Re: [protege-user] DL query bug ?
Jean-Louis Janin <[email protected]> Sun, 23 Nov 2025 11:59:55 +0100
| Newsgroups | gmane.comp.misc.ontology.protege.owl |
|---|---|
| Message-ID | <[email protected]> |
This is a multi-part message in MIME format. --===============8218081193510590079== Content-Type: multipart/alternative; boundary="------------1i0t7DeBfZ9mDWbrGulWXF7f" Content-Language: fr, en-GB This is a multi-part message in MIME format. --------------1i0t7DeBfZ9mDWbrGulWXF7f Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Hi Samson, I reduced the number of axioms down to 16 920 axioms (getting rid of unnecessary data properties of lexicon entities), still without success with HermiT reasoner for 'or' queries. As you said, Elk reasoner returned quickly the result of "or" queries, which is time-saving for me in the process of exporting (copy/paste) lists of lexical units edited with DL queries. I revised the axioms of 'Lexicalised lemma' mentioned in your last message. Loops should be avoided in the process of editing these lemma in Protégé, along with auxiliary lemmas. I will look in the future for unnecessary axioms but I don't think I will be able to reduce them massively : the aim of the ontology is to generate a stratified lexicon as lexical units issued from class names, controlled entities and data properties through a limited number of knowledge graphs in a the non-specific and specific (non technical, technico-administrative and techno-scientific) semantic fields of "Water, climate and sustainable development", including the graphs in the final edition of the lexicon. The ontology LEXEAU is available on the AgroPortal deposit in Montpellier in an obsolete version of February 2025 (2.2 version). It will be updated at the end of this month (2.3 version) after a complet edition of non redondant conceptual graphs in French. The Word edition of the bilingual Lexicon is available as a pdf file on the site of the French 'Académie de l'Eau' in an obsolete version of February 2025. It will be updated hopefully early next year together with the deposit of 3.0 edition of LEXEAU on AgroPortal. Thanks very much for your help. Best regards, Jean-Louis Janin Le 22/11/2025 à 19:49, Samson Tu a écrit : > Hi Jean-Louis, > > Your problem is not a Protege bug but a reasoner issue. The ontology > has a large number of axioms (702 classes, but 18,865 axioms). The > HermiT reasoner tied itself in knots trying to resolve the query. > Interestingly, using a less expressive reasoner (Elk) designed for a > profile that doesn’t include disjunction returns the query results > quickly. I am not an expert on reasoners or the internals of Protege. > My (wild) guess is that, in the absence of inference results built by > the reasoner, Protege uses some kind of lookup to resolve the simple > disjunctive query. > > (Initially, I thought the problem might be due to limited Java memory > allocation. However, after allocating 8 GB of memory to the process, I > observed that the Protege process did not use more than 3 GB of memory > when it attempted to resolve the HermiT-based query. It simply hung.) > > I am not sure anyone is maintaining the reasoners anymore. They would > be the right people to look into this, not Protege developers. > > /Best regards,/ > > /Samson/ > > > > PastedGraphic-1.tiff > >> On Nov 20, 2025, at 7:52 AM, Jean-Louis Janin >> <[email protected]> wrote: >> >> Hello, >> >> I tried to execute this query (see DL_query_1) for the instances of >> two tables of my ontology (see DL_query_4) but it did not work. After >> some time, I was told that Protégé "does not respond" (see DL_query_2). >> >> Why ? Is it a question of "live" memory ? >> >> I attach the beginning of the report (see DL_3) >> >> Yours, >> >> Jean-Louis Janin >> <DL_query_1.png><DL_query_2.png><DL_query_3.png><DL_query_4.png>_______________________________________________ >> protege-user mailing list >> [email protected] >> https://mailman.stanford.edu/mailman/listinfo/protege-user > --------------1i0t7DeBfZ9mDWbrGulWXF7f Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: 8bit <!DOCTYPE html> <html> <head> <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> </head> <body> <p>Hi Samson,</p> <p>I reduced the number of axioms down to 16 920 axioms (getting rid of unnecessary data properties of lexicon entities), still without success with HermiT reasoner for 'or' queries.</p> <p>As you said, Elk reasoner returned quickly the result of "or" queries, which is time-saving for me in the process of exporting (copy/paste) lists of lexical units edited with DL queries.</p> <p>I revised the axioms of 'Lexicalised lemma' mentioned in your last message. Loops should be avoided in the process of editing these lemma in Protégé, along with auxiliary lemmas. I will look in the future for unnecessary axioms but I don't think I will be able to reduce them massively : the aim of the ontology is to generate a stratified lexicon as lexical units issued from class names, controlled entities and data properties through a limited number of knowledge graphs in a the non-specific and specific (non technical, technico-administrative and techno-scientific) semantic fields of "Water, climate and sustainable development", including the graphs in the final edition of the lexicon.</p> <p>The ontology LEXEAU is available on the AgroPortal deposit in Montpellier in an obsolete version of February 2025 (2.2 version). It will be updated at the end of this month (2.3 version) after a complet edition of non redondant conceptual graphs in French. </p> <p>The Word edition of the bilingual Lexicon is available as a pdf file on the site of the French 'Académie de l'Eau' in an obsolete version of February 2025. It will be updated hopefully early next year together with the deposit of 3.0 edition of LEXEAU on AgroPortal.</p> <p>Thanks very much for your help.</p> <p>Best regards,</p> <p>Jean-Louis Janin </p> <div class="moz-cite-prefix">Le 22/11/2025 à 19:49, Samson Tu a écrit :<br> </div> <blockquote type="cite" cite="mid:[email protected]"> <div>Hi Jean-Louis,</div> <div><br> </div> <div>Your problem is not a Protege bug but a reasoner issue. The ontology has a large number of axioms (702 classes, but 18,865 axioms). The HermiT reasoner tied itself in knots trying to resolve the query. Interestingly, using a less expressive reasoner (Elk) designed for a profile that doesn’t include disjunction returns the query results quickly. I am not an expert on reasoners or the internals of Protege. My (wild) guess is that, in the absence of inference results built by the reasoner, Protege uses some kind of lookup to resolve the simple disjunctive query. </div> <div><br> </div> <div>(Initially, I thought the problem might be due to limited Java memory allocation. However, after allocating 8 GB of memory to the process, I observed that the Protege process did not use more than 3 GB of memory when it attempted to resolve the HermiT-based query. It simply hung.)</div> <div><br> </div> <div>I am not sure anyone is maintaining the reasoners anymore. They would be the right people to look into this, not Protege developers.</div> <div><br> </div> <div> <p> <i>Best regards,</i></p> <p> <i>Samson</i></p> </div> <div><br> </div> <div><br> </div> <img width="890" alt="PastedGraphic-1.tiff" src="cid:AB3F5AE8-06C8-48CC-883C-9B313E79BCF2" moz-do-not-send="true"><br id="lineBreakAtBeginningOfMessage"> <div><br> <blockquote type="cite"> <div>On Nov 20, 2025, at 7:52 AM, Jean-Louis Janin <a class="moz-txt-link-rfc2396E" href="mailto:[email protected]"><[email protected]></a> wrote:</div> <br class="Apple-interchange-newline"> <div> <div>Hello,<br> <br> I tried to execute this query (see DL_query_1) for the instances of two tables of my ontology (see DL_query_4) but it did not work. After some time, I was told that Protégé "does not respond" (see DL_query_2).<br> <br> Why ? Is it a question of "live" memory ?<br> <br> I attach the beginning of the report (see DL_3)<br> <br> Yours,<br> <br> Jean-Louis Janin<br> <span id="cid:53054045-E6A7-46EB-82E8-7A4734A5F0F8"><DL_query_1.png></span><span id="cid:F055C90E-205B-40EB-86F8-0DE80BA58EAD"><DL_query_2.png></span><span id="cid:644520F1-72CC-4BAE-80EF-5A2B432B6635"><DL_query_3.png></span><span id="cid:65CA6681-B910-4438-A62F-17FCFDD4B54C"><DL_query_4.png></span>_______________________________________________<br> protege-user mailing list<br> <a class="moz-txt-link-abbreviated moz-txt-link-freetext" href="mailto:[email protected]">[email protected]</a><br> <a class="moz-txt-link-freetext" href="https://mailman.stanford.edu/mailman/listinfo/protege-user">https://mailman.stanford.edu/mailman/listinfo/protege-user</a><br> </div> </div> </blockquote> </div> <br> </blockquote> </body> </html> --------------1i0t7DeBfZ9mDWbrGulWXF7f-- --===============8218081193510590079== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ protege-user mailing list [email protected] https://mailman.stanford.edu/mailman/listinfo/protege-user --===============8218081193510590079==--