Social Media Data Extraction Method Benchmarking Comparison
Issue:
Volume 5, Issue 2, June 2019
Pages:
40-44
Received:
7 April 2019
Accepted:
13 August 2019
Published:
28 August 2019
Abstract: Social media has become more and more widely used nowadays. As the most popular media, a lot of information spread through Twitter, especially given the fact that U.S. President Trump has used Twitter as his main official free news publication outlet. Therefore, social media platforms like Twitter have become the important sources to extract information and then the information could be further analyzed through text analytics models for decision-making problems. In this paper, we first investigate several text analytics methods and then multiple tweets retrieving methods/software will be investigated: Twitter Analytics, Application for Twitter, Python plus Tweepy, and Next Analytics. Seven criteria related to features are applied to compare the methods for ease of use, extraction timing and capability to accommodate big data. Given that our results may be approximate because we might not be able to observe all the capability and features of the software, our results show that Python plus Tweepy method is the most ideal one when applying to big data projects (millions of tweets or above) and real time text data extraction. Next Analytics is the software that could retrieve historical text message in a more convenient way through Excel and is able to trace back further in time period, which could give much better capabilities in social media analysis.
Abstract: Social media has become more and more widely used nowadays. As the most popular media, a lot of information spread through Twitter, especially given the fact that U.S. President Trump has used Twitter as his main official free news publication outlet. Therefore, social media platforms like Twitter have become the important sources to extract inform...
Show More
Effect of Faults on Kalman Filter of State Vectors in Linear Systems
Issue:
Volume 5, Issue 2, June 2019
Pages:
45-56
Received:
25 July 2019
Accepted:
14 August 2019
Published:
28 August 2019
Abstract: Kalman filter (KF) is composed of a set of recursion algorithms which can be used to estimate the optimal state of the linear system, and widely used in the control system, signal processing and other fields. In the practical application of the KF, it is an unavoidable problem that how faults or anomalies are infectious to the estimation value of state vectors in the linear system, which must be paid much attention to and solved down. In this paper, the effect of sensor faults and control input anomalies on the Kalman filtering values of state vectors is discussed, the transmission relationship is established to analyze the estimation deviation of state vectors which comes from pulse or step faults/anomalies, and a sufficient condition is deduced for the convergence of the estimation deviation of state vectors; Four different system models with 3-dimension state vector and 2-dimension observation vector are selected for simulation calculation and comparative analysis, simulation results show that sensor faults and control input anomalies in linear systems may cause significant deviations in the estimation value of state vectors for a long time, and there are distinct differences in the estimation value of state vectors. The research results provide a certain theoretical reference for us to analyze system fault types and to identify fault.
Abstract: Kalman filter (KF) is composed of a set of recursion algorithms which can be used to estimate the optimal state of the linear system, and widely used in the control system, signal processing and other fields. In the practical application of the KF, it is an unavoidable problem that how faults or anomalies are infectious to the estimation value of s...
Show More
