New Evaluation Index for Application of Machine Learning Algorithms to Determine Trust in Skewed Social Media Data
By: Fazili, Shifaa Basharat.
Contributor(s): Ahmad, Manzoor.
Publisher: New Delhi STM Journals 2018Edition: Vol 5 (3), Sep - Dec.Description: 49 -57p.Subject(s): Computer EngineeringOnline resources: Click Here In: Journal of artificial intelligence research and advances (JoAIRA)Summary: paper we study the problem of accuracy paradox which arises on the application of machine learning algorithms for inference of trust in skewed social media data. Skewness is defined as the under representation of one class over another in a binary classification problem. We achieved our purpose of identifying the accuracy paradox problem in various algorithms by identifying a new evaluation index called predictive index. The dataset used was that of Twitter one of the most commonly used collaborative system which has experienced enormous growth in a small amount of time. It has evolved from a microblogging service to a major news source used by people as a platform to share and disseminate information about current events. However, not all information posted on Twitter is trustworthy or useful in providing information about the event. Gossips, fake news etc. are also a part of genuine news. The main aim of this paper is to tackle the issue of accuracy paradox, a major problem when dealing with social media research, were the data extracted by us was highly skewed. This high skewness in the dataset gives us biased information about the performance of our machine learning algorithms.Item type | Current location | Call number | Status | Date due | Barcode | Item holds |
---|---|---|---|---|---|---|
Articles Abstract Database | School of Engineering & Technology Archieval Section | Not for loan | 2021-2021403 |
paper we study the problem of accuracy paradox which arises on the application of machine learning algorithms for inference of trust in skewed social media data. Skewness is defined as the under representation of one class over another in a binary classification problem. We achieved our purpose of identifying the accuracy paradox problem in various algorithms by identifying a new evaluation index called predictive index. The dataset used was that of Twitter one of the most commonly used collaborative system which has experienced enormous growth in a small amount of time. It has evolved from a microblogging service to a major news source used by people as a platform to share and disseminate information about current events. However, not all information posted on Twitter is trustworthy or useful in providing information about the event. Gossips, fake news etc. are also a part of genuine news. The main aim of this paper is to tackle the issue of accuracy paradox, a major problem when dealing with social media research, were the data extracted by us was highly skewed. This high skewness in the dataset gives us biased information about the performance of our machine learning algorithms.
There are no comments for this item.