Check out Safety Emporium for your N95, N99, and face shield needs.
Date: Thu, 4 Aug 2011 13:17:00 -0400
Reply-To: DCHAS-L Discussion List <DCHAS-L**At_Symbol_Here**LIST.UVM.EDU>
Sender: DCHAS-L Discussion List <DCHAS-L**At_Symbol_Here**LIST.UVM.EDU>
From: ILPI <info**At_Symbol_Here**ILPI.COM>
Subject: Re: Recent Accidents in College Level and beyond Chemistry Lab
In-Reply-To: <4E3A8AAC020000BA0002FFB7**At_Symbol_Here**gwsmtp1.smith.edu>
For well over ten years
now, I've envisioned a web-based tool in which a researcher could input
their chemicals/procedure and get output that recommends PPE, warns
about incompatibilities, recommends waste disposal procedures, finds
alternatives, displays accident data, etc. etc. My baby step in
that direction was the Material Safety Data Sheet Demystifier:
http://www.ilpi.com/m
sds/ref/demystify.html And
http://hyperglossary.org/hg/abo
ut extends the ideas originally developed in the
MS-Demystifier.
Of course, to extrapolate to
the tool I describe would require tremendous effort.
The Chemical Safety Headlines =46rom Google=99 that get posted to
this list give an insight into just how difficult it can be to have a
computer collate and digest information. Ralph has set up a
Google feed that scans for keywords and collects the articles.
That then gets run through an ever-growing JavaScript that I wrote
to make a first guess at tagging the keywords in the article.
Those then get a manual review before they are collected and sent
off the list. While the parsing algorithms work well for what they
are, there is always some wording/construction/meaning that the code has
not anticipated, so I'm constantly tweaking it. For example, when
we look to see whether the incident involved injury or death, it's not
as simple as looking for keywords. The article might say "no one
was injured" or "chemical X can cause death" and the code needs to
recognize not to raise the Injury or Death flag for those kinds of
constructions. Even for something as basic as the location, the
algorithm does its best to determine where the incident occurred by
looking first for a proper article byline (rare), scanning the text of
the article for the state (which can fail as we had the other day with
an incident in the city of Indiana, Pennsylvania), and the URL itself
(but a Boston newspaper may be reporting on an incident in a bordering
state), etc.
Still, once trained, the system
works fairly well. My experience so far indicates that with
correctly parsed information this kind of safety wizard can be built.
The way to achieve that most effectively and affordably (safety
information should not be out of the reach of schools that can not
afford an expensive subscription) is through a wiki system. Wikis
are web-based frameworks that let users contribute to the content, as in
WIkipedia. By having members of the chemistry/safety community
contribute (anonymously or otherwise) to a chemical safety system, we do
not have to rely on search engine logic. And when we want to have
the system make recommendations, those should be fairly accurate because
the information will have been rigorously parsed by
humans.
If the chemical community (presumably
under the aegis of the ACS) could establish a chemical safety wiki, it
would provide the necessary central resource for the tool I described
above. Imagine a system where you could input n-butyllithium and
not only get the usual sources of safety information (Bretherick etc.)
but be able to view a list of laboratory incidents/injuries/deaths,
handling procedures from Aldrich's web site, links to safety videos,
reports on the AIHA site, old DCHAS post, titration methods etc.
Truly comprehensive and *relevant* information without having to
do an extensive web/literature search.
This is
doable today. It needs a commitment, a back-end database, and a
core of volunteers to get it off the ground. The wiki nature makes
it self-maintaining.
Rob
Toreki
=========================
=========================
====
Safety Emporium - Lab & Safety Supplies
featuring brand names
On Aug 4, 2011, at 12:03 PM,
Margaret Rakas wrote:
Re:
[DCHAS-L] Recent Accidents in College Level and beyond Chemistry
Lab
Thanks to Ralph's ability to comb through and present a wide
variety of chemical accidents from around the globe, we frequently have
access to first reports about accidents in academic institutions.
However, it is relatively rare that these present the kind of
information that would help improve chemical safety at other
institutions, because chemical names are omitted or mangled, full
details of the type of process occurring aren't included (may not be
known at the time of publication), etc. And follow-up disclosures
frequently are quite difficult to find: they either don't happen,
aren't published (who really wants to read about yesterday's news,
especially when it's the dry topic of a chemical accident that isn't
otherwise sensational), or otherwise don't surface. I am sure the
incidents reported to C&E News are a small fraction of what has gone
wrong in labs, although I am very grateful to the writers and editors
for taking the time and space to make colleagues aware.
I have no desire to make any researcher, at any
level or institution, 'look bad'. I also understand the reluctance
of the institution's PR and legal offices to release details that they
feel compromise the institution. But I would really hope the
Councillors would discuss the implementation of an anonymous reporting
system that would include the details, because for some incidents, that
is the only way to understand you may have a problem at your own
instititution--it just (luckily) hasn't happened yet. I think it
is also important to realize these academic chemical incidents don't
just happen in chemical departments; biologists, neuroscientists, and
engineers use hazardous materials above and beyond the standard
corrosives and flammables, and those of us who work in
academic safety need to know about incidents and near-misses in those
areas as well, so we can warn our researchers.
sincerely,
Margaret
My
personal opinion only, not legal or business advice, and may not
represent the opinion of my employer or any group to which I
belong..
Previous post | Top of Page | Next post