1887

oa Educor Multidisciplinary Journal - Classifying search results using neural networks and anomaly detection

Volume 2 Number 1
  • ISSN : 2520-4254
  • E-ISSN: 2663-2349

 

Abstract

Searching for something on the internet may not always yield relevant results due to the inability of current search engines to distinguish non-advertisement from advertisement. The purpose of this research was to detect and classify search results from search engines involving advertising or shop sales, so that they may be ignored. Artificial neural networks make use of multiple layers of nodes, receiving search result input at one end, transforming the features, and then outputting a hypothesis at the other end of whether the search result is an advertisement or not. In contrast, anomaly detection algorithms attempt to separate data into normal (non-advertisement) and anomalous (advertisement) data based on a specific parameter value. The problem being solved in this research was search results classification using machine learning algorithms (supervised learning). A manual method was implemented to collect 1GB of search results from the internet and process features, extract common words from results that were similar to advertisements, and then train algorithms to look for these features. The results from this experiment showed that both types of algorithms classified the data well. Both types of algorithms are thus viable for use in this scenario and can be implemented in tandem.

Loading full text...

Full text loading...

Loading

Article metrics loading...

/content/journal/10520/EJC-13d317b93a
2018-12-01
2020-02-27

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error