-----Original Message-----
From: John O'Gorman [mailto:jogorman@...]
Sent: Monday, November 23, 2009 07:35 AM
To: TaxoCoP@yahoogroups.com
Subject: [TaxoCoP] Taxonomies and Relational Databases

Good morning;   I once heard the relational theory of data referred to a 'set-based', and since taxonomists' bread and butter could be thought of as the formal division of things into set and subsets, I was wondering if anyone had any thoughts on a November morning about possible connections between the two.   John O'

Hi. I just joined, so this is the requested self-introduction. I'm presently
sitting at the TBC workshop in San Jose.

I'm the Director of Linguistics for OpenAmplify (http://www.openamplify.com), a
small text analytics / semantic extraction platform vendor. We work especially
in the ad-placement space, and our products are especially strong in natural
language processing and linguistic analysis. Our products report topics,
polarity/sentiment, named entities, and the like, but also more actionable data
such as an author's expressions of intention toward various topics, time frames,
and more.

My academic background is in Computer Science and Linguistics. I am perhaps best
known for developing the methods for part-of-speech assignment that everybody
now uses (based on corpus analysis, Markov modeling, etc.) and for my work on
electronic information standards. I've been heavily involved in W3C, OASIS, ISO,
and other standards organizations, on XML, XPath, XLink, TEI, the Encoded
Archival Description standard familiar to librarians, and many others, often as
editor.

In 1989 I co-founded Electronic Book Technologies and created an SGML-based
browser and structured search engine for delivery of very large-scale
documentation (customers like Boeing, SGI, Sun, DEC, etc). One of my favorite
anecdotes is that when the US Navy deployed our "DynaText" system to replace the
printed maintenance manuals on destroyers, the ships rode 6 inches higher in the
water from the saved weight of paper.

Currently, I'm especially interested in crowd-sourcing, construction of
large-scale practical taxonomies and ontologies, and anything related.

Steve DeRose

Hi all,
For those of you in San Jose for TBC, meet after the SharePoint talk I'll be
giving today (Thursday) with Jeff Carr and we can all go out for a drink!
See you tonight,
S.

Hello,
If you're attending the Taxonomy Boot Camp this week, stop by the SLA Taxonomy
Division table in the hall just outside the sessions and pick up information.

The Taxonomy Division addresses ways to organize and structure information so
that content is accessible and useful. It offers a practical context for
exploring issues and sharing experiences related to planning, creating and
maintaining taxonomies, thesauri, authority files, and other controlled
vocabularies and information structures.

If you're not going to the conference, you can get more information from
http://wiki.sla.org/display/SLATAX or information on joining SLA at
http://www.sla.org/content/membership/joinsla/index.cfm

Thanks,
Jennifer Borrell
PR Committee Chair
SLA Taxonomy Division

Heather -

A Mac application that I like is DEVONthink (www.devon-technologies.com) for
$150.

It produces a "concordance" (word count & frequencies) from a body of text.

I've never tried for phrases, but will spend some time in Help to see if it does
that
too.

- David

Hi all,

Let us know if you're going to be at TBC this year - would love to organize a
group dinner or drinks...

Jeff, Seth and I are doing a whopping 6 prsentations - hope to see you there!

Stephanie
Earley & Associates

One challenge I have seen with these tools is related to performance. Sampling
large amounts of content can bog down systems and take inordinate amounts of
time to process.

Our colleague Prakash Govindarajulu has created a tool to do this sampling
called Disk Bee. I am sure he would welcome opportunities to get feedback. He
can be reached at Prakash@...

Seth

Seth Earley
President
Earley & Associates
781-820-8080
www.earley.com

Sent from my Blackberry

________________________________
From: TaxoCoP@yahoogroups.com <TaxoCoP@yahoogroups.com>
To: TaxoCoP@yahoogroups.com <TaxoCoP@yahoogroups.com>
Sent: Wed Nov 11 12:36:48 2009
Subject: Re: [TaxoCoP] word/phrase extracting/sorting tools



Collecting basic statistics on individual words should be utterly trivial for
any competent programmer (assuming it's easy to access the content).  It could
be done in a scripting language (perl, python, etc.), a statistics language (R),
or a regular programming language (Java, C#, etc.). Maybe even in Visual Basic.
There are also free packages available online for this sort of thing.

Collecting phrases (n-grams) is a bit harder (and many of the free toys don't do
it), because you don't want to look at the raw frequency of n-grams (in which
case things like "of the" are the most common) but their frequency relative to
what you'd expect from sequences of random words (divide the frequency of the
n-gram by the product of the frequencies of the component words).  You will
still get "of the" near the top, but this will filter out many noise cases.

Though it is hard to do this accurately across a variety of subtle cases: is
building the same word as buildings? is mine the same word as mining? is
Coca-Cola one word or two? is Carpenter (presumably someone's name) the same
word as carpenter (presumably a job description)? does Santa Fe refer to the
city or to the railroad? etc.).  In fact, it covers a lot of the problem areas
of computational linguistics, but you don't need that level of sophistication. 
In fact, you're probably OK treating all non-alphanumeric characters as word
boundaries.

There are also suites of computational linguistic tools (many of them free) that
will do all this and more, but many of them will be too complicated to install
and learn for such a simple task.

Good luck!

                -s

On Wed, Nov 11, 2009 at 9:34 AM, Heather Hedden
<heather@...<mailto:heather@...>> wrote:


Does anyone have recommendations for simple/inexpensive tools that a taxonomist
could use to automatically gather words and phrases from content and sort or
count them (then to generate histograms, for example), as a method to gather
candidate terms from text when building a taxonomy? A colleague mentioned Word
List from MechanicWords http://www.mechanicwords.com/ but I'd like to know of
others like it.  This would be an automated supplement to a content audit. 
Also, is this something simple enough to code that any in-house software
developers in a large enterprise (with time resources) might be able to code
for?

Thanks.

-- Heather

--
Heather Hedden
Hedden Information Management
www.Hedden-Information.com<http://www.Hedden-Information.com>

-- Heather Hedden
Hedden Information Management
www.Hedden-Information.com

-----Original Message-----
From: Nick Berry [mailto:infoglutton@...]
Sent: Thursday, November 5, 2009 09:36 AM
To: TaxoCoP@yahoogroups.com
Subject: Re: [TaxoCoP] Re: Most common facets used within organisations

These are not just facets - they are also master data elements that typically end up in data warehouses for business intelligence purposes.  Right now we're spearheading an MDM initiative, and our top "facets" or entities are: Customer (attributes include name, address/location, activities, purchases, member/nonmember, etc) Article (which is the same as Product - attributes include size, color, price, fabric/material, brand, gender, seasonality, status, etc...but an Article could be a rain jacket or an Adventure Trip or a ski rental) Site (refers to stores, channels, distribution centers, trailheads, etc; attributes include name, location, type, etc) Vendor (includes manufacturers, distributors, etc; attributes are name, address, associated Articles, etc) We are applying standard IA principles to these elements - interviewing users to understand their needs, doing inventories of the data and how it is used, setting up taxonomies to manage relationships between facets and attributes, etc.  Except in the MDM world these tasks are called data profiling, data integration, data quality, data governance, etc.   If there are other people on this list doing MDM and/or BI, I'd love to chat with you about data governance, org structure, tools, etc.  Cheers, Nick nick berry, content architect & taxonomist Recreational Equipment, Inc. (REI) 253-437-7860 (p) / 253-395-8201 (f) On Thu, Nov 5, 2009 at 12:49 AM, stephaniefkmg <stephanie@earley.com> wrote:   This has been a fascinating discussion, weaving between business uses and theoretical concepts. I'll be summarizing the main points in a blog post next week and on the TaxoCoP wiki, quoting from this discussion. If any of the participants prefer not being quoted, please email me at stephanie@earley.com Thanks! S.

This has been a fascinating discussion, weaving between business uses and
theoretical concepts.

I'll be summarizing the main points in a blog post next week and on the TaxoCoP
wiki, quoting from this discussion.

If any of the participants prefer not being quoted, please email me at
stephanie@...

Thanks!
S.

On Wed, 4 Nov 2009 at 10:01:31, Gabriel Tanase <gabtanase@...>
wrote
>Leonard, all,
>
>In the data(base) conceptual modeling arena <Party> is the most
>frequently used term, as a super-type of both Person and Organisation.
>See Silverston's books on Data Model Patterns. Furthermore, the concept
>of <Party Role> is closely associated with <Party>.
>
>E.g. a "Customer" is a <Party Role>, i.e. the role that a <Person>
>party (Joe SixPack) plays in the context of an <Company> organization
>party (ACME Inc., whose customer Joe SixPack becomes at a given point
>in time). The role of <Customer> may or may not depend on the existence
>of a legal arrangement (contract) between the company Party and the
>person Party. A <Party Role> is always related to a <Party> that plays
>that role in a specific context of at least one other <Party>.
>
>Admittedly, as a term <Party> is not much better than <Agent> or
><Actor>. The conceptual data models make heavy use of subtyping under
>both <Party> and <Party Role> to represent the usual business terms of
>Customer, Supplier, Service Provider, Beneficiary, Company,
>Not-for-Profit Organization, Household etc. etc.
>
>
>Regards,
>Gabriel

Gabriel -

Thanks. I have come across this use of <party>, and perhaps it is the
best term we can find for the job - so long as it is clear to everybody
what it means.

I see that the CIDOC relational data model, at

<http://cidoc.mediahost.org/data-model%28en%29%28E73%29.xml>

uses PEOPLE as the overall type, with

PEOPLE-PERSON
and
PEOPLE-GROUP

for individuals and groups respectively. These are rather cumbersome,
but I suppose that GROUP is more general than "organisation" as a group
may not be "organised".

<Party role> is more difficult, because we have to decide whether to
treat it as a coordination or link between separate PARTY and ROLE
facets,

e.g. "PARTY" & "ROLE"

or whether to treat ROLE as a characteristic of division, giving an
array within the PARTY facet introduced by the node label

<Party by role>

In neither case can it be used as a simple element of a citation order
formula for expressing compound subjects, because its location in such a
sequence will depend on the nature of the role.

Leonard

--
Willpower Information     (Partners: Dr Leonard D Will, Sheena E Will)
Information Management Consultants            Tel: +44 (0)20 8372 0092
27 Calshot Way                              L.Will@...
ENFIELD                                Sheena.Will@...
EN2 7BQ, UK                            http://www.willpowerinfo.co.uk/

-----Original Message-----
From: Leonard Will [mailto:L.Will@...]
Sent: Tuesday, November 3, 2009 04:00 AM
To: TaxoCoP@yahoogroups.com
Subject: Re: [TaxoCoP] Most common facets used within organisations

On Mon, 2 Nov 2009 at 16:23:51, Matt Moore <innotecture@yahoo.com> wrote >Leonard, > >What would appear on your list? > >Cheers, > >Matt I don't think that it is possible to have a "definitive" list that suits all circumstances. The references Patrick Lambe has given discuss the reasons. Two issues are sometimes confused when discussing facets: 1. The need to distinguish between * "facets" (sometimes called "fundamental facets"), which are the mutually exclusive categories we are talking about here, such as objects, materials, activities, places, abstract concepts, and so on, and * "characteristics of division", such as "colour" or "age" which are used to subdivide a facet into _arrays_, headed by a "node label" such as "automobiles by colour" or "people by age"; these would be part of the "objects" or "people" facets respectively. 2. The fact that in specifying a citation order of facets in a compound subject you may have to list a single facet more than once, depending on its role. For example, in the CRG / BC2 list quoted in William Denton's paper * thing/entity * kind * part * property * material * process * operation * patient * product * by-product * agent * space * time terms from the "people" facet may be used as both "patient" and "agent", and terms from the "object" facet may be used as "thing", "product", "by-product", and so on. This makes is difficult to specify a consistent combination order which can be mechanised based solely on the names of facets - the entries in the above list are not really names of facets in the sense defined in 1 above.. This issue is not so much of a problem in post-coordinate applications, but as this list is talking about "taxonomies" (i.e. classification schemes), rather than thesauri, I presume that some pre-coordination will often be required. 3. It is difficult to find a term to name a facet to which people and organisations both belong: "agent" can give the misleading impression that it needs to be the pro-active member in a process; "actor" is liable to misinterpretation. Has anyone found a good term for this? Leonard -- Willpower Information (Partners: Dr Leonard D Will, Sheena E Will) Information Management Consultants Tel: +44 (0)20 8372 0092 27 Calshot Way L.Will@Willpowerinfo.co.uk ENFIELD Sheena.Will@Willpowerinfo.co.uk EN2 7BQ, UK http://www.willpowerinfo.co.uk/

On Mon, 2 Nov 2009 at 16:23:51, Matt Moore <innotecture@...> wrote
>Leonard,
>
>What would appear on your list?
>
>Cheers,
>
>Matt

I don't think that it is possible to have a "definitive" list that suits
all circumstances. The references Patrick Lambe has given discuss the
reasons.

Two issues are sometimes confused when discussing facets:

1. The need to distinguish between

*  "facets" (sometimes called "fundamental facets"), which are the
mutually exclusive categories we are talking about here, such as
objects, materials, activities, places, abstract concepts, and so on,

and

* "characteristics of division", such as "colour" or "age" which are
used to subdivide a facet into _arrays_, headed by a "node label" such
as "automobiles by colour" or "people by age"; these would be part of
the "objects" or "people" facets respectively.


2. The fact that in specifying a citation order of facets in a compound
subject you may have to list a single facet more than once, depending on
its role. For example, in the CRG / BC2 list quoted in William Denton's
paper

      * thing/entity
      * kind
      * part
      * property
      * material
      * process
      * operation
      * patient
      * product
      * by-product
      * agent
      * space
      * time

terms from the "people" facet may be used as both "patient" and "agent",
and terms from the "object" facet may be used as "thing", "product",
"by-product", and so on. This makes is difficult to specify a consistent
combination order which can be mechanised based solely on the names of
facets - the entries in the above list are not really names of facets in
the sense defined in 1 above..

This issue is not so much of a problem in post-coordinate applications,
but as this list is talking about "taxonomies" (i.e. classification
schemes), rather than thesauri, I presume that some pre-coordination
will often be required.

3. It is difficult to find a term to name a facet to which people and
organisations both belong: "agent" can give the misleading impression
that it needs to be the pro-active member in a process; "actor" is
liable to misinterpretation. Has anyone found a good term for this?

Leonard

--
Willpower Information     (Partners: Dr Leonard D Will, Sheena E Will)
Information Management Consultants            Tel: +44 (0)20 8372 0092
27 Calshot Way                              L.Will@...
ENFIELD                                Sheena.Will@...
EN2 7BQ, UK                            http://www.willpowerinfo.co.uk/

John,

I think we'll wait for the white paper. My initial comment is that while your
approach seems quite rigorous it also seems a little divorced from the language
that people use in businesses - no one talks about agents (unless they work for
the CIA) but they do talk aboutcustomers, clients, suppliers, and partners.

Cheers,

Matt



________________________________
From: John O'Gorman <jogorman@...>
To: TaxoCoP@yahoogroups.com
Sent: Tue, November 3, 2009 8:31:38 AM
Subject: Re:  [TaxoCoP] Most common facets used within organisations


Hi Stavros;

As a method for managing collections, Q6 is based on Ranganathan' s system of
classification. Interestingly, it also has a loose resemblence to the Zachman
Framework for enterprise architecture, but was independently derived.

The difference with Q6 is in its granularity and its geometry. In addition to
using the concepts of colon classification, the principles of Q6 start at word
level and can be extended right up through any digital asset. Photos, documents,
diagrams and construction drawings can all be deconstructed and reassembled
using the same nineteen facets and Q6 axioms.

I'm working on a whitepaper outlining the approach and the results of the first
few applications. It should be finished fairly shortly.

John O'