Showing papers by "Parmeet Kaur published in 2019"

A Machine Learning Approach for Enhancing Defence Against Global Terrorism

Abstract: The objective of this work is to predict the region and country of a terrorist attack using machine learning approaches. The work has been carried out upon the Global Terrorism Database (GTD), which is an open database containing list of terrorist activities from 1970 to 2017. Six machine learning algorithms have been applied on a selected set of features from the dataset to achieve an accuracy of up to 82%. The results suggest that it is possible to train machine learning models in order to predict the region and country of terrorist attack if certain parameters are known. It is postulated that the work can be used for enhancing security against terrorist attacks in the world.

Efficient Computation Offloading in Mobile Cloud Computing with Nature-Inspired Algorithms

Abstract: The ubiquitous presence of smart phones and other hand-held computing devices has resulted in a growing feasibility to utilize them as computing resources. However, these mobile devices are constra...

Efficient computation offloading using grey wolf optimization algorithm

Abstract: The widely available smart-phones as well as other mobile devices can be utilized as valuable compute resources with the support of the powerful and elastic paradigm of cloud computing. This is possible by performing computations of a limited scale on mobile devices and migrating or offloading complex services to the cloud. Use of mobile devices for computations will enable a plethora of applications to execute on mobile devices alone without dependence on static infrastructure. However, this entails the cost of transfer of computations to the cloud along with the cost of cloud computing resources. This paper formulates computation offloading as an optimization problem and utilizes nature-inspired approach of Grey Wolf Algorithm (GWO) to achieve near-optimal solutions. Results precisely depict that although the best cost solutions are attained by the brute force technique but the number of computations is significantly higher as compared to Grey Wolf Algorithm. Moreover with the small increase in number of tasks, there is exponential increase in number of computations. Considering these tradeoffs, its more appropriate to use nature-inspired algorithms for computation offloading in a mobile cloud computing environment.

Scheduling Data Intensive Scientific Workflows in Cloud Environment Using Nature Inspired Algorithms

Abstract: Workflows are a commonly used model to describe applications consisting of computational tasks with data or control flow dependencies. They are used in domains of bioinformatics, astronomy, physics, etc., for data-driven scientific applications. Execution of data-intensive workflow applications in a reasonable amount of time demands a high-performance computing environment. Cloud computing is a way of purchasing computing resources on demand through virtualization technologies. It provides the infrastructure to build and run workflow applications, which is called ‘Infrastructure as a Service.' However, it is necessary to schedule workflows on cloud in a way that reduces the cost of leasing resources. Scheduling tasks on resources is a NP hard problem and using meta-heuristic algorithms is an obvious choice for the same. This chapter presents application of nature-inspired algorithms: particle swarm optimization, shuffled frog leaping algorithm and grey wolf optimization algorithm to the workflow scheduling problem on the cloud. Simulation results prove the efficacy of the suggested algorithms.

A Multitenant Data Store Using a Column Based NoSQL Database

Abstract: As the world is frantically moving towards digitization, handling huge volumes of data is becoming complex and difficult to optimize. This is mainly because of increasing dimensions of data, its unstructuredness and increased need for storage space. To increase the efficiency of the cloud environment, many service providers have moved towards multi-tenant architecture. In a multi-tenant architecture, a number of individual applications termed as tenants work under a shared environment. In such a scenario, it is imperative that data of one tenant be isolated from that of other tenants. Further, it is also desirable to optimize the use of storage space. With these considerations, this paper proposes a multi-tenant database architecture for applications using a column-based NoSQL data store. The architecture has been implemented with the NoSQL database, Cassandra.

Analysis of Comments on Youtube Videos using Hadoop

Abstract: YouTube is a popular video sharing portal of today's time which also allows users to provide their feedback in form of comments, likes and subscriptions. Analysis of this feedback plays a crucial role in enhancement of video content to suit the need of the time. The comments, in specific, can highly benefit a new user in forming an informed and decisive opinion about a particular video. The work presents a system to perform a detailed analysis and classification of the comments on YouTube videos. The comments corresponding to videos have been ingested into the Hadoop Distributed File System (HDFS) and queried with the Hadoop analytical software, Hive. Further, sentiment analysis of these comments has been performed using Python. The proposed system has been evaluated by executing multiple self-designed queries on the YouTube data. Results of execution times of these queries have been tabulated and presented in the form of graphs. Further, results of sentiment analysis of comments that determine the reaction to the video, which are encouraging.

Analyzing IPL dataset with MongoDB

Abstract: Development has emerged use of NoSQL databases for storage and retrieval of big data. A highly popular NoSQL database is the document-oriented database, MongoDB which has superseded the relational databases in numerous applications. This paper investigates the querying performance of MongoDB using a dataset from the popular Indian cricket league, IPL (Indian Premier League). The present work analyzes the IPL dataset using MongoDB queries for determining the attributes that are important for a team in order to win a match. The experimental results have shown that for considered dataset, it was observed from the performed analysis that the parameters of winning a toss, playing at home grounds and application of Duckworth Lewis rule plays an important role in determining the outcome of a match.

Dependency Graph Based Detection of Semantically Equivalent Questions in Online Forums

Abstract: In the present, people increasingly rely on online user forums for clearing their doubts and seeking answers to varied questions. On many online forums, users encounter a similar question asked in different formats. This results in confusion for the user and he may not be able to find appropriate answer to his question even though the appropriate answer exists on some other page, i.e., on the page resultant of a differently formed question. Currently, online forums like Quora only give suggestions to the user about the questions he could ask but do not show all the semantically equivalent questions. This article eases the work of the users searching for answers on online user forums. The proposed technique will allow the users to look at all the existing semantically equivalent questions. A dependency graph-based matching algorithm is applied to accomplish this objective.

News Data Analysis from Facebook Through MongoDB and Hive

Abstract: The research aims at doing a comparative study for news data analysis through MongoDB and Hive. The news posts or feeds are gathered from the official Facebook page of The Times of India with the help of Facebook Graph API. The data is stored in a NoSQL database, MongoDB and the Hive data warehouse of Hadoop ecosystem. For better data handling and quick query processing, the data is sharded (partitioned) in MongoDB as well as partitioned in Hive using the date field as the sharding key and the partitioning key in the respective databases. A comparative study is done between the two platforms by running various queries on the data and comparing the execution time. The queries aim to search the news posts on the basis of date ranges, particular keywords in the heading or through any other field like maximum laugh reactions. The study was done on a single machine system for both MongoDB and Hive. The results indicate less execution time for all queries with MongoDB as compared to Hive.

A Distributed Storage System Using a Mobile Ad hoc Network: Distributed Storage System

Abstract: The Ambient Assisted Living (AAL) domain aims to support the daily life activities of elders, patients with chronic conditions, and disabled people. Several AAL platforms have been developed over the last two decades. Hence, there is a need to identify Quality Criteria (QC) and make it well defined in order to achieve the AAL system purposes. To be able to convince all stakeholders including both technologies and end users of AAL systems, high quality must be guaranteed. The goal of this article is to obtain a set of data quality characteristics that would be applicable to AAL system, and have its performance evaluated using sensor data. To this end, this work uses the ISO/IEC 25012 and ISO/IEC 25010 standards to extract the most relevant criteria that are apt for AAL systems. As a result, an evaluation approach on an indoor localization platform was made, and an evaluation procedure has been established. This is done by first generating a hierarchical data quality model, and have it evaluated using the metrics, based on the sensor data and the concept of fuzzy logic.