[00:42:30] New patchset: Jqnatividad; "set SMWDIProperty show flag of subobjects to false to suppress display in factbox. Doesn't make sense to show subobjects in factbox since the display name is non-unique." [mediawiki/extensions/SemanticMediaWiki] (1.8.x) - https://gerrit.wikimedia.org/r/64878 [12:37:02] Change merged: Yaron Koren; [mediawiki/extensions/SemanticDrilldown] (master) - https://gerrit.wikimedia.org/r/64535 [12:39:53] Change merged: Yaron Koren; [mediawiki/extensions/SemanticDrilldown] (master) - https://gerrit.wikimedia.org/r/64536 [12:40:12] Change merged: Yaron Koren; [mediawiki/extensions/SemanticDrilldown] (master) - https://gerrit.wikimedia.org/r/64537 [13:19:58] New patchset: Yaron Koren; "Fixes for new logarithmic tag cloud display" [mediawiki/extensions/SemanticDrilldown] (master) - https://gerrit.wikimedia.org/r/64925 [13:21:15] Change merged: Yaron Koren; [mediawiki/extensions/SemanticDrilldown] (master) - https://gerrit.wikimedia.org/r/64925 [13:49:22] New review: Jeroen De Dauw; "Does what it says it does. I'm however not sure this makes sense from a user perspective as I'm not ..." [mediawiki/extensions/SemanticMediaWiki] (1.8.x) C: 1; - https://gerrit.wikimedia.org/r/64878 [13:56:24] New patchset: Yaron Koren; "More fixes for removal of '_str' datatype in SMW 1.9" [mediawiki/extensions/SemanticDrilldown] (master) - https://gerrit.wikimedia.org/r/64928 [14:27:40] New patchset: Mwjames; "Fix fatal when predefined properties are no longer exists" [mediawiki/extensions/SemanticMediaWiki] (master) - https://gerrit.wikimedia.org/r/64938 [16:30:20] I need some advice about Semantic Mediawiki. [16:30:41] Say, we have a number of relations written in this way: [16:31:18] [[user::Fomenko]] [[reader::Fomenko]] [[driver::Ivanov]] [[driver::Fomenko]] [[writer::Ivanov]] [16:32:40] I may write and #ask query to find items by their property. This shows me that: user is --> Fomenko, driver is --> Fomenko and Ivanov, writer is --> Ivanov. [16:33:23] May I reverse the query? May I ask what properties does the item have? I would like to see that Fomenko <-- is user and reader and driver. [16:33:55] Dear people, could you please share some advice with me? Thanks. [17:12:26] fomenko: if you know the properties, you can get their values [17:12:33] the command is #show [17:12:55] {{#show: |?user}} [17:13:00] New review: Nischayn22; "(1 comment)" [mediawiki/extensions/SemanticMediaWiki] (1.8.x) - https://gerrit.wikimedia.org/r/64878 [17:13:06] Yes. But my question is about the inversed way. I know the value -- how can I get the properties? [17:13:23] you can't [17:13:41] Your #show command will show me who are users. I want the opposite: [17:13:56] what properties the value has? [17:14:37] So I need to make two relations at once? [[reader::Fomenko]] and [[Fomenko::reader]] in the same place in text? [17:15:16] well it's ugly but it works [17:15:33] but you should not need it at all [17:15:49] apparently your information model is incomplete [17:16:49] Let's discuss more general view. I have a long historical text. It tells me about different people holding different roles. I want to format it in the way to logically see the relations between people and their roles. [17:17:16] that sounds reasonable, and you can do that [17:18:32] For example: [17:18:33] Sir Alexandrov, being a head of the council, raised the discussion about city streets condition. He said, it would be a great help for the city to arrange paving the streets with stone. [17:18:44] This is how I see the markup: [17:19:37] Sir [[speaker::Alexandrov|Alexandrov]], being a head of the [[head::council|council]], raised the discussion about city streets condition. He said, it would be a great help for the city to arrange [[pavement::streets|paving]] the streets with stone. [17:19:54] that won't work. [17:20:12] your relations are not between parts of the sentence, but between the page as a whole, and other pages [17:20:36] so if this sentence is on page Longtext, then your saying in the above things like [17:20:37] My relations are between historical facts explained in text. [17:20:45] Longtext --speaker--> Alexandrov [17:21:02] Longtext --head--> Council [17:21:04] I don't want to find pages by facts. I want to find facts by facts, to find the relations between facts in text. [17:21:37] I don't see how that would be possible without some fact-discovering parser [17:21:46] The parser is myself. [17:21:54] From the example text I want to know: [17:22:37] who were speakers? what role Alexandrov played? what is known about the streets? [17:22:54] for that you need some text interpreter, and SMW isn't it [17:23:33] If you wanted to ask SMW these questions, you'd have to cut up the text into separate pages, each representing a bit of knowledge from the text [17:23:36] speaker::Alexandrov helps me query that Alexandrov was among speakers. But how can I search about that man's properties and roles? [17:23:57] Sorry, cannot catch. Why cut the text? [17:24:25] because SMW works with knowledge about (the subject of) the page [17:25:09] in your example, Alexandrov spoke about the street [17:25:16] you know that, I know that, SMW doesnt [17:25:35] SMW can't see if Alexandrov spoke about the council, the street, or something else [17:25:37] Another example. The text about the Council lists the people taking part in that Council. Markup is, member::Alexandrov, member::Gusev, member::Glinsky. So I can get the list of members from this text. [17:25:48] no you can't [17:25:59] because they aren't member of the council [17:26:14] you're creating Longtext --member--> Alexandrov [17:26:21] They are marked up with those properties. [17:26:24] yes [17:26:40] but the relation is between the page and the person, not between a specific portion of your text and the person [17:26:59] Sample: "In the Council stayed present: [[member::Alexandrov|Sir Alexandrov]], [[member::Gusev|Sir Gusev]]. [17:27:00] SMW doesn't know that relation Member pertains to the word Council [17:27:15] it knows that relation Member originates from page Longtext [17:27:36] I want to duscuss relation between "Alexandrov" and "member", that is marked as member::Alexandrov. [17:27:52] yes, but there's no "council" in sight [17:27:54] This helps me to get the list of members, with Alexandrov in this list. [17:28:12] yes, but you've annotated that Alexandrov is a member of Longtext [17:28:29] I think you don't have an information model at all [17:28:32] Don't mind about the Council, it doesn't act now. I annotated that Alexandrov is simply a member of something. [17:29:17] then you can ask {{#ask: [[member::Alexandrov]]}} and it will give you all pages where Alexandrov is annotated as a member [17:29:38] Will it show me that Alexandrov was annotated as member? [17:29:56] indirectly it already does [17:29:56] Also, may the query find all pages where Alexandrov was annotated as something? [17:30:03] no you can't do that [17:30:18] afaik [17:30:19] That's the main problem. I want bidirectional relations. [17:30:35] they are bidirectional by means of the concept of "inverse relations" [17:31:00] If I see Alexandrov is member, I'd like to markup that he was member, so I could query for the list of members and for the properties of Alexandrov. [17:31:32] you can, but you have to already know that you're looking for [[member:: ]] [17:31:34] In other words, search for properties with the given value. [17:31:41] you can't do that [17:32:22] This means I will look for members. Or will search for the page where Alexandrov is member. Yes, this is obvious. And my idea is, how to find WHOM that Alexandrov was. [17:32:40] In the 1st step I will markup his roles, in the 2nd I'd search for those roles. [17:32:44] again, you already have to know the roles [17:33:20] to be clear, you cannot ask [[*::alexandrov]] [17:33:30] you CAN ask [[member::alexandrov]] [17:34:21] Yes, the roles are stated in text somewhere. They are explained in text and they are marked up. Somewhere is [[member::Alexandrov]], or [[death::Alexandrov]], or [[job::Alexandrov]]. The fantasy is to find the places about Alexandrov and to show there are places about his death, job, and membership. [17:34:48] you can do that as long as you have the list of properties that you want to check for Alexandrov [17:35:13] they're all of the form {{#ask: [[death::Alexandrov]]}} etc [17:35:34] So at first I have to know that list — to know what properties are attached to him. As you said, this cannot be queried. [17:35:41] correct [17:35:54] but you can go to the page and look at the factbox at the bottom [17:36:00] Too bad for me. [17:36:19] well it's exactly like that in the real world [17:36:25] Factbox at the bottom tells me to find a program to query those factbosex :-) [17:36:36] factboxes, sorry :-) [17:36:40] if I show you a vase, you cannot query what properties it has [17:37:04] you have to ask yourself (or somebody else) "does it have the property "weight" and then the answer is yes [17:37:24] If you show me the text about vase with marked properties, I may search for those properties. I blindly hoped semantic wiki may search for the properties. [17:38:05] You see, this is not about real thing, but about the story of thing. About the properties shown in the story. [17:38:20] I got that [17:38:44] It still appears to me that you try to have SMW automatically extract meaning [17:39:54] SMW to automatically extract the marked meaning from the market text. [17:40:05] yes, it cannot do that [17:40:19] it extracts facts, not their meaning [17:40:23] Not to find the meaning in the raw text, but to find the annotated meaning from the annontated pages. [17:40:44] again, SMW can extracts facts, not their meaning [17:40:56] if you don't know what you're looking for, SMW can't find it for you [17:41:11] Example: [[member::Alexandrov]]. There are 2 facts: Alexandrov was in "member" group and "member" group included Alexandrov. [17:41:25] neither is true [17:41:31] Why? [17:41:36] there is no group "member" [17:41:55] there is a relation between the page on which the annotation sits, and the page Alexandrov [17:41:57] The property "member" means "the one in 'member' group" [17:42:01] and the relation is named "member" [17:42:40] So there are 3 pages: text, Alexandrov, member. And there have to be 2 relations: [[member::Alexandrov]] and [[Alexandrov::member]]. [17:43:04] i'm getting confused by the examples [17:43:13] what is on the page "member" [17:43:30] and what is the relation between the page "member" and the relation "member", according to you? [17:43:49] The page "member" tells there were members, and lists the people who were members. [17:44:02] The page "Alexandrov" tells there was a man, and lists his roles. [17:44:13] ok [17:44:22] The page "text" tells us the document about Sir Alexandrov being a member of something. [17:45:06] have you tried creating a simple semantic wiki, and recording some information that's less complicated than this? [17:45:10] With [[member::Alexandrov]] I may #ask on the page "member" and get the list of members, and the "text" pages about their being members. [17:45:15] it sounds like you're mixing up all sorts of concepts [17:45:47] your page "member" duplicates the information that SMW already shows on page Property:member [17:46:37] Yes, the page "member" duplicates the page "Property:member". [17:46:47] say, may I suggest to you the book by Yaron: http://workingwithmediawiki.com/ [17:47:01] it has concise but rather complete information on SMW [17:47:23] I'm pretty sure I can't help you with your task [17:47:41] So, generally speaking, I hope to make all values to be properties. When I read about Alexandrov being a member, there are two relations -- between group and its element. [17:50:33] You were correct saying I am not asking about semantics of wiki pages. Yes: I am thinking about semantics of the text in those wiki pages. [17:50:49] that is a really, really, REALLY tough nut to crack [17:51:28] The plain attempt I see now is to write 2 relations, member::Alexandrov and Alexandrov::member. [17:52:03] that is NOT the answer to your problem. [17:52:08] Why? [17:52:46] I really cannot understand the complete logic of your argument. [17:52:49] if the second statement is on the same page as the first one, you're creating properties that correspond directly to pages. Your wiki would explode in size with each page and with each proeprty you'd add [17:53:09] if the second statement is on the Alexandrov page, then you're not creating a semantic wiki at all [17:53:34] furthermore, the second statement can already be interpreted by means of "inverse property", i.e. [[-member:Alexandrov]] [17:53:36] Well, as far as I understood, semantic wiki = wiki with semantic markup. [17:54:02] well that's only true up to a point [17:54:13] No-no, the 2nd statement is on the same page. Example goes: [17:54:46] In the Council stayed present: [[member::Alexandrov|Sir Alexandrov]], [[member::Gusev|Sir Gusev]] [[Gusev::member]] [[Alexandrov::member]] [17:54:57] fomenko: please stop explaining it to me, my head starts to hurt [17:55:12] So am I :-) [17:55:14] SMW can handle 3 things for you: classes, attributes and relations [17:55:36] it cannot interpret, it can't handle much negation, it cannot magically find meaning [17:55:56] I will markup the meanings. [17:56:10] you can't markup meaning [17:56:14] you can markup facts [17:56:19] and the markup IS the meaning [17:56:32] and the wiki doesn't know the meaning, the users do (ideally) [17:57:10] and you can only add classes, attributes of pages, and relation between pages. [17:57:32] In this way I don't want to search for meanings. I want to query symmetric facts. Relations between attributes. [17:57:33] (well there are subobjects and other trickeries, but still...) [17:58:18] fomenko: read http://semantic-mediawiki.org/wiki/Inverse_Properties [17:58:55] furthermore, if page A has relation B with page C, then you can ALWAYS find on page C which pages have relation B with it by querying [17:59:18] but you cannot find what relations there are to page C [18:01:32] The thing "I cannot find" is the aim. Woe to me. [18:02:27] Yes, I wanted to find relations. So my task is, to convert relations into pages. This can be done, for the relations (in this sample) are facts, and each fact produces its page. [18:02:28] I'd suggest not continuing on this path, but go back and rethink your problem [18:03:06] The problem is, to markup the semantics of the document, so it would be easier to work with its facts. [18:03:26] Something like a developed book index. [18:03:45] everything you've told me tells me that SMW won't be able to help you [18:04:02] but ofcourse I could be wrong :-) [18:04:15] What problem is about two relations in one text? Example: [18:04:43] In the Council stayed [[member::Alexandrov|Sir Alexandrov]]. [[Alexandrov::member|]] [18:05:27] I've explained all I can. [18:05:57] I've suggested some reasonable courses of actions, I can do no more [18:06:32] As far as I understood, the action is not to use SMW. [18:08:37] well let's just say "yes" to that :-/ [18:12:23] OK, let's try other way. Is there a query that gives the pages with some value? — Markup goes: [[member::Alexandrov]], value is [[Alexandrov]]. — How to query those links? [18:12:57] {{#ask: [[member::Alexandrov]]}} [18:14:49] And for any [[Alexandrov]], member:: or not? [18:20:41] {{#ask: [[member::Alexandrov]]|format=count}} [18:20:56] would return 0 if no page has the "member" relation to Alexandrov [18:25:56] I mean, a query to find any links to [[Alexandrov]] not caring about properties? [18:26:21] I. e. not caring about relations? [18:29:28] you cannot [18:32:24] Saruman: Well, if you collected additional data, you could. You'd have to create a subobject for every relation {{#subobject:has relation|predicate=member|object=Alexandrov}} [18:32:42] Somewhat cumbersome, thoguh [18:32:48] "somewhat"? [18:32:52] :D [18:33:57] my mantra in cases like these is either "you don't have your information model thought out" or "you're trying to solve the wrong problem the wrong way" [18:34:50] Saruman: Or with the wrong tool anyway. Although being able to query properties would be nice [18:35:04] Ftrott: this just occurs to me: what if you create EVERY relation of type page a subproperty of relation "has relation"? [18:35:30] Saruman: What do you mean? [18:35:52] hmm that'd give you all pages that have _a_ relation to your value, but not _what_ relation [18:36:00] ok never mind [18:40:37] New patchset: Jqnatividad; "set SMWDIProperty show flag of subobjects to false to suppress display in factbox. Doesn't make sense to show subobjects in factbox since the display name is non-unique and is the name of the current page. If there are several subobjects in the current p" [mediawiki/extensions/SemanticMediaWiki] (1.8.x) - https://gerrit.wikimedia.org/r/64878 [18:41:39] New patchset: Jqnatividad; "set SMWDIProperty show flag of subobjects to false to suppress display in factbox. Doesn't make sense to show subobjects in factbox since the display name is non-unique and is the current page name. If there are several subobjects in the current page, i" [mediawiki/extensions/SemanticMediaWiki] (1.8.x) - https://gerrit.wikimedia.org/r/64878 [18:47:46] New patchset: Mwjames; "SMW\ParserTextProcessor add test and improve CRAP index" [mediawiki/extensions/SemanticMediaWiki] (master) - https://gerrit.wikimedia.org/r/64969 [18:48:00] Dear Saruman, thanks for the explanations and discussion. [18:48:09] you're welcome [18:48:19] just trying to help [18:48:20] Elen sila lumenn omentielvo. [18:48:28] eh.. if you say so... [18:48:37] je bent meer dan welkom, graag gedaan [18:55:59] New patchset: Mwjames; "SMW\ParserTextProcessor add test and improve CRAP index" [mediawiki/extensions/SemanticMediaWiki] (master) - https://gerrit.wikimedia.org/r/64969 [19:06:24] Change merged: jenkins-bot; [mediawiki/extensions/SemanticMediaWiki] (master) - https://gerrit.wikimedia.org/r/64969 [19:29:47] New review: Jeroen De Dauw; "How did you find the CRAP index? Ran PHPUnit coverage?" [mediawiki/extensions/SemanticMediaWiki] (master) - https://gerrit.wikimedia.org/r/64969 [19:49:18] New review: Mwjames; "Yes exactly (PHP_CodeCoverage 1.2.7 using PHP 5.4.7 and PHPUnit 3.7.13) but unfortunately it takes a..." [mediawiki/extensions/SemanticMediaWiki] (master) - https://gerrit.wikimedia.org/r/64969 [21:27:26] New review: Jeroen De Dauw; "Why not just run the coverage for SMW?" [mediawiki/extensions/SemanticMediaWiki] (master) - https://gerrit.wikimedia.org/r/64969 [23:01:48] Hi! I'm trying to get the gallery working like in http://semantic-mediawiki.org/wiki/Demo:Berlin. But the caption/description thingy isn't behaving. http://wiki.ultimacodex.com/forever/User:Fenyx4/Sandbox [23:04:58] I've installed Semantic Result Formats and Widgets in an attempt to get it working. And although it helped I'm still not quite there...