If Ms. Hetrea was using the term /algorithm/ to indicate artificial computational analysis you may be right; and it helps clarify anyway. Yet seeing as she is a linguist I am not sure why she did that.

The term algorithm refers to a step-by-step problem-solving procedure, especially an established, recursive procedure for solving a problem in a small, fixed number of steps. Any other use of the term would be a neologism.

Also, an algorithm is not necessarily mathematical nor can it always be implemented on computers. Whether or not you are conscious of it, you often use an algorithm as a means to accomplish tasks in daily life Consider how you find your keys or how you mow your lawn.

It is “disinforming” to use the term algorithm to refer to *artificial” ways, e.g., Boolean keyword, Machine Learning (ML), and Bayesian methods of implementing full-text search and classification algorithms on computers. Why not just say artificial methods or call it artificial where it applies.

What I bring are formal methods of implementing established and natural algorithms that we know of, such as:

— learning (the processes of recognizing, assessing and inducing or deducing a resolution or understanding of something presented to the attention of the senses),
— writing (the iterative and recursive process of encoding the results of cognitive mechanisms that transform sounds and translate thoughts, language and experience into interpersonal and conventional symbolic representations) and;
– reading (text) (the processes of repeated apprehension and comprehension by means of recursively decoding the interpersonal and conventional symbolic representations).

These algorithms are not artificial, they are natural and they depend on underlying semantic mapping mechanisms. In my view these mechanisms are marked by their explanatory power.

As far as I see, no one is implementing natural algorithms or logical mechanisms like this. I have surveyed the literature and the modern technology and methods being used in search technology. I do not see texts being understood by natural algorithms that are formalized and implemented on computers.

People solve the problem of classifying a text by means of applying the processes of reading it, simultaneously decoding and learning about and comprehending interpersonal relationships while apprehending the topics, facts and entities encoded in the conventional symbolic representations. This is done in an established, step-by-step, sequential and recursive manner.

In other words, it is done by way of algorithm. Therefore the algorithm is the means to reading, to learning and to searching text about natural and interpersonal relationships of and relating to people and their experience. Implementing such algorithms in computer software is a matter of creating formal methods (usually mathematical or logical) that imitate the same mechanisms that underly human conventions (established and agreed upon procedures).

Findability in the world of online text may be the ability of the resource to be recognized and apprehended (being interpreted and appreciated) from a sea of noise and artificial (and meaningless) structures that are only impersonating relevant signs for private gain. It does not change the fact that finding any resource follows a step-by-step procedure of sometimes iterative search, selecting among results and further review.

Thank you all for your critics. I do apologize if I showed bias toward or against any of the fine products being discussed here while talking about the NLP methods and semantic search in particular.

I am biased against claims of semantic search where semantic acuity is absent and such absence is verified by a lack of suitable correspondence between the input and the result.

I suppose that anyone using RDF and other elements of the semantic web can call their methods semantic; even that is a pretty low bar. I just don’t see the semantics in the NLP methods used to map search terms onto parts of speech and indexed sentences.

This is a hard topic to address in a few comments so I will write a post on that and put it up a little later today or tomorrow.

-Ken Ewell