Data Mining, Custom Research, Industry and Company Information Listings

An aspect of free information which enables huge advantages for small business, academic research and web development in a distributed, agile world exists in coordinating reference databases and deep analysis.

techBio is a resource to help develop your data context in any information domain.

Current Projects at techBio

Update:
New Projects in 'The Works':
The Yummy List
BoatBoss Boating, Marine, Watersports Portal
Kooky Coconut, Cafe, Indian Rocks Beach, FL

---

We are working diligently on:
Snapspans
Code|Dub
The Empty Fridge

While others languish in stability:
Budget Blinds of Southern Minnesota, Bob Cotton Owner
techBio Development Group
techBio:wiki
C. Moore Fine Art
Memagon: The Shape of Memes and Memory

techBio Quietly Knows About SEO and SEM

Caveat Emptor: Neither I nor techBio are practicing search optimization or marketing specialists providing such services.

Search engine optimization is described elsewhere. Search engine marketing is a separate but typically conflated set of services.

Two unsupported conclusions come to mind.

---

1. As major search engines tune their algorithms to provide useful, relevant and authoritative results, the only realistic long play strategy for SEO and SEM is to provide a concise and communicative website, with simple close navigation to concise and informative webpages.

My assumptions, without any special knowledge, are that textual content* is analyzed for keyword prevalence (addressed by keyword optimization) by a simple, deceptively powerful heuristic.

a) key words and terms are collected by matching against a less-common-words dictionary, and scored by relative universal-usage uniqueness (inverse of word frequency in text containing that word)

b) score for each keyterm is weighted proportionally to position in document, with more points added for position in heading text such as title, url, h1, h2,h3, position in a list, emphasis, inverse frequency throughout domain and keyword/text ratio is scaled by this proportion

c) scores for ranked and analyzed sites and pages linking in are summed and multiplied by each keyterm, resulting in an ordered list of keywords by computed score

d) the summed value of keyterms is page is scaled by down a readability analysis score

Everything else on the page will be all but irrelevent with respect to content and design optimization. The last entry (d) in the above list checks for reasonably natural prose, loosely defined as normal distribution of words and phrases and grammatical correctness.

---

2) Web sites that have built some traffic will be acquired by companies with deep pockets, and their links, traffic, data, content and users merged into larger corporation's domains and services. These sites will be bought at purely economical valuations. Sites which are well developed and run smoothly, provide simple and useful tools and content, and are organically search optimized will be valued far higher than guerilla coded and jungle optimized websites.

Anyway, that is what I am betting my time and effort on.

---

* Assuming for text content, in order of priority: title, url, included URLs, remote incoming links and inline local links, and all page text between markup

Arrived at the party

Technorati Profile

TechBio is spending lots of time and experiment with passive revenue from Google AdSense, affiliate referrals and long tail development through Snapspans.

It is a test bed for consulting work in the present and near future.

Website As Application Service Provider and Smart Access Point

So, a website is what serves webpages, right? Yes. And WAP, RSS, XML-RPC, URN-database-requests, SOAP, REST, API calls.

Oh.
And Web apps.
Oh.
How about all of these? Uh-oh.
I am pondering the flexibility of websites in light of:
http://go-pear.org
http://cheat.errtheblog.org

Mass marketing is not for computers. Trust me.

I wrote a script which:

- downloads classed webpages
- counts unique words on each page
- counts by class and site
- pretty good approximation for textdiff
- outputs weighted rankings

I need to see if the generated list is a good classifier of pages by class.

I would like a statistician to develop a metric of this performance.

A brain cell is the same as the universe

First, one may need speech word recognition manifold hashtables.

Secondly, an association. with an arbitrary file: image, URL, text document.
Thirdly a source file database.

The system presents files for commentary--these files may have an existing cross-reference via tags on a site such as flickr, digg, del.icio.us, Google Earth, Wikipedia, or a statistical correlation with other sources in the database, such as text files with statistically unlikely phrases, markup structures in SGML, color palette or file compression signature.

The result of this becomes not only a richer tag set, of varied files, but the transverse intelligence to display photographic examples alongside text, maps and images alongside a history, product illustrations alongside VOIP conversations or the transverse navigation interaction whereby mentioning a keyword brings onscreen files containing that and other keywords--a picture of a cat and a bird for instance--when the person describes an image of a cat, and this predator prey image appears, the person's next utterance, let's say this is "bird"--brings up pictures of birds, or the range of the ospreys seasonal migrations

Much of this data can be drawn from fairly open cheap or free sources, such as Wikipedia, Flickr, Google or Yahoo image search, maps and other apis.

The graphical visualization and the speech recognition should be sourced to technical or academic APIs and algorithms.