The Google Myth - LSI Revealed

Thousands of web documents, even sites across the web, have been dedicated to publishing content on the concept of Latent Semantic Indexing (LSI) as part of Google’s large and complex algorithm. Many of whom have been lead to believe that LSI, an advanced search engine ranking technique, was [a] core variable to determine rankings in Google’s natural search results. For a while, at least in the search engine community, LSI was a prominent buzz word (or shall we say acronym) amongst experts and professionals alike, however a new buzz word is quickly emerging to overshadow LSI’s popularity. For some it may shock and for many it will shake the way best practice SEO is executed.

There are plenty of web documents available for your viewing if you wish to learn more about LSI. If you like, Wikipedia’s Latent Semantic Analysis (LSA) provides a highly relevant yet thorough explanation of what is involved in the LSI process and how such a formula analyses web documents to match up with a particular search query. To explain to you what LSI really is, would go far beyond the intent and purpose of this paper. However, as part of this paper to encourage you to seek truth and relevance to LSI and its association (or non-association) with Google’s algorithm, for your benefit we will briefly discuss what LSI is.

According to Professor Thomas Hofman from the International Computer Science Institute at Berkley University California; LSI is an automated approach to document indexing through Singular Value Decomposition (SVD). His research on ‘Probabilistic Latent Semantic Indexing’ (PLSA) also suggested that, although LSI has been applied with remarkable success – “it has a number of deficits”. Other researchers; Thomas K. Landauer and Darrell Laham from University of Colorado and Peter W. Foltz from New Mexico State University agree that LSI, to a certain extent, is a highly effective and efficient system to “correctly match queries to (an only to) documents of similar topical meaning when query and document use different words”. Their research ‘An introduction to Latent Semantic Analysis’ provides a highly comprehensive analysis of how statistical computations decode and determine the contextual usage and meanings of words.

If you decide to read further into the above papers you should have a more in-depth understanding of LSI principals and how they are formulated to create a system of information retrieval. Based on these findings and Google’s track record for providing its users the most relevant search results, it’s no wonder many people (including experts and professionals) in the search engine community were commonly lead to think and assume LSI mechanisms were heavily involved in Google’s ranking algorithm.
So if Google’s not using LSI what are they using? How are their search results determined? Until recently it’s safe to assume no one outside of Google actually knew. That is, the best answer people could give was ‘LSI’; however faculty member and search engine expert Leslie Rohde from StomperNet.com, based on his research and testing, has come up with a proven solution. He suggests Google is not using LSI but rather what he calls “Referential Integrity (RI)”.

Get the full latent semantic index article at amanticode.com

It’s been said, that to be good at a system you need to know how the system works. The concept of Referential Integrity is a core mechanism of Google’s ranking algorithm. It’s advanced, it’s sophisticated and here’s how RI works and how you can improve your search engine rankings:

[…] To read the full article go to amanticode.com home of advanced search engine optimisation and internet marketing techniques