Search

Grammatical Error Correction (GEC) is the task of detecting and correcting various grammatical errors in texts. Many previous approaches to the GEC have used various mechanisms including rules, statistics, and their combinations. Recently, the performance of the GEC in English has been drastically enhanced due to the vigorous applications of deep neural networks and pretrained language models. Following the promising results of the English GEC tasks, we apply the Transformer with Copying Mechanism into the Korean GEC task by introducing novel and effective noising methods for constructing Korean GEC datasets. Our comparative experiments showed that the proposed system outperforms two commercial grammar check and other NMT-based models.

We previously described the use of a cell-based screening approach to identify small molecules that regulate adipocyte differentiation. Here we identify the amiloride derivative phenamil as an adipogenic compound. Phenamil acutely induces expression of the key transcription factor of adipogenesis, peroxisome proliferator-activated receptor γ (PPARγ) and, consequently, promotes the differentiation of multiple preadipocyte cell lines, including 3T3-L1 and F442A. Interestingly, the adipogenic action of phenamil is distinct from and additive with both PPARγ ligands and the previously identified adipogenic small molecule harmine. To identify signaling pathways mediating phenamil's effects, we performed transcriptional profiling of 3T3-F442A preadipocytes. ETS variant 4 (ETV4) was identified as a gene rapidly induced by phenamil but not by other adipogenic small molecules or PPARγ agonists. Transient expression of ETV4 in preadipocytes enhances the expression of PPARγ. Stable overexpression of ETV4 promotes expression of PPARγ and its downstream target genes and enhances morphological differentiation. Finally, knockdown of PPARγ expression by shRNA blocks the effects of phenamil on adipocyte differentiation and gene expression, but it does not block phenamil induction of ETV4, which suggests that ETV4 acts upstream of PPARγ in differentiation processes. These results identify a phenamil as new small molecule tool for the probing of adipocyte differentiation that acts, at least in part, through induction of ETV4 expression.

One of the key issues for providing users user-customized or context-aware services is to automatically detect latent topics, users’ interests, and their changing patterns from large-scale social network information. Most of the current methods are devoted either to discovering static latent topics and users’ interests or to analyzing topic evolution only from intrafeatures of documents, namely, text content, without considering directly extrafeatures of documents such as authors. Moreover, they are applicable only to the case of single processor. To resolve these problems, we propose a dynamic users’ interest discovery model with distributed inference algorithm, named as Distributed Author-Topic over Time (D-AToT) model. The collapsed Gibbs sampling method following the main idea of MapReduce is also utilized for inferring model parameters. The proposed model can discover latent topics and users’ interests, and mine their changing patterns over time. Extensive experimental results on NIPS (Neural Information Processing Systems) dataset show that our D-AToT model is feasible and efficient.

Sensor data is structured and generally lacks of meaning by itself, but life-logging data (time, location, etc.) out of sensor data can be utilized to create lots of meaningful information combined with social data from social networks like Facebook and Twitter. There have been many platforms to produce meaningful information and support human behavior and context-awareness through integrating diverse mobile, social, and sensing input streams. The problem is that these platforms do not guarantee the performance in terms of the processing time and even let the accuracy of output data be addressed by new studies in each area where the platform is applied. Thus, this study proposes an improved platform which builds a knowledge base for context awareness by applying distributed and parallel computing approach considering the characteristics of sensor data that is collected and processed in real-time, and compares the proposed platform with existing platforms in terms of performance. The experiment shows the proposed platform is an advanced platform in terms of processing time. We reduce the processing time by 40% compared with existing platform. The proposed platform also guarantees the accuracy compared with existing platform.

Recently, due to the popularization of the smartphone and social network service (SNS), many SNS users write their opinions for social events. According to these social phenomena, social sensor network which analyzes social events by utilizing those users' text data is proposed. Parsing is essential module to analyze user's text contents because it gives the understanding of semantics by extracting the words and their classes from texts. However, parsing requires much time because it needs to analyze all context information from the users' text. In addition, as users' text data are generated and transferred in streaming, the required parsing time increases too. This situation occurs that it is hard to parse the text on the single machine. Therefore, to drastically enhance the parsing speed, we propose distributed and parallel parsing system on the MapReduce. It applies the legacy parser to the MapReduce through loose coupling. Also, to reduce communication overheads, the statistical model used by the parser is resided on local cache in each mapper. The experimental result shows that the speed of proposed system is 2–19 times better than that of the legacy parser. As a result, we prove that the proposed system is useful for parsing text data in social sensor network.