In this paper, we evaluate the progress of our field toward solving simple factoid questions over a knowledge base, a practically important problem in natural language interface to database. As in other natural language understanding tasks, a common practice for this task is to train and evaluate a model on a single dataset, and recent studies suggest that SimpleQuestions, the most popular and largest dataset, is nearly solved under this setting. However, this common setting does not evaluate the robustness of the systems outside of the distribution of the used training data. We rigorously evaluate such robustness of existing systems using different datasets. Our analysis, including shifting of training and test datasets and training on a union of the datasets, suggests that our progress in solving SimpleQuestions dataset does not indicate the success of more general simple question answering. We discuss a possible future direction toward this goal.
In this paper, we apply the entitymetrics model to our constructed Gene-Citation-Gene (GCG) network. Based on the premise there is a hidden, but plausible, relationship between an entity in one article and an entity in its citing article, we constructed a GCG network of gene pairs implicitly connected through citation. We compare the performance of this GCG network to a gene-gene (GG) network constructed over the same corpus but which uses gene pairs explicitly connected through traditional co-occurrence. Using 331,411 MEDLINE abstracts collected from 18,323 seed articles and their references, we identify 25 gene pairs. A comparison of these pairs with interactions found in BioGRID reveal that 96% of the gene pairs in the GCG network have known interactions. We measure network performance using degree, weighted degree, closeness, betweenness centrality and PageRank. Combining all measures, we find the GCG network has more gene pairs, but a lower matching rate than the GG network. However, combining top ranked genes in both networks produces a matching rate of 35.53%. By visualizing both the GG and GCG networks, we find that cancer is the most dominant disease associated with the genes in both networks. Overall, the study indicates that the GCG network can be useful for detecting gene interaction in an implicit manner.
In this paper we investigated the socio-cultural behavior of users reflected in the two different social media channels, YouTube and Twitter. We conducted the comparative analysis of the networks generated from the two channels. The relationship we set for each network is the relatedness on YouTube and the co-links on Twitter. From the results, we revealed that the social media influenced the distinct socio-cultural behaviors of their users. Specifically, Twitter network better showed the actual consumption of contents in the field of the k-pop culture than YouTube. From this study, we contributed to offer a novel approach for exploring the socio-cultural behavior of users on the social media.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.