- From: Hugh Glaser <hugh@glasers.org>
- Date: Sun, 9 Sep 2018 21:20:20 +0100
- To: Kingsley Idehen <kidehen@openlinksw.com>
- Cc: public-lod@w3.org
Hi. Probably my last message :-) And thanks for trying to help me understand, but... :-) > On 9 Sep 2018, at 20:15, Kingsley Idehen <kidehen@openlinksw.com> wrote: > > On 9/9/18 1:07 PM, Hugh Glaser wrote: > ... >> >> I still don't understand why you can't just download Sage and run SPARQL queries on it, and compare the performance with Virtuoso doing the same thing - surely that would enable you to test their claim? > > Because my point is that they don't understand what our "Anytime > Feature" is and thus don't understand its relevance to the problem at > hand i.e., ad-hoc SPARQL query services published to the Web. > > We have many live instances that support our claims about the > aforementioned capability. Thus, a "like for like" comparison has to > offer a query endpoint that supports ad-hoc queries by anything (human > or machine). Otherwise, what are we comparing? > > My demonstrable claim is simple: > You can publish a SPARQL Query Service on the Web that allows ad-hoc > SPARQL Queries from anywhere by humans or macnines. If an alternative > claims it can do better, then at the very least it should produce a > comparable endpoint, right? No, basically, it isn't. Maybe if you had answered the question then I would understand better. For example, when I built a garbage collector and wanted to evaluate it, I downloaded other similar systems, then set up an environment that was as controlled as possible, and ran experiments to compare them. I then published the results in a paper. That is typically the way research is done and reported. And I certainly didn't go around berating the other researchers for not giving me access to their systems running in their labs. If someone wanted to disagree with my results (or indeed wanted to undertake the scientific method task of verifying them by replication), then i would expect them to ask for my software and my data and experimental methods, etc, although hopefully all that would already be available to them as open source and in the publication. It would not be normal for me to be expected to make my resources available to them to run their experiments in my lab. And in fact it would be a bad way of doing things, as there may be things wrong with the way I was running the experiments that meant the results were invalid. That is why real, independent repeatability is so important in research. So I don't understand why you seem obsessed with the idea that these researchers should give you access to use their resources, when what you presumably want to do is repeat their experiments in your own environment and control, so you can be confident off the results. Can you download their system? (Have I got that wrong? Is there some reason this can't be done?) And then run you it on your resources? As I say, thanks for trying. Best Hugh >> I guess I am missing something, but maybe you or someone else can help me. > > I am trying to :) > > Kingsley >> >> Best >> Hugh >> >>> On 9 Sep 2018, at 16:47, Kingsley Idehen <kidehen@openlinksw.com> wrote: >>> >>> On 9/8/18 7:08 PM, Hugh Glaser wrote: >>>> Hi. >>>> Not really my business (:-) - but that has never stopped me!), but I really don't understand the request for a live endpoint. >>>> If I want to evaluate different implementations of a software system, what I do is download or implement on my own server, and then run them (multiple times) over the same datasets on the same hardware configuration etc, so that I am comparing the same things while controlling as many variables as possible. >>>> Running stuff on other peoples' servers, with no reliable knowledge of the hardware or possibly even the datasets, seems of no value in terms of relative performance. >>>> So why are you asking for a SPARQL endpoint? >>>> Best >>>> Hugh >>> Hi Hugh, >>> >>> This is about proving or disproving a claim along the following lines: >>> >>> 1. Apples vs Apples -- are we comparing SPARQL Protocol endpoints or Web >>> Pages pegged to specific datasets running canned queries? >>> >>> 2. Conceptual understanding -- the concepts behind our "Anytime Query" >>> feature are only demonstrable using a live endpoint, in this context >>> i.e., the revealing claim about complete or incomplete solutions >>> >>> As I think you know, I am not going to accept questionable claims made >>> in public forums about our products. >>> >>> If I made these audacious claims be rest assured I would have a live >>> instance for objective verification. I wouldn't be directing folks to a >>> paper. >>> >>> Thanks. >>> >>> Kingsley >>>>> On 8 Sep 2018, at 19:31, Kingsley Idehen <kidehen@openlinksw.com> wrote: >>>>> >>>>> On 9/8/18 4:23 AM, Pascal Molli wrote: >>>>>> Dear Kingsley, >>>>>> >>>>>> We detail how, and in which context, we achieved the SaGe result in: >>>>>> >>>>>> Thomas Minier, Hala Skaf-Molli, Pascal Molli. SaGe: Preemptive Query Execution for >>>>>> High Data Availability on the Web. 2018. 〈hal-01806486〉 >>>>> Hi Pascal, >>>>> >>>>> You re making claims that are best verified in a live environment i.e., one of the fundamental virtues of a SPARQL endpoint. >>>>> >>>>> We have many live endpoints: >>>>> >>>>> [1] http://dbpedia.org/sparql >>>>> [2] http://dbpedia-live.openlinksw.com/sparql >>>>> [3] http://linkeddata.uriburner.com/sparql >>>>> [4] http://lod.openlinksw.com/sparql >>>>> >>>>> many others published by others. >>>>> >>>>> We even share documents regarding configuration [1]. >>>>> >>>>>> The result is confirmed with extensive benchmarks with Virtuoso, TPF and BrTPF. >>>>>> The paper details how to re-run the experimentations. >>>>> No, you are observations from a setup constructed by you. I am requesting that you demonstrate your claims via a live SPARQL endpoint, not a paper. I am requesting this for a specific reason, so can you provide this or not? >>>>> >>>>>> To improve data availability for end-users, we optimized the SaGe server for >>>>>> public LOD servers with read-only datasets. SaGe preempts running >>>>>> queries at low cost to provide a stable, responsive and unrestricted service >>>>>> in this context. >>>>>> >>>>>> The "Anytime Query" capability of a Virtuoso server is great but does not ensure complete >>>>>> answers. >>>>> Again, until you provide a live SPARQL endpoint you will not actually understand why "Anytime Query" actually exists i.e., you "..is great but does not ensure complete answers" speaks for itself. >>>>> >>>>>> If you need more details about our results, we will be happy to share with you what we >>>>>> observed on Virtuoso. >>>>> Can you provide a live SPARQL endpoint or not? >>>>> >>>>> Kingsley >>>>>> -- >>>>>> Pascal >>>>>> >>>>>> >>>>> -- >>>>> Regards, >>>>> >>>>> Kingsley Idehen >>>>> Founder & CEO >>>>> OpenLink Software (Home Page: >>>>> http://www.openlinksw.com >>>>> ) >>>>> >>>>> Weblogs (Blogs): >>>>> Legacy Blog: >>>>> http://www.openlinksw.com/blog/~kidehen/ >>>>> >>>>> Blogspot Blog: >>>>> http://kidehen.blogspot.com >>>>> >>>>> Medium Blog: >>>>> https://medium.com/@kidehen >>>>> >>>>> >>>>> Profile Pages: >>>>> Pinterest: >>>>> https://www.pinterest.com/kidehen/ >>>>> >>>>> Quora: >>>>> https://www.quora.com/profile/Kingsley-Uyi-Idehen >>>>> >>>>> Twitter: >>>>> https://twitter.com/kidehen >>>>> >>>>> Google+: >>>>> https://plus.google.com/+KingsleyIdehen/about >>>>> >>>>> LinkedIn: >>>>> http://www.linkedin.com/in/kidehen >>>>> >>>>> >>>>> Web Identities (WebID): >>>>> Personal: >>>>> http://kingsley.idehen.net/public_home/kidehen/profile.ttl#i >>>>> >>>>> : >>>>> http://id.myopenlink.net/DAV/home/KingsleyUyiIdehen/Public/kingsley.ttl#this >>>>> >>>>> >>>>> >>> -- >>> Regards, >>> >>> Kingsley Idehen >>> Founder & CEO >>> OpenLink Software (Home Page: http://www.openlinksw.com) >>> >>> Weblogs (Blogs): >>> Legacy Blog: http://www.openlinksw.com/blog/~kidehen/ >>> Blogspot Blog: http://kidehen.blogspot.com >>> Medium Blog: https://medium.com/@kidehen >>> >>> Profile Pages: >>> Pinterest: https://www.pinterest.com/kidehen/ >>> Quora: https://www.quora.com/profile/Kingsley-Uyi-Idehen >>> Twitter: https://twitter.com/kidehen >>> Google+: https://plus.google.com/+KingsleyIdehen/about >>> LinkedIn: http://www.linkedin.com/in/kidehen >>> >>> Web Identities (WebID): >>> Personal: http://kingsley.idehen.net/public_home/kidehen/profile.ttl#i >>> : http://id.myopenlink.net/DAV/home/KingsleyUyiIdehen/Public/kingsley.ttl#this > > > -- > Regards, > > Kingsley Idehen > Founder & CEO > OpenLink Software (Home Page: http://www.openlinksw.com) > > Weblogs (Blogs): > Legacy Blog: http://www.openlinksw.com/blog/~kidehen/ > Blogspot Blog: http://kidehen.blogspot.com > Medium Blog: https://medium.com/@kidehen > > Profile Pages: > Pinterest: https://www.pinterest.com/kidehen/ > Quora: https://www.quora.com/profile/Kingsley-Uyi-Idehen > Twitter: https://twitter.com/kidehen > Google+: https://plus.google.com/+KingsleyIdehen/about > LinkedIn: http://www.linkedin.com/in/kidehen > > Web Identities (WebID): > Personal: http://kingsley.idehen.net/public_home/kidehen/profile.ttl#i > : http://id.myopenlink.net/DAV/home/KingsleyUyiIdehen/Public/kingsley.ttl#this -- Hugh 023 8061 5652
Received on Sunday, 9 September 2018 20:20:47 UTC