Re: Enterprise information system

I think Gio is right about reinventing the wheel but there certainly is a
desire to move beyond the status quo,  which people aren't all that happy
with that either.

I'd go back to Microsoft Access to highlight the problem with status quo.

I think Access is a pretty good tool that punches above its weight,  but
when businesspeople want to make a "database" they frequently pick Excel.
Even though you can use Excel to visually create database tables and forms
and sprinkle in just a little bit of Visual Basic to make a real app,
 untrained people have a difficult time with data modelling.  Even if you
could make all the "coding" go away,  you'd still have extreme difficulties
because people would (i) pick a bad data model,  and (ii) realize it later
when there is a lot of data in the system.

Two angles to these problems are:

(1) The OMG has developed a large number of standards centered around the
"model driven architecture",  which,  like the Semantic Web,  is a work in
process.  The OMG has more of an enterprise focus so it is worth
understanding what they've done.  UML started out as a diagramming tool,
 but eventually wants to become executable,  because so long as you have a
separate map and territory,  these will diverge.

(2) "Business rules engines",  primarily based on forward chaining,  have
been promised as another technology that lets businesspeople express their
will in something human readable but this too is a challenge.  To be
specific,  I've built some systems that are based on a first-order logical
theory,  and many of these rules engines don't have a real query optimizer
so the order that I write the conditions in can be the difference between
something that runs in 1ms and something that takes 20 seconds.  If
somebody who isn't hep to that makes a change to the system,  they can
break it.

The are multiple angles of attack on this problem,  like ultimately you
need the query optimizer,  but I think semantics have a lot to offer both
(1) and (2) in the sense of being able to start with a general domain model
for something like CRM and then specialize for a particular company without
doing a lot of programming.



On Thu, Feb 26, 2015 at 1:38 AM, Giovanni Tummarello <g.tummarello@gmail.com
> wrote:

> Hugh,
>
> i think if you send them down a route where you have to write bespoke
> software (which uses RDF concept, hard to find developers to write and
> maintain) for purposes for which mature widely tested and widely spread
> software exists you'd be doing them a disservice.
>
> Eventually they'll find someone showing them how "normally these things
> are done" they'd say "hey but this is what we need really - give it to us
> now". This will at that point both possibly spoil your reputation with them
> and their perception toward LD technologies, which could on the other hand
> be useful if used in moderation - or in domain where data "variability" is
> indeed extreme.
>
> I would recommend look for good open source personnel or project
> management system (Groupware etc) and see if it makes sense to introduce
> concepts such as unique identifiers used across the organization (Which
> could be resolvable URI thus giving you a "homepage" for every core concept
> of the company). But be flexible even in this case if you are to add any LD
> at all.. people often prefer typ+number (e.g. personnel ID, project code)
> than URIs  so if you do a global lookup interface for all, dont insist they
> must use URIs to find something. However if anything does in fact "show" on
> a stable and nice URI in their browser, they'll naturally refer to it when
> passing each other references in emails etc.  but this is the same than
> what they would be doing with any reputable content management system.
>
> my2c
> Gio
>
> On Wed, Feb 25, 2015 at 10:06 PM, Hugh Glaser <hugh@glasers.org> wrote:
>
>> So, here’s a thing.
>>
>> Usually you talk to a company about introducing Linked Data technologies
>> to their existing IT infrastructure, emphasising that you can add stuff to
>> work with existing systems (low risk, low cost etc.) to improve all sorts
>> of stuff (silo breakdown, comprehensive dashboards, etc..)
>>
>> But what if you start from scratch?
>>
>> So, the company wants to base all its stuff around Linked Data
>> technologies, starting with information about employees, what they did and
>> are doing, projects, etc., and moving on to embrace the whole gamut.
>> (Sort of like a typical personnel management core, plus a load of other
>> related DBs.)
>>
>> Let’s say for an organisation of a few thousand, roughly none of whom are
>> technical, of course.
>>
>> It’s a pretty standard thing to need, and gives great value.
>> Is there a solution out of the box for all the data capture from
>> individuals, and reports, queries, etc.?
>> Or would they end up with a team of developers having to build bespoke
>> things?
>> Or, heaven forfend!, would they end up using conventional methods for all
>> the interface management, and then have the usual LD extra system?
>>
>> Any thoughts?
>>
>> --
>> Hugh Glaser
>>    20 Portchester Rise
>>    Eastleigh
>>    SO50 4QS
>> Mobile: +44 75 9533 4155, Home: +44 23 8061 5652
>>
>>
>>
>>
>


-- 
Paul Houle
Expert on Freebase, DBpedia, Hadoop and RDF
(607) 539 6254    paul.houle on Skype   ontology2@gmail.com
http://legalentityidentifier.info/lei/lookup