Neural network models usually suffer from the challenge of incorporating commonsense knowledge into open-domain dialogue systems. In this paper, we propose a novel knowledge-aware generation model (called TransDG), which transfers question representation and matching abilities base answering (KBQA) task to facilitate utterance understanding factual selection for generation. addition, response guiding attention multi-step decoding strategy steer our focus on relevant features Experiments two...

The large language models based on transformers have shown strong text generation ability. However, due to the need for significant computing resources, little work has been done generate emotional using such as GPT-2. To address this issue, authors proposed an affective prompt-tuning-based model (APT-LM) equipped with decoding (AD) method, aiming enhance limited resources. In detail, incorporates attributes into soft prompt by NRC emotion intensity lexicon and updates additional parameters...

Summary Item‐based collaborative filtering (CF) is a model‐based algorithm for making recommendations. In the algorithm, similarity between items are calculated by using number of measures, and then these values used to predict ratings users. However, if users grows millions, scalability processing efficiency item‐based CF can be hindered some hardware constraints. To solve this problem, we propose an optimized MapReduce integrated with empirical analysis. Through extensive experiments on...

Clustering is a classical research field due to its broad applications in data mining such as emotion detection, event extraction and topic discovery. It aims discover intrinsic patterns which can be formed clusters from collection of data. Significant progress have been made by the Density-based Spatial Applications with Noise (DBSCAN) variants. However, there major limitation that current density-based algorithms suffer linear connection problem, where they perform poorly discriminate...

Existing end-to-end task-oriented dialog systems struggle to dynamically model long context for interactions and effectively incorporate knowledge base (KB) information into generation. To conquer these limitations, we propose a Dual Dynamic Memory Network (DDMN) multi-turn generation, which maintains two core components: memory manager KB manager. The expands the turn by keeps track of history with an updating mechanism, encourages filter irrelevant memorize important newly coming...

Online social networks (OSNs) have received a lot of attentions recently since they provide new platform for product promotion and online viral marketing. Influence maximization problem has been extensively studied on some existing influence diffusion models in number domains. However, most the studies consider OSNs as friendly only containing friendship relationships, whereas hostile relations do exist many real life, e.g., Epinions Slashdot. In this paper, we integrate PageRank signed use...

Contact detection is a general problem of many physical simulations. This work presents O(N) multigrid method for contact problems (MGCD). The idea integrated with problems. Both the time complexity and memory consumption MGCD are O(N). Unlike other methods, whose efficiencies influenced strongly by object size distribution, performance insensitive to distribution. We compare no binary search (NBS) multilevel boxing in three dimensions both consumption. For objects similar size, as good NBS...

With the advancement of cloud computing, high energy consumption computing data centers has become a prominent problem. In this paper, we propose new virtual machine consolidation framework for achieving better efficiency. The proposed two main contributions: (1) underloaded host decision step, paper proposes method based on overload threshold hosts and average utilization all active hosts, which is named Improved Underload Decision (IUD) algorithm; (2) And in migration target selection puts...

Internet is becoming a spreading platform for the public opinion. It important to grasp opinion in time and understand trends of their correctly. Text classification plays fundamental role number information management retrieval tasks. But Web-page much more difficult than pure-text due large variety noisy embedded Web pages. In this paper, we propose system scheme analysis (IPO). We apply through summarization extract most relevant content from pages then pass them standard text algorithms...

Sleep monitoring typically requires the uncomfortable and expensive polysomnography (PSG) test to determine sleep stages. Body movement cardiopulmonary signals provide an alternative way perform staging. In recent years, long-short term memory (LSTM) networks convolutional neural (CNN) have dominated automatic staging due their better learning ability than machine classifiers. However, LSTM may lose information when dealing with long sequences, while CNN is not good at sequence modeling. As...

With the advent of era cloud computing, high energy consumption computing data centers has become a prominent problem, and how to reduce center improve efficiency research focus researchers all world. In environment, virtual machine consolidation (VMC) is an effective strategy that can efficiency. However, at same time, in process consolidation, we need deal with tradeoff between excellent service performance meet level agreement (SLA). this paper, propose new framework for achieving better...

Service integration and orchestration are important ways to make use of existing services build larger systems. In the real world, heterogeneous. The most popular SOAP/WSDL Web applications. heterogeneous bring challenges service orchestration. Prior attempts adapters or converters mask heterogeneity host alien services, which potentially lead inefficiency. This paper reviews current techniques for Comparing different techniques' strengths weaknesses, this presents a hybrid approach that...

In the discrete element method, damping coefficient, which is a key parameter for rational simulation, uncertain and often chosen based on researchers experience. To make physical model reflect reality better, in this paper acoustic technology used to calibrate coefficient. Oscillogram of wave generated by collisions analyzed calibrated coefficient obtained be 0.5. The parameters are simulate random packing process 2000 monosize spherical particles. final density 0.625, accordance with...

This paper reports a comparative study for medical text categorizations on four machine learning methods: k Nearest Neighbor (kNN), Support Vector Machines (SVM), Naïve Bayes (NB) and Clonal Selection Algorithm Based Antibody Density (CSABAD). CSABAD is an improved immune algorithm proposed by us. According to the clonal selection principle density control mechanism, only those cells that have higher affinity lower are selected proliferate. In addition, we propose approach, called Term...

With the prevalence of cloud storage, more and users store their files in storage. There are a large number duplicate which makes file deduplication important for saving storage space. After analyzing features files, this paper proposes new method based on differential bloom filter. Besides utilizing application aw chunking index similarity theory, filter is designed to accelerate speed lookups reduce total computational costs. The experiments demonstrate that proposed has good overa11...