Search

Leveraging knowledge graph will benefit question answering tasks, as KG contains well-structured informative data. However, training knowledge graph-based simple question answering systems is known computationally expensive due to the complex predicate extraction and candidate pool generation. Moreover, the existing methods based on convolutional neural network (CNN) or recurrent neural network (RNN) overestimate the importance of predicate features thus reduce performance. To address these challenges, we propose a time-efficient and resource-effective framework. We use leaky n-gram to balance recall and candidate pool size in candidate pool generation. For predicate extraction, we propose a soft-histogram and self-attention (SHSA) module which serves the role of preserving the global information of questions via feature matrices. And this leads to reduce the RNN module as the simple feedforward network in predicate representation. We also designed a Hamming lower-bound label encoding algorithm to encode the label representations in lower dimensions. Experiments on benchmark datasets show that our method outperforms the competitive work for end-tasks and achieves better recall with a significantly pruned candidate space.

This research, which investigates a set of fundamental relationships in the R&D literature, is based on an unusually rich set of panel data covering the population of China’s large and medium-size manufacturing enterprises. Using a recursive three-equation system, we investigate the determinants of firm-level R&D intensity, the process of knowledge production, and the impact of innovation on firm performance. Several results stand out. Overall, the statistical relationships within the model are surprisingly robust, including the contributions of R&D expenditure to new product (NP) innovation, productivity, and profitability. The roles of firm size, market concentration, and profitability in driving R&D effort parallel to those found in the US literature. We find that new product (NP) innovation accounts for approximately 12% of the total returns to R&D. Also, returns to industrial R&D in China appear to be at least three to four times the returns to fixed production assets.

total of 41 patients were enrolled: 19 patients (46%) had known CD, 13 patients (32%) were being evaluated for suspected CD, and 9 patients (22%) had an endoscopy for evaluation of GERD or abdominal. The total number of biopsy pieces available for analysis was 85 (mean 2.1 per patient) in the 1-pass group and 74 (mean 2.0 per patient) in the 2-pass group. A total of 49% of specimens were well-oriented: 17 patients (41%) had well-oriented specimens with the 1-pass method and 23 patients (56%) had well-oriented specimens with the 2-pass method. A repeated measures logistic regression analysis was strongly suggestive of improved orientation with the 2-pass method for all patients (OR 1.80; 95% CI 0.84 3.86). In subset analyses of patients with and without CD, the trend towards improved orientation with the 2-pass method persisted. In 8 of 19 (42%) patients with known CD, the Marsh score differed between the 1and 2pass methods; this included 3 patients (16%) in whom the Marsh score differed by two or more stages. Conclusions: Our study suggests a trend towards improved orientation of duodenal biopsy specimens with the 2-pass method compared to the 1-pass method. Practitioners should consider using the 2-pass method among patients undergoing endoscopy for known or suspected CD.