Building a Semantic Relations Index

Creating high quality content that provides valuable information without distraction from or dilution of the intended subject requires a unique set of skills that are supported by knowledge and experience.

The search optimization professional is faced with an additional dimension of complexity in that (s)he must produce text that the search engines will find to be just as valuable. This is complicated by the fact that search engines simply don't read like humans do. In fact by the time the search engines are actually "reading" your text, it doesn't much resemble your original text at all; it's just a string of simple word tokens which would be confusing to a human reader.

The search engine is faced with the task of determining the subject, sense and context of text information, without the benefit of subjective interpretation. To do this the search engine employs a variety of heuristics and language disambiguation technologies. A significant part of the text interpretation process is concerned with measures of relevancy based upon informational integrity and the level of authoritativeness that can be inferred from the writing style of the author. Words are studied for their context within surrounding words and evaluated against statistical data related to the topical community which the text (webpage) belongs to.

As an example if we consider two distinct fields of study (say topics A and B.). Topics A and B are unrelated fields which share no direct common relationship. So let's say that a simple library for each topic exists (like a corporate knowledgebase). Each library contains the most important written information related to either field. It can be said that each library in its totality contains important information and statistical data about writing styles, common word usage and context, etc. that are unique to that topic. Furthermore this information could be used to effectively discriminate between high and low quality works within the field as well as help to detect low level authorship.

Search engines employ a similar method of analysis by comparing a given webpage's text against information gained by studying a vast corpus of related works.

It is therefore essential when writing optimized text for a website to work within the unique vocabulary associated with the topical community which the website belongs to.

When it comes to optimized text copy; focus is crucial. A semantic word relation index can be employed as a tremendous tool for developing the core vocabulary for a web page's optimized copy. A semantic word relation index is basically a map of technical linguistic and syntactic word relations built around your targeted search phrases. To develop a semantic word relations index you need 3 things:

I would hope that number 1 was satisfied long before any optimized text development began. Reliable language resources are freely available on the Internet; my personal preference is to use Princeton's WordNet and Google's advance search operators when compiling a relations index. The last item is furnished below. The following relationship list is based upon the CIRCA semantic space architecture.

The technique is simple enough and involves working your way down the list by researching words that meet each of the relationship type definitions in relation to your targeted search phrases. For example if my targeted search phrase was "Alternative Health" my semantic relations index might look like this:

Synonymy Antonymy
Homeopathy Allopathy

Similarity
Homeopathic

Hypernymy
Herbal Medicine

Try to build as many terms for each relation type as you can. The completed index can serve as a core pool of words to try to include contextually in your website's text. Of course text quality always assumes importance over usage of specific words. It has been my experience that the semantic relations index is best consulted as a thesaurus when writing for a particular subject, in this way I am able to improve word diversity while simultaneously improving topical strength.