Inspired by previous work, where decipherment is used to improve machine translation, we propose a new idea to combine word alignment and decipherment into a single learning process. We use EM to estimate the model parameters, not only to maximize the probability of parallel corpus, but also the monolingual corpus. We apply our approach to improve Malagasy-English machine translation, where only a small amount of parallel data is available. In our experiments, we observe gains of 0.9 to 2.1 Bleu over a strong baseline.

No abstract

The temporal distribution patterns of bacterial communities, as an important group in mountain soil, are affected by various environmental factors. To improve knowledge regarding the successional seasonal dynamics of the mountain soil bacterial communities, the rhizospheric soil of a 30-year-old natural secondary Pinus tabulaeformis forest, located in the high-altitude (1900 m a.s.l.) of the temperate Qinling Mountains, was sampled and studied during four different seasons. The bacterial community composition and structure in the rhizospheric soil were studied using an Illumina MiSeq Sequencing platform. Furthermore, the edaphic properties and soil enzymatic activities (urease, phosphatase, and catalase) were measured in order to identify the main impact factors on the soil bacterial community. According to the results, all of the edaphic properties and soil enzymatic activities were significantly affected by the seasonal changes, except for the C/N ratio. Although the biomasses of soil bacterial communities increased during the summer and autumn (warm seasons), their Shannon diversity and Pielou’s evenness were decreased. Proteobacteria, Acidobacteria, Actinobacteria, Planctomycetes, and Bacteroidetes were the predominant bacterial groups in all of the soil samples, and the genera of Ktedonobacter, Sphingobium as well as an unclassified member of the Ktedonobacteria were the keystone taxa. The composition and structure of soil bacterial communities were strongly impacted by the edaphic properties, especially the temperature, moisture, ammoniacal nitrogen, available phosphorus and total phosphorus which were the crucial factors to drive the temporal distribution of the soil bacterial community and diversity. In conclusion, the soil temperature, moisture and the nutrients N and P were the crucial edaphic factors for shaping the rhizospheric soil bacterial communities as season and climate change in a P. tabulaeformis forest of Qinling Mountains.

We introduce into Bayesian decipherment a base distribution derived from similarities of word embeddings. We use Dirichlet multinomial regression (Mimno and McCallum, 2012) to learn a mapping between ciphertext and plaintext word embeddings from non-parallel data. Experimental results show that the base distribution is highly beneficial to decipherment, improving state-of-the-art decipherment accuracy from 45.8% to 67.4% for Spanish/English, and from 5.1% to 11.2% for Malagasy/English.

In terrestrial ecosystems, mycorrhizal roots play a key role in the cycling of soil carbon (C) and other nutrients. The impact of environmental factors on the mycorrhizal fungal community has been well studied; however, the seasonal variations in the root-associated fungal microbiota affected by environmental changes are less clear. To improve the understanding of how environmental factors shape the fungal microbiota in mycorrhizal roots, seasonal changes in Pinus tabuliformis root-associated fungi were investigated. In the present study, the seasonal dynamics of edaphic properties, soil enzymatic activities, root fungal colonization rates, and root-associated fungal microbiota in P. tabuliformis forests were studied across four seasons during a whole year to reveal their correlations with environmental changes. The results indicate that the soil functions, such as the enzymatic activities related to nitrogen (N) and phosphorus (P) degradation, were varied with the seasonal changes in microclimate factors, resulting in a significant fluctuation of edaphic properties. In addition, the ectomycorrhizal fungal colonization rate in the host pine tree roots increased during warm seasons (summer and autumn), while the fungal colonization rate of dark septate endophyte was declined. Moreover, the present study indicates that the fungal biomass increased in both the pine roots and rhizospheric soils during warm seasons, while the fungal species richness and diversity decreased. While the Basidiomycota and Ascomycota were the two dominant phyla in both root and soil fungal communities, the higher relative abundance of Basidiomycota taxa presented in warm seasons. In addition, the fungal microbial network complexity declined under the higher temperature and humidity conditions. The present study illustrates that the varieties in connectivity between the microbial networks and in functional taxa of root-associated fungal microbiota significantly influence the soil ecosystem functions, especially the N and P cycling.