Sentiment Analysis

My, as usual a bit off center, take on Sentiment Analysis is posted here: Pinwheel Time, taking the Time out of Social Networking (and leaving the Trends in) [1].

It is a classic "Tides" harmonic analysis, but gives some insights into how to avoid being enslaved by the "news".  It does include some cool ways to simplify calculations and do away with the "fortnight" (1/7) harmonic and the observation that groups need no "ordering" alphabetical of otherwise - but they do need C4 Group Symmetry.

--Gannon

[1] http://www.rustprivacy.org/2012/roadmap/soc-net-trends.pdf






________________________________
 From: Gannon Dick <gannon_dick@yahoo.com>
To: eGov IG (Public) <public-egov-ig@w3.org>; Soon Ae Chun <Soon.Chun@csi.cuny.edu>; "carlos.ivan.chesnevar@gmail.com" <carlos.ivan.chesnevar@gmail.com> 
Sent: Friday, August 24, 2012 2:58 PM
Subject: Re: [Minutes] 2012-08-24 (was Re: W3C eGov Meeting (Los Angeles  time))
 

Great call :-)

1.  I had an observation about claims for the accuracy of Sentiment Analysis.  It rather depends on the expectation that opinion is in fact polarized.  Many years ago a friend described a similar problem ...  She returned home after her Living Room was re-carpeted.  So she looked, and found a Husband-sized lump, ignored it, and found a cat sized lump and called the workmen back.  The lump size gives an indication of the form underneath a Gaussian.  If the expectation is that the form underneath is two poles then the width of the Gaussian shape above is not a reflection of the standard deviation of the sample, but rather a reflection that the sample has no Central Limit (average), something which makes a Gaussian tall and skinny.  Three sigma is a sampling rate ~ 99.7% of Tweet Context (Subject), whereas one sigma (~ 68%) assumes every lump has one peak underneath.  My method is easier - approximate the lump underneath by an isosceles triangle
 and pivot each half to "see" the expected poles.   Hint: your approximation to the cat sized Gaussian just became an approximation to a catenary.
http://www.rustprivacy.org/2012/roadmap/carpet-lump.png

2. <olyerickson> Overarching question: for all of these papers, do they focus on their immediate countries/regions, the whole" world or a specific interesting country (e.g. USA, UK, EU countries, etc)

I too would like to hear an answer to Mr. Erickson's overarching question ... in this Crowdsourcing Context:  Canberra is 13 hours ahead of Texas at the moment.  It will soon be Saturday night, here.  Nothing says piety like a Sunday morning in Sidney with the contemporaneous input of 20 something million liquored up Cowboys.  I foresee some problems in the offing with real time data.

3.  Ms. Flagg ... is there a "Time Zero" on the Archival Data Sets from the various suppliers ?  The reason I ask is because the archives might be crucial to validate the expectations for success of a Government program, pre-launch.  It seems to me that this would make the risk of using a new technology for the Agency, unnecessarily greater, and for you a harder sell.

--Gannon  



________________________________
 From: "Holm, Jeanne M (1760)" <jeanne.m.holm@jpl.nasa.gov>
To: Phil Archer <phila@w3.org> 
Cc: eGov IG (Public) <public-egov-ig@w3.org>; Soon Ae Chun <Soon.Chun@csi.cuny.edu>; "carlos.ivan.chesnevar@gmail.com" <carlos.ivan.chesnevar@gmail.com>; "Aisenberg, Michael A." <maisenberg@mitre.org>; "Strawn, George" <Strawn@nitrd.gov>; "Geraghty, John P." <jgeraghty@mitre.org>; "Irven, Candy" <cirven@egov.com>; "jeffkaplan88@gmail.com" <jeffkaplan88@gmail.com>; Dave McAllister <dmcallis@adobe.com>; Bobbi Martin <Bobbi.Martin@tn.gov>; Malcolm Crompton <mcrompton@iispartners.com>; Sylvia Webb <sylvia.webb@vision4standards.com>; "Khoshafian, Setrag" <Setrag.Khoshafian@pega.com>; "Lee, Deirdre" <Deirdre.Lee@deri.org>; Eriksen Gorm Haug <Gorm-Haug.Eriksen@dss.dep.no> 
Sent: Friday, August 24, 2012 11:52 AM
Subject: Re: [Minutes] 2012-08-24 (was Re: W3C eGov Meeting (Los Angeles  time))
 
Phil and Deirdre--

Thanks so much for your help in getting the minutes together and out so
quickly!

--Jeanne

**********************************************************
Jeanne Holm
Evangelist, Data.gov
U.S. General Services Administration
Cell: (818) 434-5037
Twitter/Facebook/LinkedIn: JeanneHolm
**********************************************************




On 8/24/12 9:51 AM, "Phil Archer" <phila@w3.org> wrote:

>The minutes of today's eGov IG call are online at
>http://www.w3.org/2012/08/24-egov-minutes.html and I've updated the wiki
>accordingly.
>
>Thanks again to the three very interesting speakers on today's call.
>
>Next meetings are:
>3 September (Eurasian timing)
>21 September (Atlantic
 timing)
>
>Cheers
>
>Phil.
>
>On 23/08/2012 15:43, Holm, Jeanne M (1760) wrote:
>> Event Invitation
>>
>> Title:
>>
>>     
>>
>> W3C eGov Meeting (Los Angeles time)
>>
>> Location:
>>
>>     
>>
>> +1-617-761-6200 or sip:zakim@voip.w3.org then conference code 3468#
>> ("EGOV#")
>>
>> When:
>>
>>     
>>
>> 24 August 2012 16:00 ­ 17:30
>>
>> Organizer:
>>
>>     
>>
>> Holm, Jeanne M (1760)
>>
>> Description:
>>
>>     
>>
>> When: Friday, August 24, 2012 11:00 AM-12:30 PM. (UTC-05:00)
 Eastern
>> Time (US & Canada) Where: +1-617-761-6200 or sip:zakim@voip.w3.org then
>> conference code 3468# ("EGOV#") The next meeting of the W3C eGOV
>> Interest Group will take place on Friday, August 24 from 8:00 to 9:30
>> a.m. Los Angeles (11 a.m.-12:30 p.m. EST) time. The discussion will
>> focus on the use of social media by government and other state actors to
>> interact with citizens, busines,s and with each other, and related
>> technology and policy issues. If you are interested in sharing your
>> experience with social media development and use in the public space or
>> know about an initiative, experience, research findings, etc. worth
>> sharing, please present it to the group or encourage your colleagues to
>> present. Please contact Jeanne at Jeanne.Holm@jpl.nasa.gov to coordinate
>> any future presentations. AGENDA 08:00 - 08:05 Welcome and introductions
>> 08:05 - 08:15 Defining the discussion space 08:15 - 09:15 Social Media
>> case studies --Carlos Ivan Chesnevar--DECIDE 2.0: Intelligent Processing
>> of Citizen Opinions in Social Media (slides can be downloaded at
>> http://cs.uns.edu.ar/~cic/w3c-decide20.zip ) --Soon Ae Chun--Social
>> Media in Government: Findings of the Special Issue of Government
>> Information Quarterly 09:15 - 09:25 Summary and next steps 09:25 - 09:30
>> Next meeting We will also need a member of the group to help scribe.
>> This is essentially taking notes in the IRC chat during the
>> teleconference to capture the key points of the discussion. If you are
>> able to volunteer to
 scribe for the next meeting (or interested for
>> future meetings), please contact Tomasz or Jeanne directly. Looking
>> forward to talking to you tomorrow! Ways to connect: --Telecon line:
>> Dial +1-617-761-6200 or
>> 
>>sip:zakim@voip.w3.org<mailto:zakim@voip.w3.org><mailto:zakim@voip.w3.org>
>> then
>> conference code 3468# ("EGOV#") --W3C IRC channel #egov, see
>> http://www.w3.org/Project/IRC/ or use http://irc.w3.org/?channels=egov
>> --Scribe: (please volunteer, if you've done this before) --Group access
>> via the W3C at http://www.w3.org/egov/ and also via LinkedIn at the W3C
>> eGovernment Interest
>>
 Group:http://www.linkedin.com/groups?gid=1800648&trk=hb_side_g
>>
>
>-- 
>
>
>Phil Archer
>W3C eGovernment
>http://www.w3.org/egov/
>
>http://philarcher.org
>+44 (0)7887 767755
>@philarcher1