Attribute Relevance

Attribute Relevance

DOI: 10.4018/978-1-6684-4849-6.ch007
OnDemand:
(Individual Chapters)
Available
$37.50
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

A random forest study can help the investigator determine the degree of relevance of some attribute set that is under scrutiny, thus leaving the investigative study with a greater degree of insight into where ultimately that compass actually points towards. This chapter performs a random forest study with the data sets that were derived in previous chapters, and they are leveraged so that under a supervised learning model a degree of relevance over the attribute set may be ascertained. By following this process, it becomes feasible to determine the similitude that may exist with additional bodies of work and to ultimately be able to ascertain a goodness of fit for posterior models. This process of affirmation helps to guide the discussion to an ultimate state of truth, which is what is the only matter of concern for the scientist.
Chapter Preview
Top

Introduction

In the previous chapter an approximation formula to the Yahoo search engine was created, which resulted in an almost 80% parity with the search results that were derived from the search provider. What the previous chapter did not disclose however was the degree of significance of the independent variables to the dependent variable. In order to be able to determine the degree of significance of each element on the system it becomes imperative to be able to perform a different type of study, which is what is undertaken in this chapter. This degree of feature relevance may be determined through a random forest study which is facilitated through the statistical programming software package R. The previous chapter was a solid first step to understanding the indexing paradigm that is subscribed to by two of the major search engines, but it left a void in the focus with regards to the development of software for the web, this chapter seeks to close this gap by helping to shed light on how to quantify the indexing domain from a development context. While you may understand that link equity for example is relevant, how relevant is it compared to title tags? Where does the focus need to be when designing searchable applications? How exactly is HTML supposed to be structured to facilitate the indexing of the applications that are to be built? These are the questions that are sought to be answered in this chapter. What is sought in this chapter is the structure of web content for Micro Distributed Applications to enable a web presence to be enhanced and in doing so help the business win the indexing race. Ultimately what is sought here is to determine the focal point of web development; how specifically can MDAs be structured to find alignment with the search purveyors given their focus of the system attributes. What is currently understood between the relationship of indexing and development is the semblance of truth, the development community perceives the silhouette to be actually the person because of the outline but cannot identify the person from this shadow because truth is in fact hidden. This chapter is important because it leads the discussion forward to the identification of the architectural framework underpinnings by which development needs to adhere to in order to be able to create those small JavaScript components that will lead domains to win the indexing race. There can be no doubt that what has been done until now is significant as the disclosing of a supervised learning model allowed for the approximation of search engine relevance of one of the major search purveyors with an almost 80% degree of accuracy, but unfortunately this is not quite enough because it does not afford that blueprint that needs to be followed by development teams in general in order to be able to create systems of subsistence. These systems of relevance will only ever come to fruition if the silhouette of truth will simply not suffice.

Complete Chapter List

Search this Book:
Reset