Scientific claim verification can help the researchers to easily find the target scientific papers with the sentence evidence from a large corpus for the given claim.Some existing works propose pipeline models on the three tasks of abstract retrieval, rationale selection and stance prediction.Such works have the problems of error propagation among the modules in the pipeline and lack of sharing valuable information among modules.We thus propose an approach, named as ARSJOINT, that jointly learns the modules for the three tasks with a machine reading comprehension framework by including claim information.In addition, we enhance the information exchanges and constraints among tasks by proposing a regularization term between the sentence attention scores of abstract retrieval and the estimated outputs of rational selection.The experimental results on the benchmark dataset SCI-FACT show that our approach outperforms the existing works.

The data imbalance problem is a crucial issue for the multi-label text classification.Some existing works tackle it by proposing imbalanced loss objectives instead of the vanilla cross-entropy loss, but their performances remain limited in the cases of extremely imbalanced data.We propose a hybrid solution which adapts general networks for the head categories, and few-shot techniques for the tail categories.We propose a Hybrid-Siamese Convolutional Neural Network (HSCNN) with additional technical attributes, i.e., a multi-task architecture based on Single and Siamese networks; a category-specific similarity in the Siamese structure; a specific sampling method for training HSCNN.The results using two benchmark datasets and three loss objectives show that our method can improve the performance of Single networks with diverse loss objectives on the tail or entire categories.

Workflow management systems (WFMS) facilitate the routine operation of business processes and gain popularity in recent years. To ensure the correctness of business process specification and execution, model verification (especially structure verification) must be conducted so that we can identify any violations and consequently take proper action to remove them in time. However, little progress has been made on the perfect business process model syntactic and semantic structure verification in the workflow management environment, especially simple efficient algorithm is lack. In this paper, we present a biologically-inspired algorithm: Wolf-pack algorithm (WPA) for workflow syntactic and semantic structure verification. Our algorithm responds to distinguishing syntactic structure inconsistencies (such as deadlock, lack of synchronization, dead loop) and semantic structure inconsistencies (such as data source conflict) from workflow specification. Furthermore, we apply the algorithm to JTangFlow workflow management system and achieved good results.

Production system which accepts the facts and draws conclusions by repeatedly matching facts with rules plays an important role of improving the business by providing agility and flexibility. However, rule matching in production is badly time-consuming, and single computer limits the improvement for current matching algorithm. To address these problems, we proposed a MapReduce-based architecture to implement the distributed and parallel matching in different computers running with Rete algorithm. The architecture would benefit production system in performance, large scale of rules and facts are for special. This paper firstly formalizes some definitions for an accurate description, then not only discusses the details of implementation for different stages of the architecture but also shows the high efficiency through the experiment. At the end, we mention some complex factors which will be considered in the future for better performance.