[Corpora-List] how many subscribers are there on corpora-list?

Marco Guerini marco.guerini at gmail.com
Thu Aug 22 10:48:02 CEST 2013


Dear Hieu,

sorry but I disagree with your last point.

Click-Through cannot be taken as a measure of readership. At most it can be taken as a measure of how "catchy" and "persuasive" the text of your mail was. CT can vary drastically depending on message wording and audience, and a CT-rate of 5% in email-marketing is already a success (3% on average). If we consider active users on forums, 3% is also a reasonable ratio (i.e. 97% of forums registered users are just "lurkers"). In my opinion, another viable solution to estimate readership is to inspect the analytics of Corpora List archives.

-- Marco Guerini, PhD www.marcoguerini.eu

On Wed, Aug 21, 2013 at 3:22 PM, Hieu Hoang <Hieu.Hoang at ed.ac.uk> wrote:
>
>
>
> On 21 August 2013 13:23, John F Sowa <sowa at bestweb.net> wrote:
>>
>> On 8/21/2013 7:08 AM, Trevor Jenkins wrote:
>>>
>>> The single number doesn't mean anything.
>>
>>
>> I agree with that statement. What I believe people want to know is
>> the impact of various lists relative to other methods of sharing
>> their thoughts with colleagues in related fields.
>
>
> we once tried to measure the impact by measuring the URL click-through when
> we released a LM software that was of interest to the NLP community. The
> email went out to several mailing lists.
>
> However, we didn't set up the experiment correctly (wasn't simultaneous,
> same URL etc) so the result was highly suspect.
>
> I think someone advertising a conference or software can do the same and it
> would be a good measure of readership
>
>>
>> The simple answer is that nobody really knows how to measure impact
>> in any absolute sense. Even relative impact is extremely difficult
>> to measure. And any numbers that are generated are suspect.
>>
>>
>> On 8/21/2013 6:24 AM, Hieu Hoang wrote:
>>>
>>> imo, it is an imperfect but useful bit of information to gauge the
>>> relative popularity of the mailing list. In Oct 2010, i asked the same
>>> question for some popular NLP/SMT mailing list:
>>> Corpora list : 3600
>>> EAMT - 792
>>> Moses - 630
>>
>>
>> But what is the relative percentage of email that subscribers actually
>> (a) read, (b) delete within N seconds of being opened, (c) do anything
>> about -- such as reply to, click on a URL, use as a starting point for
>> further searches, etc. ?
>>
>> John
>>
>>
>> _______________________________________________
>> UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
>> Corpora mailing list
>> Corpora at uib.no
>> http://mailman.uib.no/listinfo/corpora
>>
>
>
>
> --
> Hieu Hoang
> Research Associate
> University of Edinburgh
> http://www.hoang.co.uk/hieu
>
>
> _______________________________________________
> UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
> Corpora mailing list
> Corpora at uib.no
> http://mailman.uib.no/listinfo/corpora
>



More information about the Corpora mailing list