The Suboptimal WMT Test Sets and Its Impact on Human Parity

With the advent of Neural Machine Translation, the more the achievement of human-machine parity is claimed at WMT, the more we come to ask ourselves if their evaluation environment can be trusted. In this paper, we argue that the low quality of the source test set of the news track at WMT may lead to an overrated human parity claim. First of all, we report nine types of so-called technical contaminants in the data set, originated from an absence of meticulous inspection after web-crawling. Our empirical findings show that when they are corrected, about 5% of the segments that have previously achieved a human parity claim turn out to be statistically invalid. Such a tendency gets evident when the contaminated sentences are solely concerned. To the best of our knowledge, it is the first attempt to question the “source” side of the test set as a potential cause of the overclaim of human parity. We cast evidence for such phenomenon that according to sentence-level TER scores, those trivial errors change a good part of system translations. We conclude that to overlook it would be a mistake, especially when it comes to an NMT evaluation.

Mineralization treatment of European oak heartwood with calcium oxalate for improved fire retardancy

Within the focus to apply substances for wood protections, here fire retardants, with low hazardous and low environmental impact is of interest. Additionally, European oak is an attractive species for various interior and exterior applications. However, oak is classified as very heavily treatable and thus impregnation is challenging. However, the focus of this study was to treat European oak with a new fire retardant based on an in-situ calcium oxalate deposition. Thin oak specimens with a thickness of 4 mm were investigated with two various formulations of aqueous salt solutions (potassium oxalate and calcium chloride, and potassium oxalate and calcium acetate) to obtain an in-situ mineralization of calcium oxalate during a two-step impregnation process. The uptake, the distribution, and the penetration of the salts for both applied formulations were investigated. Additionally, fire retardant properties were investigated in a single flame source test. It could be demonstrated that an acceptable degree of treatability was achieved for both applied formulations. The fire retardancy of the so mineralized material was clearly improved. Synergetic effects which might be caused by the reaction side products of the various formulations were found to be neglectable.

Plagiarism of the implicit concept in interior design projects: Does it exist?

Design spaces can be read in two complementary structural and compositional dimensions: the implicit dimension, and the formal dimension. This study aimed to answer the following questions: What are the phases of the original idea transformation in the design process? Does taking advantage of the implicit dimension of others’ interior spaces count as inspiration or plagiarism? Participants comprised nine postgraduate students, who were asked to copy others’ project ideas by applying four design phases: idea source, test, choice, and output. The results showed that plagiarism of the concept of others’ work could not occur unless the concept copies the structural, functional, and aesthetic aspects of the design space at the level of its components and the relationships between them. Otherwise, taking advantage of the implicit dimension of others’ project ideas can be an inspiration. The results are expected to increase awareness among interior design instructors in higher education.

176 Evaluation of a novel threonine source for nursery pig diets

Four hundred and eighty (PIC 337 X 1050, PIC Genus, Hendersonville, TN) pigs were used to evaluate a novel threonine source (ThrPro, CJ America-Bio, Fort Dodge, IA) for nursery pigs from approximately 7 kg to 20 kg. At weaning, pigs were sorted by gender and fed a common diet for one week. Upon completion of the first week, pigs were sorted into randomized complete blocks, equalized by weight, within 16 replications. Pigs were allocated to one of three dietary treatments: positive control (POS)-SID LYS:THR 0.60, negative control (NEG)-SID LYS:THR ≤0.46 and alternative threonine source (TEST)-SID LYS:THR 0.60. All other nutrients met or exceeded the NRC (2012) recommendations. Growth and intake data were analyzed using PROC MIXED procedure in SAS. The experimental unit was the pen. During the first 14 days, pigs fed TEST had similar ADG (0.417 vs. 0.414 kg/d, P=0.81) and G:F (0.977 vs. 1.030, P=0.18) compared to POS, and increased ADG (0.417 vs. 0.387 kg/d, P=0.01) and G:F (0.977 vs. 0.898, P=0.05) compared to NEG. Over days 14-28, pigs fed TEST had similar ADG (0.523 vs. 0.532 kg/d, P=0.49) and G:F (0.712 vs. 0.707, P=0.71) compared to POS, and increased ADG (0.523 vs. 0.479 kg/d, P=0.002) and G:F (0.712 vs. 0.627, P< 0.0001) compared to NEG. Overall (d 0 to 28), pigs fed TEST had similar ADG (0.466 vs. 0.474 kg/d, P=0.48) and G:F (0.808 vs. 0.816, P=0.55) compared to POS, and increased ADG (0.466 vs. 0.433 kg/d, P=0.002) and G:F (0.808 vs. 0.725, P< 0.0001) compared to NEG. ADFI was not significantly different across treatments for the entirety of the study. In conclusion, the replacement of L-threonine with a novel threonine source resulted in similar growth performance in nursery pigs from approximately 7 kg to 20 kg.