[developers] DELPH-IN Discourse (Q&A)

Sat Nov 18 22:16:38 CET 2017

dear all,

i realize i know very little about how search engines decide to rank their
suggestions, but it appears google at least has indexed our mailing lists
(see below).  i now suspect folks might have had the impression that
DELPH-IN lists are not indexed because there will often be seemingly
‘better’ matches suggested by a search engine, possibly because there are
comparatively few web links /to/ these pages.  on this view, the forum will
likely have a similar challenge regarding search engine visibility, and
distributing information over multiple platforms may arguably actually add
to that problem.

—seeing as the ‘site:’ prefix is such an efficient way of making google
only return DELPH-IN matches, i think we should mention that technique
prominently on the wiki, and i shall look into creating a ‘DELPH-IN search’
master page, where users can submit queries to google such that they will
be automatically conjoined with ‘site:delph-in.net’ or maybe a disjunction
of more specific DELPH-IN addresses.

for that to work transparently, i have created ‘discourse.delph-in.net’ as
an alias (CNAME) for the UW address and would suggest that you configure
the server there to serve the forum under that address.  this way, we also
strengthen brand recognition and leave open the possibility of moving the
forum at some point, if need be, without having to change its address.
 does that sound like a feasible approach?

best wishes, oe

On Sat, 18 Nov 2017 at 17:20 Olga Zamaraeva <olzama at uw.edu> wrote:

> Dear all,
>
> > I believe this is the reason why Olga is talking about importing the
> mailing lists to new Delph-in Discourse forum.
>
> More because some people were worried about fragmenting the existing
> support. Fwiw, the new forum is not yet super visible from the outside
> (e.g. Google search), but that may change if we keep using it?
>
> I think however that one reason the googling results are inconsistent
> (sometimes one finds something that is linked to DELPH-IN resources but
> often times one doesn't) is that there are too few queries, and so Google
> is inclined to point the user to some other site where a similar topic is
> discussed? In that case, that will be the case with the forum also, though
> of course you can search within the forum.
>
> The difference with the mailing archives will then primarily be the
> salience of the working solution and overall usability, including the
> search experience and the formatting. I think the salience of the solution
> is important. For example, I wouldn't think it necessarily worthwhile to
> introduce a simple forum in addition to the existing resources, despite
> perhaps better formatting.
>
> Olga
>
> On Sat, Nov 18, 2017 at 2:50 AM Luis Morgado da Costa <
> luis.passos.morgado at gmail.com> wrote:
>
>> Dear all,
>>
>> I believe it was the mailing lists, and not the Wiki, that people were
>> finding not being properly indexed by search engines.
>>
>> I did a quick test with a couple of known headers in some mail sent to
>> the developers' list and I couldn't get any meaningful results. E.g.: "Docu
>> of the MaxEnt .mem file structure?"
>>
>> But I do get mixed results... when searching "and typical
>> unknown-word-aware DELPH-IN grammars use constraints on generic", I did get
>> a result that linked to the developers' list.
>>
>> I'm not sure the reason behind this inconsistency.
>>
>> Anyway, I believe this is the reason why Olga is talking about importing
>> the mailing lists to new Delph-in Discourse forum.
>>
>> Cheers,
>> L
>>
>>
>>
>> On Sat, Nov 18, 2017 at 6:33 PM Francis Bond <bond at ieee.org> wrote:
>>
>>> My apologies, I had thought they weren't but clearly they are.
>>>
>>> On Sat, Nov 18, 2017 at 6:10 PM, Stephan Oepen <oe at ifi.uio.no> wrote:
>>>
>>>> thanks for the reminder, francis.
>>>>
>>>> i believe someone voiced the impression at the summit already that the
>>>> DELPH-IN wiki is not indexed by search engines.  i admit i never looked
>>>> into interactions with search engines specifically, but i would be very
>>>> surprised if there were any settings on the oslo DELPH-IN servers that
>>>> would scare off search engines.
>>>>
>>>> why do you believe we are not indexed?  i just did a quick test, which
>>>> seems to give the result one would expect (see below).
>>>>
>>>> all best, oe
>>>>
>>>>
>>>>
>>>>
>>>> On Sat, 18 Nov 2017 at 10:08 Francis Bond <bond at ieee.org> wrote:
>>>>
>>>>> Thank you for taking the initiative Olga!   I hope that this make
>>>>> information flow even better.
>>>>>
>>>>> I think one of the problems with the wiki is that it does not seem to
>>>>> be indexed by, e.g. google.   Stephan: is there a setting you can
>>>>> change to get it indexed?   I think this would make the information
>>>>> there a lot more discoverable, ...
>>>>>
>>>>> On Sat, Nov 18, 2017 at 7:02 AM, Olga Zamaraeva <olzama at uw.edu> wrote:
>>>>> > Dear developers,
>>>>> >
>>>>> > In the last summit in Oslo, we discussed a possibility of having a
>>>>> Q&A forum
>>>>> > in stackexchange style.
>>>>> >
>>>>> > Many people were enthusiastic, and, thanks to University of
>>>>> Washington
>>>>> > staff, we were able to set up a forum powered by Discourse:
>>>>> >
>>>>> > https://delphinqa.ling.washington.edu/
>>>>> >
>>>>> > I would like to invite each of you to create an account and try it
>>>>> out!
>>>>> >
>>>>> > The forum was created very recently, but at this point, there should
>>>>> not be
>>>>> > too many hiccups. We need to decide a few things soon (see below),
>>>>> so, we
>>>>> > want to involve the entire community at this point, even if we are
>>>>> still
>>>>> > testing the forum.
>>>>> >
>>>>> > A few notes, for those inclined to read them :).
>>>>> >
>>>>> > 1) The primary purpose of this forum is to accumulate concrete
>>>>> solutions to
>>>>> > technical problems and make them easily discoverable. So the primary
>>>>> > scenario is for one to encounter an error message, and, instead of
>>>>> trying to
>>>>> > locate the answer somewhere on the wiki or in the email archives and
>>>>> ending
>>>>> > up emailing the list for the 100th time anyway, one would query the
>>>>> forum
>>>>> > (using e.g. the text of the error message). If their question was
>>>>> already
>>>>> > answered, the assumption is they will be able to immediately find
>>>>> out what
>>>>> > the working solution is. This is, in my opinion, the most important
>>>>> feature
>>>>> > of stackexchange Q&A style, and one that other types of support
>>>>> lack.**
>>>>> > Moreover, the forum is also a fine platform for open-ended
>>>>> discussions, with
>>>>> > a nice addition of the upvote button which makes better, clearer
>>>>> posts more
>>>>> > salient in the thread.
>>>>> >
>>>>> > **(Some people oppose this style since, in their opinion, finding
>>>>> the right
>>>>> > solution "too easily" leads to lack of understanding of why the
>>>>> right answer
>>>>> > is the right answer. I agree  that there may be some truth to that
>>>>> but I
>>>>> > also think that this ultimately depends on the user and how much and
>>>>> in what
>>>>> > style they are willing to learn; there is also a question of
>>>>> priorities:
>>>>> > sometimes what is needed the most is to fix the problem asap.)
>>>>> >
>>>>> > 2) I will be the admin of the forum for now. This means I am the
>>>>> person to
>>>>> > whom you can complain about things if something is not working for
>>>>> you. It
>>>>> > also means I can do various things for you, such as create/merge
>>>>> categories
>>>>> > and so forth. If you would like to also be an admin or a moderator,
>>>>> let me
>>>>> > know. It will probably help to have more admins, especially in other
>>>>> time
>>>>> > zones.
>>>>> >
>>>>> > 3) You do not need to monitor the forum or anything like that,
>>>>> unless you
>>>>> > would like to. Instead, you can subscribe to categories or tags, and
>>>>> the
>>>>> > forum will email you when a new post is created in the category or
>>>>> with that
>>>>> > tag. You can choose whether you will be notified about the first
>>>>> post only
>>>>> > or each reply.
>>>>> >
>>>>> > 4) Some developers expressed concern that we are thus fragmenting our
>>>>> > support platforms. I understand this concern and would like to
>>>>> address it by
>>>>> > both making an effort to import the email archives into the new forum
>>>>> > (unless it proves too messy) and by pointing out that the email+wiki
>>>>> do not
>>>>> > and cannot really serve the purpose that I think this forum will
>>>>> serve. It
>>>>> > seems that the wiki, the email list, and the forum can very happily
>>>>> coexist,
>>>>> > just like python documentation and python-related sections of
>>>>> stackoverflow
>>>>> > (and presumably some people exchange emails about python when they
>>>>> want to
>>>>> > discuss something in a manner they don't think the general public
>>>>> will
>>>>> > necessarily benefit from). Obviously, the wiki should be linked to
>>>>> from the
>>>>> > forum (and vice versa) and instructions on the wiki can be updated
>>>>> whenever
>>>>> > it makes sense to do so.***
>>>>> >
>>>>> > ***(Note however that it can be rather impractical to try and
>>>>> incorporate
>>>>> > every single detail into the wiki, especially if the detail only
>>>>> makes sense
>>>>> > in the context of a concrete question. In fact, every time one adds
>>>>> > something to the wiki, one in a way is adding a new dimension to the
>>>>> space
>>>>> > in which people will be searching for solutions... without adding a
>>>>> starting
>>>>> > point for the search, necessarily.)
>>>>> >
>>>>> > 5) The forum is hosted by the University of Washington, so, there is
>>>>> no
>>>>> > danger of some external host interfering in our affairs. Discourse is
>>>>> > open-source software. It is also a very flexible system already, with
>>>>> > various plugins etc. We can customize it as we like.
>>>>> >
>>>>> > If you would like to discuss any of this, we can do it here over
>>>>> email or on
>>>>> > the forum; there is already a topic there regarding the email
>>>>> archives:
>>>>> >
>>>>> https://delphinqa.ling.washington.edu/t/importing-email-archives/28/5
>>>>> In
>>>>> > particular, that topic invites you to participate in a poll.
>>>>> >
>>>>> > Thank you for considering,
>>>>> >
>>>>> > Olga
>>>>> >
>>>>> > P.S.: You will see that there are already some users and topics; I
>>>>> hadn't
>>>>> > advertised the forum here earlier because it was not yet working in
>>>>> a stable
>>>>> > fashion, but it should be now :). Note also that for now, this is
>>>>> all in the
>>>>> > testing stage. For example, if we do decide to import the mailing
>>>>> list
>>>>> > archives, we may need to purge the existing content.
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Francis Bond <http://www3.ntu.edu.sg/home/fcbond/>
>>>>> Division of Linguistics and Multilingual Studies
>>>>> Nanyang Technological University
>>>>>
>>>>
>>>
>>>
>>> --
>>> Francis Bond <http://www3.ntu.edu.sg/home/fcbond/>
>>> Division of Linguistics and Multilingual Studies
>>> Nanyang Technological University
>>>
>> --
>>
>> *Luis MORGADO DA COSTA (Mr) |* Ph.D. Student *| *Global Asia * |* Nanyang
>> Technological University
>>
>> Interdisciplinary Graduate School* |* 50 Nanyang Avenue
>> <https://maps.google.com/?q=50+Nanyang+Avenue&entry=gmail&source=g>,
>> S(639798)
>>
>> Tel: (65) 90541507  GMT+8h *|* Email: depa0003 at e.ntu.edu.sg
>> <XXXXX at e.ntu.edu.sg>* |* Web: http://igs.ntu.edu.sg/
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.delph-in.net/archives/developers/attachments/20171118/c5e47d6f/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: IMG_1263.jpg
Type: image/jpeg
Size: 157547 bytes
Desc: not available
URL: <http://lists.delph-in.net/archives/developers/attachments/20171118/c5e47d6f/attachment-0001.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: IMG_1268.jpg
Type: image/png
Size: 181023 bytes
Desc: not available
URL: <http://lists.delph-in.net/archives/developers/attachments/20171118/c5e47d6f/attachment-0001.png>