Thursday, July 24, 2008

Tiptoe Towards PR Meassurement

This weekend I anticipate being able to analyse all the pages of designated web sites using semantic analysis. The objective is to be able to identify values systems evident in the text in the form of semantic chunks.

Using this approach one can identify those elements of text (such as sentences) that are, by virtue of containing words (concept words) identified as having enhanced value by virtue of their strength of connectedness among the words in the corpus, of greater significance than others.

There are a number of approaches one can take. For example one can identify the strength of concept words by page or from the combined texts of all the pages.

My requirement is to be able to identify those concepts that are most connected throughout the web site to yield chunks of text ranked from most significant to least significant (and to identify the URL of the pages from which they are derived).

This is one of my approaches to help provide empirical proofs to support the Relationship Value Model .

The model posits that relationships are based on values shared between actors and semantic chunks of text have characteristics akin to values. For all intent and purposes, semantic chunks in web sites are expressions of values. They are not the complete set of because other elements such as design, site uptime, photographs, video and other images are also expressions of organisational (and personal) values.

This new development will take the hard work out of identifying the value systems inherent in a web site and is the first stepping stone towards being able to identify common values between organisations and actors.

I guess that a lot of people will be interested in the values their web site presents to the public (and those of competitors ) but this is only the first step in this journey.

Semantic concepts are valuable in other directions too.

Search engines use semantic analysis of web pages as part of their algorithms to match up search terms to web sites and an example of how this works is provided by Yahoo. Its 'Search Assist' service provides lists of semantic concept words to help people using its search engine.

Thus there is a commercial value in my research at an early stage. It can show people trying to optimise web site content how effectively their content has contributed to accessibility of their site to the public through both common values and search.

The next research aim using semantically derived values is the be able to compare the commonality of values as between different web sites. Thus one can combine the web page corpus of two sites to identify all the concepts for both and the extent to which there are common concepts, unique concepts and the relative significance (lets call it rank for the time being) of different concepts and their associated semantic chunks of text.

So far so good. But can this approach go further? Let us imagine comparing the values of an organisation as expressed through its web site (the place where more important visit most often) and the values expressed in, say, the media or blog posts or in social networks.

In theory, and we will be able to test this in a few weeks, we will be able to identify those values that these media have in common with an organisation and the values that are expressed that are unique to either the media in question or the organisation. This will offer a very powerful view of whether the message is 'getting through'.

The extent to which there is convergence and divergence is, surely, a test of how close the relationship is between the organisation and its stakeholders.

Is this a measure of the effectiveness of public relations as a whole?

It certainly has potential.