آرش غفوری

Today, one of the most popular and challenging tasks in natural language processing is answering complex user questions. Question-answering systems, as a new generation of search engines, receive user questions in natural language without semantic limitations and provide precise answers. In recent years, most research in the field of question-answering systems has been focused on the English language, and not much effort has been made in languages with limited resources, such as Persian. This limitation prevents these systems from efficiently handling questions in languages like Persian.In this regard, this article aims to enhance the efficiency of question-answering systems in the Persian language by creating a dataset for answering complex multi-turn questions. Multi-hop questions, require at least two steps of reasoning to reach an answer. This dataset, called PersianMHQA, is the first open-domain question-answering dataset in Persian and includes 7,000 multi-hop questions. It was generated using the Persian Wikipedia as a knowledge source. To evaluate and benchmark this dataset, it has been fine-tuned on the latest pre-trained language models that support the Persian language.The best results obtained based on F1 score and exact match on this dataset are 75.92% and 71.73%, respectively. These results indicate that this dataset is a powerful starting point for improving multi-hop complex question-answering for Persian language systems.In this regard, this article aims to enhance the efficiency of question-answering systems in the Persian language by creating a dataset for answering complex multi-turn questions. Multi-hop questions, require at least two steps of reasoning to reach an answer. This dataset, called PersianMHQA, is the first open-domain question-answering dataset in Persian and includes 7,000 multi-hop questions. It was generated using the Persian Wikipedia as a knowledge source. To evaluate and benchmark this dataset, it has been fine-tuned on the latest pre-trained language models that support the Persian language.The best results obtained based on F1 score and exact match on this dataset are 75.92% and 71.73%, respectively. These results indicate that this dataset is a powerful starting point for improving multi-hop complex question-answering for Persian language systems.

Question-answering systems, as the next generation of search engines, have the capability to retrieve relevant answers to queries posed in natural language. These systems generally consist of three main components: question processing, information retrieval, and answer extraction, with various methods introduced for each component. One of the most important aspects is information retrieval and selecting relevant paragraphs. Nowadays, most user queries in question-answering systems are complex. To respond to such queries, it initially requires understanding the question and then retrieving various documents that are lexically and semantically related to the query. In recent years, advancements in deep neural network-based learning on one hand and the introduction of high-quality datasets on a large scale on the other have attracted researchers' attention to this field. In this research, a method for selecting relevant paragraphs for complex question-answering systems on the HotpotQA dataset is introduced. To select relevant paragraphs, the type of question is first recognized using a deep neural network. Then, using BERT language models, relevant paragraphs that show evidence of the answer are selected in several steps. This process uses keyword extraction from the question. The results obtained indicate that the outcomes are better compared to the baseline method

Question Answering Systems (QAS) are one of the most important intelligent systems that have the ability to provide instant and explicit answers to input questions. One of the new challenges of these systems is the ability to answer multi-hop complex questions that require information collection from multiple documents. In this paper, a method is proposed to solve the challenge of answering multi-hop complex questions. In this method, first, documents related to the question are retrieved in a two-step process. Then, to facilitate answering the question, knowledge is extracted from the retrieved documents and represented in the form of a graph. Finally, to find the answer to the question, reasoning is performed on the graph using a combination of heterogeneous graph neural network and transformer. To evaluate the effectiveness of the proposed method, this method and other related works have been tested on the HotpotQA open-domain dataset. The results obtained in answering questions based on F1 score and exact match are reported as 86.51 and 78.71 respectively, indicating superiority of this method over similar works.

In this study, the effect of limited inclination angle on two-phase flow in a vertical pipe was investigated, focusing on the formation of liquid film thickness using image recording and processing techniques. To generate annular flow patterns, non-homogeneous air-water flow was utilized in three transparent pipes with diameters ranging from 25 to 75 millimeters. After analyzing the flow pattern results for various superficial gas and liquid velocities, Reynolds numbers, and different pipe diameters, both the maximum and minimum liquid film thicknesses were calculated and presented, along with their ratios. Additionally, the dimensionless equivalent liquid film thicknesses for inclined pipes were provided concerning the uniform film thickness in the fully vertical condition. The results indicated an increase in the ratio of maximum to minimum film thickness with a change in inclination angle from 90 to 85 degrees. Furthermore, the dimensionless equivalent liquid film thickness decreased with inclination angle, consistent with prior research findings. The uncertainty analysis revealed a maximum uncertainty of 4.9 percent in the experimental assessment of the liquid film thickness, which falls within an acceptable range.

In the experimental study, the formation of the annular flow pattern in a vertical pipe with the counter-current two-phase flow has been investigated with the help of image recording and processing techniques. After separating the created two-phase flow regime, the range of superficial velocity of air (upward) and water (downward) is 3.66-20.94 m/s and 0.06-0.31m/s for annular flow, respectively. The interfacial friction factor (liquid and gas phase) has been evaluated according to the hydrodynamic parameters of the flow. Comparing the average deviation of the results obtained from the current research with the previous research shows that the results are in good agreement. Also, the interfacial shear stress has been calculated and evaluated for two test pipes in the center-current two-phase flow pattern in the annular flow regime. In this research, the interfacial friction factor (liquid and gas phase) is also presented as a new correlation depending on the Reynolds number of the gas flow and the Reynolds number of the liquid flow with a coefficient of determination of R2=0.98.

With the advent of big data technology as a service, small organizations can provide services they need in the case of big data through corporate services. This work reduces the resources required for pilot projects and reduces costs. Due to the layers of the big data framework as a service, leading companies each offer their services on some layers of this framework, and users can choose the service provider according to their needs. In this paper, eight important technologies that are presented as big data as a service by different companies were reviewed and compared with each other. An important issue to consider is the variety of services provided by different companies so that one can not judge the superiority of one service over other services, but the goal of service providers is to provide more diverse and demand-based services that users according to their needs Choose them.

In recent years, cloud computing is becoming eminent in the field of information technology. In a cloud computing environment, there is a potential for faults. There are different methods for dealing with faults, but with regard to the features and characteristics of the cloud computing environment, the use of fault tolerance methods is the best choice for this environment.One of the biggest issues in fault tolerance methods is the efficient use of resources. The optimal use of resources is important for cloud providers and customers. Unfortunately, the optimal use of resources in fault tolerance methods has not been much considered by researchers and cloud service providers. In this paper taking into account the dependence between tasks, an attempt has been made to provide a fault tolerance method on virtual machines, which in addition to being tolerant of fault, achieves optimum use of resources. In this method, by using a priority scheduler, each task is assigned a priority, then tasks are sent by the order of priority to their virtual  machines for processing. The results of simulation by the cloudsim simulator show that the proposed method has been able to improve the use of resources more than other methods and with 95% confidence intervals it has achieved (29.15% and 22.74%) improvement in the number of processors, (30.76% and 22.34%) improvement in memory usage and (29.71% and 22.88%) improvement in the use of bandwidth.

Enterprise IT infrastructures endure a wide range of costs including software and hardware costs. However, the advent of cloud computing environments have provided a tangible perspective for reduce some of these costs. Software as a Service (SaaS) is type of services that provided by the cloud. It provide a new paradigm, which goal is to provide an effective and intelligent way to support end users’ on-demand requirements to computing resources. In this regard, multitenancy is one of the most important features of SaaS tier that due to increasing rate of resource utilization and decreasing overall cost of IT infrastructure by IT organization have been considered. Because of the emerging nature of this concept and the advantages which is offered by it, study of characteristics, related concepts and its challenges are important. This paper first introduces Multi-tenancy, features and related concepts and then examine technical layers and variety of date management approaches in multi-tenancy architecture and finally offers multi-tenancy's challenges and existing solutions in order to resolve theirs challenges.

Human errors in design and configuration of networks and systems are potentials for attacks. Security Operation Center often used in wide networks, is a solution for continuous monitoring and detection, and human workers have key role in it. Through study of visualization subject and comparison between commercial samples of SOCs, this paper proposed a method that helping early detection in wide networks. The proposed method (MAPSA) is adding a cyber-attack real-time visualization module in SOC which SOC's analyzers may use it to early decide about modifications requirement in networks. This method leads to human error reduction, growth of personnel's effectiveness and increase in speed of modification. Therefore decreases the effects of attacks on wide networks.

Regarding the limitations of wired networks, such as geographical constraints, positioning limitations, high cost of building the necessary infrastructure, and construction lag of infrastructures etc, wireless networks have received considerable attention. In the meantime, adhoc mobile networks have received special attention due to the features such as lack of central management and infrastructure, as well as high mobility. A mobile network is a network independent of router nodes connected to each other by wireless links. In this network, routers move randomly and independently organize themselves; therefore, the wireless network topology is changing frequently and unpredictably, which has become one of the challenges of using these networks. If the routing protocol is not selected according to the application’s circumstances, it will cause many problems. In this paper, we explore mobile adhoc networks and then we will propose an optimal routing protocol for mobile military communications. This proposal has been presented with regard to the characteristics of a military connection. The simulation results demonetrate optimal performance of this protocol compared to other protocols.

Mobile AD hoc Networks as a special type of wireless networks have received special attention due to having some special features such as no need for central management, no need for infrastructure and high mobility capability. They can be used in cases where creating an effective communication infrastructure is neither cost-effective nor practically feasible, such as conferences, battles and post-disaster                   communications. Several routing protocols are proposed for these networks. ODMRP protocol is one of the most famous and widely used protocols in Mobile AD hoc networks. This study was carried out to discuss this routing protocol and aimed to provide a new routing method for increasing this protocol’s efficiency. In the ODMRP protocol, the optimal route is selected based on the shortest route. In wireless communications with the nodes getting away from each other, the received signal levels are weakened and may result in loss of data, and in practice, the shortest path that works based on a smaller number of hops loses its effectiveness. In the proposed protocol, the route is selected based on the received signal strength level.  According to the simulation results, control overhead decreases by %30 and packet delivery rate increases by %10, demonstrating performance improvement of the devised protocol compared to the original one.

Multi-Tenant software acts out based on auto-scale instance applications due to fluctuating workloads affected by increasing or decreasing the number of tenant demands and requested resource volumes. The aim of auto-scaling is better resource utilization of resource computing as well as provision of good Quality of Service (QoS) to tenants. Auto-scaling approach for meeting its used goal has challenges that consist of predicting workload, identifying correct resource requirement according to workload and selecting the policy of system for resource allocation with the lowest cost. In this research, a runtime auto-scaling mechanism is provided for multi-tenant software in order to answer mentioned challenges. First, due to lack of comprehensive description of a runtime system of multi-tenant software and based on found evidence of various studies, an architecture for runtime system of multi-tenant software is provided and a computing model is introduced that has three various granularity for its module of multi-tenant software. Then by providing an architecture for auto-scaling module of runtime system of multi-tenant software and using sub -modules that have appropriate functionality, mentioned challenges of auto-scaling are answered. To evaluate this, the proposed auto-scaling mechanism is implemented in the cloudism simulation environment and is tested and compared to the recent similar work in the same environmental conditions. The obtained results indicate that the provided auto-scaling mechanism compared to similar work despite the close number of violation of QoS, has better utilization and wastes computing resources less [13/28%-16/19%] with 99% confidence level.