W3C home > Mailing lists > Public > semantic-web@w3.org > March 2019

Re: Solomon''s curse and search Bias

From: Paola Di Maio <paoladimaio10@gmail.com>
Date: Mon, 4 Mar 2019 19:54:38 +0800
Message-ID: <CAMXe=SobcjFKJ7U8CVTgdHnoL8+2JZW=ke7onPzYcKmjvagHSw@mail.gmail.com>
To: Sherman Monroe <sdmonroe@gmail.com>
Cc: ontolog-forum <ontolog-forum@googlegroups.com>, "schema.org Mailing List" <public-schemaorg@w3.org>, SW-forum <semantic-web@w3.org>, public-aikr@w3.org, Kingsley Idehen <kidehen@openlinksw.com>, Hugh Williams <hwilliams@openlinksw.com>, vios <vios@dev-team.com>
Sherman and Thomas P
thanks a lot for sharing your search journey with friendly narrative and
about this project. Looks really neat!

I am still thinking about the need to carry ot some structured data search
on open web via  general search engines tho.  I hope Google may consider
adopting your architecture or at least some of these ideas which in
principe work at least within a closed data set,

Thomas P

2. Tagging information by area/topic/etc is infernally hard as a general

thing, and all search engines like Google's have to go on is the URL and

text of a web page.  It's hard for people, and it's a lot harder to

figure out how to write software to do it.  No wonder we don't have it yet.


Tagging is one way, but there are other approaches-
i view this not as a software problem, but way the results of the search
algo is
Ășnsorted, rather than logically sorted and contextualized

 I read so many papers about context search.
why is that not happening?

I am not sure this is a technology problem at all, but a serivce design
issue,

I dont think there is any technical obstacle to returning sorted results
according
to some logical search criteria



PDM
So it's partly a user interface issue, but more deeply it's just really
hard.


On Mon, Mar 4, 2019 at 1:29 AM Sherman Monroe <sdmonroe@gmail.com> wrote:

> Hello Paola,
>
> You might be interested in VIOS Network <http://poc.vios.network>
> project, and its folder-driven approach (it is still in prototype phase and
> quite buggy).
>
> I started with keywords Solomon Curse <http://bit.ly/2GWOmEc>. Since the
> results were sparse (it only checked the records' rdfs:label), I clicked Expand
> Search <http://bit.ly/2Tf367V> button to apply search to all text fields.
> I press the R key <http://bit.ly/2H9yjSu> to view the roles (property-of
> relations) of the things in the results set, as I am trying to pivot out to *historical
> causes and conditions* related to the *House of David*. Primary topic
> sounds interesting, so I click the number next to Primary topic role
> <http://bit.ly/2GXI4nQ>. I discover nothing new under topics, so I click
> the words "Solomon Curse" in the breadcrumbs to move back to root
> <http://bit.ly/2C13cFE>, then I press the S key <http://bit.ly/2EK4RRK>
> to view subject badges. *Jewish mythology* sounds promising, so I click
> that badge <http://bit.ly/2Ud7v7H> to add the filter. I then click Jewish
> Mythology <http://bit.ly/2ErckUd> in the breadcrumbs to "go there", and press
> R <http://bit.ly/2C3QINl> to view its roles, and click the expand button
> next to "subject" to preview the list of things under subject *Jewish
> Mythology*.
>
> Unfortunately, I was not able to resolve your query due to incompleteness
> in the data space, but I hope to demonstrate that it is possible to allow
> high precision, accuracy, and most importantly, granular user control over
> the result sets, by publishing our data as linked data. The onus is with
> data publishers to link the data in such a way as to provide enough filters
> to allow users to *configure* the answers to their questions.
>
> As another example, I checked a different data space (URI Burner
> <http://linkeddata.uriburner.com/>), for keywords House of David
> <http://bit.ly/2NEN80J>, to continue the research. I then created a topic
> chart <http://bit.ly/2NGtzoL> on the folder contents, which results in
> some rather toxic news items. I was delighted to see the content was marked
> "news" or "opinion"...
>
> [image: image.png]
>
> ... which lets me restrict the results to "news" column only. I press the
> G key <http://bit.ly/2Uh4AL9> to show the libraries (the Named Graph)
> these results came from, and if I trust the library, then I know the items
> in the "news" column are real/verified items. So, one important part of
> address your issue is the notion of how do you assign trust to data
> publishers. I believe one way is to link Named Graphs to identities in the Web
> of Trust <https://en.wikipedia.org/wiki/Web_of_trust>.
>
> *Notes and caveats:*
>
>    - This is a proof-of-concept site only, use at your own risk
>    - Some of the data servers in the above example are prone to a filter
>    bug in Virtuoso
>    <https://github.com/openlink/virtuoso-opensource/issues/823> that is
>    currently being addressed
>    - The server resources for both the proxy server and the data server
>    are minimum and crash easily (use browser console log and toggle on Debug
>    to troubleshoot)
>    - The functionality of this PoC is not fully implemented, some
>    features work, some not entirely, some not at all
>
> For those interested, here is some more information about VIOS Network
> project <https://medium.com/@sdmonroe/vios-network-99488f5bf29d>. The
> framework is open-source and highly extensible (e.g. the Compare Chart is a
> "plugin"), so if you know Javascript/HTML5/SPARQL and are willing to lend a
> hand, please reach out :)
>
> Cheers,
> -sherman
>
>
>
>
> On Sat, Mar 2, 2019 at 9:14 PM Paola Di Maio <paola.dimaio@gmail.com>
> wrote:
>
>>
>> I wanted to share a concern, as I know posts gets read and issued picked
>> up and addressed in time
>>
>> I searched Google today for Solomon Curse, trying to find some references
>> to some historical cause and conditions in the first house of David - not
>> in relation to a specific race, but more in relation to the history of the
>> modern world
>> to see if anyone is following up the courses and recourses of history
>> https://www.iep.utm.edu/vico/
>>
>>
>> Well, I was shocked to see that the first page of results were all about
>> a book and its author, and nothing
>> about history came up at all.  I had to add additional words to create
>> some context to dig up some
>> historical references.
>>
>> Just wanted to point out that I am very concerned about future
>> generations receiving a distorted
>> version of history by heavily commercially biased search results when
>> typing some search terms and
>> getting only/mostly the results from one entity, rather than a
>> representation of the plurality of meanings and contexts
>>
>> Bias is a known problem in searches, however I was hoping that by now we
>> would have
>> some mechanisms to reduce this bias? Doesn't look like it.
>>
>> I hope that schema.org could help that by creating metaschemas for
>> disambiguation
>> or other mechanism, such a representation of context which should include
>> at least
>> two perspectives: the domain a search term is present, and the
>> time/chronology (to show which came first)
>>
>>  Just a sunday morning note before digging in more confusing knowledge
>> from search results
>>
>> PDM
>>
>>
>> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=icon> Virus-free.
>> www.avast.com
>> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=link>
>> <#m_-979156246711179234_m_-2020286702584855163_m_9019427200774627252_m_6661572909156064707_m_6942727257523792062_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
>>
>
>
> --
>
> Thanks,
> -sherman
>
> Every good gift and every perfect gift is from above, and cometh down from
> the Father of lights, with whom is no variableness, neither shadow of
> turning.
> (James 1:17)
>


image.png
(image/png attachment: image.png)

Received on Monday, 4 March 2019 11:55:43 UTC

This archive was generated by hypermail 2.3.1 : Monday, 4 March 2019 11:55:43 UTC