowadays PCI test rigs are the best option of reactors for evaluation of coals and other fuels for injection. In this context, the Iron and Steelmaking Laboratory of the Federal University of Rio Grande do Sul (LaSid-UFRGS) designed and developed a PCI test rig with innovative features, such as collection of char, quantitative analysis of combustion gases by gas chromatography and acquisition of temperature and pressure data by ultrafast sensors. Unlike the existing models, the equipment is fully automated, being the only one in the country in vertical arrangement, minimizing pressure loss and allowing operation with greater stability. The present work aims to present the characteristics of the LaSid-UFRGS PCI test rig and preliminary experimental results. Key-Words: Ironmaking. Blast furnace. Raceway. Fuels. Pulverized coal. V CONGRESSO BRASILEIRO DE CARVÃO MINERAL CRICIÚMA SC – BRASIL 29 DE MAIO A 01 DE JUNHO DE 2017