{"id":437,"date":"2025-12-01T15:06:34","date_gmt":"2025-12-01T12:06:34","guid":{"rendered":"https:\/\/www.datuskola.lv\/?p=437"},"modified":"2025-12-01T15:06:36","modified_gmt":"2025-12-01T12:06:36","slug":"ka-maksligajam-intelektam-paredzet-nakotnes-kadastra-vertibas","status":"publish","type":"post","link":"https:\/\/www.datuskola.lv\/index.php\/2025\/12\/01\/ka-maksligajam-intelektam-paredzet-nakotnes-kadastra-vertibas\/","title":{"rendered":"K\u0101 m\u0101ksl\u012bgajam intelektam paredz\u0113t n\u0101kotnes kadastra v\u0113rt\u012bbas?"},"content":{"rendered":"\n

Ceturtdien, 27. novembr\u012b tik\u0101mies uz p\u0113d\u0113jo \u0161\u012b gada domapmai\u0146u, kur\u0101 datu zin\u0101tnieks Aivis Brut\u0101ns par\u0101d\u012bja divas pieejas, k\u0101 m\u0101ksl\u012bgo intelektu (MI) izmantot kadastra v\u0113rt\u012bbu apr\u0113\u0137in\u0101\u0161anai. Viena no t\u0101m bija MCP (Model Context Protocol) izveide, lai lielo valodu modelis atbild\u0113tu uz lietot\u0101ja jaut\u0101jumu balsoties tikai MCP specifik\u0101cij\u0101s. \u0160\u012b metode ieliek modeli konkr\u0113tos r\u0101mjos un \u013cauj veiksm\u012bg\u0101k izvair\u012bties no fantaz\u0113\u0161anas. Tiesa, atbildot uz vienu no uzvedn\u0113m modelis piebilda, ka dotais nekustamais \u012bpa\u0161ums (adrese vai \u0123eogr\u0101fisk\u0101s koordin\u0101tas netika min\u0113tas) ir R\u012bgas centr\u0101. Atbilde bija pareiza, bet pr\u0101toj\u0101m, no kurienes tas \u0146\u0113ma \u0161o inform\u0101ciju. Secin\u0101m, ka \u0161\u012b metode ar\u012b piln\u012bb\u0101 neizsl\u0113dz fantaz\u0113\u0161anas iesp\u0113jam\u012bbu.<\/p>\n\n\n\n

Otra pieeja bija \u201cSkills\u201d funkcija, ko pied\u0101v\u0101 Claude. Tas ir kas l\u012bdz\u012bgs \u201ccustom GPT\u201d. Doma t\u0101da, ka r\u012bks \u013cauj izveidot \u201csp\u0113ju\u201d – konkr\u0113ti nodefin\u0113tas prasmes, saliktas konsekvent\u0101s darb\u012bb\u0101s. Aivis dal\u012bj\u0101s pieredz\u0113, k\u0101 bija izveidot \u0161o sp\u0113ju. Izr\u0101d\u0101s, ka vienk\u0101r\u0161i \u201ciebarot\u201d modelim normat\u012bvos aktus, kas regul\u0113 kadastr\u0101lo v\u0113rt\u012bbu noteik\u0161anu, nesan\u0101ca. Beigu beig\u0101s Aivis izb\u016br\u0101s da\u013cai no procesa un nodefin\u0113ja to pats. Rezult\u0101t\u0101 Claude \u010datbots sniedza diezgan j\u0113gpilnas un izv\u0113rstas atbildes.

Kadastr\u0101l\u0101s v\u0113rt\u012bbas apr\u0113\u0137in\u0101\u0161anai ir daudz da\u017e\u0101di scen\u0101riji (zemes vien\u012bbai, \u0113kai, telpu grupai, in\u017eenierb\u016bvei, zemes vien\u012bbas da\u013cai) un lietot\u0101js var\u0113tu pras\u012bt kadastr\u0101lo v\u0113rt\u012bbu par \u0161iem objektiem da\u017e\u0101d\u0101s kombin\u0101cij\u0101s. T\u0101 k\u0101 ir \u013coti daudz main\u012bgo parametru, kas var ietekm\u0113t kadastr\u0101lo v\u0113rt\u012bbu, tad valodu modelim j\u0101m\u0101k uzdot papildu jaut\u0101jumi, lai ieg\u016btu visu nepiecie\u0161amo inform\u0101ciju apr\u0113\u0137iniem. T\u0101p\u0113c lab\u0101kais scen\u0101rijs b\u016btu lietot\u0101ja pras\u012bbu or\u0137estr\u0113\u0161ana, izmantojot MI a\u0123enta\/darbpl\u016bsmas risin\u0101jumu. MCP gad\u012bjum\u0101 b\u016btu daudz da\u017e\u0101du funkciju, kur da\u017ei v\u0101j\u0101ki MI asistenti ar \u0161o or\u0137estr\u0113\u0161anu var\u0113tu netikt gal\u0101.
Claude Skills ir laba alternat\u012bva a\u0123entiem, bet to liel\u0101kais ierobe\u017eojums ir tas, ka nav iesp\u0113jams izveidot \u0101r\u0113ju savienojumu (API) ar papildu resursiem, lai ieg\u016btu b\u0101zes v\u0113rt\u012bbu, standartplat\u012bbu, sasaist\u012btu adresi ar v\u0113rt\u012bbu zonu u.tml. Un t\u0101 k\u0101 datu apjoms, ko Skills objekt\u0101 var glab\u0101t, ir limit\u0113ts, tad pilna risin\u0101juma izveide iek\u0161 Claude \u201cSkills\u201d b\u016btu izaicino\u0161a.<\/p>\n\n\n\n

T\u0101 k\u0101 daudzi MI entuziasti izmanto NotebookLM, Google risin\u0101jumu, kas \u013cauj pievienot konkr\u0113tus inform\u0101cijas avotus, par kuriem modeli izpra\u0161\u0146\u0101t, Aivis tik\u0161an\u0101s beig\u0101s nokoment\u0113ja ar\u012b to. Vi\u0146a pieredz\u0113 ne katru reizi tas pareizi nolasa v\u0113rt\u012bbas, vai piem\u0113ro pareizu vien\u0101dojumu.<\/p>\n\n\n\n

\"\"
Piem\u0113rs no NotebookLM atbildes, balstoties normat\u012bvajos aktos.<\/figcaption><\/figure>\n\n\n\n

\u201cVizu\u0101li apraksts izskat\u0101s iespaid\u012bgi – s\u012bki un detaliz\u0113ti aprakst\u012bts, ar atsauc\u0113m,\u201d st\u0101sta Aivis, \u201cTa\u010du t\u0101 atbild\u0113 vajadz\u0113tu s\u012bk\u0101k iedzi\u013cin\u0101ties – j\u0101saprot vai funkcija izmantota pareiza, vai izmantoti visi funkcijas parametri, vai parametru v\u0113rt\u012bbas ir noteiktas prec\u012bzi. \u0160aj\u0101 piem\u0113r\u0101 rezult\u0101ts ir gandr\u012bz pareizs. T\u0101 k\u0101 pras\u012bju par dz\u012bvok\u013ca telpu grupas v\u0113rt\u012bbu, nor\u0101dot, ka dz\u012bvoklis atrodas 1.st\u0101v\u0101, tad tas ir v\u0113rt\u012bbu samazino\u0161s faktors, kas netika iek\u013cauts. Visas p\u0101r\u0113j\u0101s v\u0113rt\u012bbas un apr\u0113\u0137ins ir pareizs.\u201d<\/p>\n\n\n\n

\u0160\u012b tik\u0161an\u0101s notika ar Valsts zemes dienesta (VZD) atbalstu vi\u0146u m\u0101j\u012bgaj\u0101s, m\u016bsdien\u012bgaj\u0101s telp\u0101s. Paldies! N\u0101kam\u0101 domapmai\u0146a notiks janv\u0101ra beig\u0101s kl\u0101tien\u0113, bet ar iesp\u0113ju v\u0113rot prezent\u0101ciju tie\u0161saist\u0113. Starp citu, tie\u0161saistes dal\u012bbas iesp\u0113ju izmantoja vair\u0101k nek\u0101 30 dal\u012bbnieku. Sekojiet l\u012bdzi inform\u0101cijai Facebook, X un datuskola.lv.<\/p>\n","protected":false},"excerpt":{"rendered":"

Ceturtdien, 27. novembr\u012b tik\u0101mies uz p\u0113d\u0113jo \u0161\u012b gada domapmai\u0146u, kur\u0101 datu zin\u0101tnieks Aivis Brut\u0101ns par\u0101d\u012bja divas pieejas, k\u0101 m\u0101ksl\u012bgo intelektu (MI) izmantot kadastra v\u0113rt\u012bbu apr\u0113\u0137in\u0101\u0161anai. Viena no t\u0101m bija MCP (Model Context Protocol) izveide, lai lielo valodu modelis atbild\u0113tu uz lietot\u0101ja jaut\u0101jumu balsoties tikai MCP specifik\u0101cij\u0101s. \u0160\u012b metode ieliek modeli konkr\u0113tos r\u0101mjos un \u013cauj veiksm\u012bg\u0101k […]<\/p>\n","protected":false},"author":1,"featured_media":439,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[14,11,16,17],"tags":[],"_links":{"self":[{"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/posts\/437"}],"collection":[{"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/comments?post=437"}],"version-history":[{"count":1,"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/posts\/437\/revisions"}],"predecessor-version":[{"id":440,"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/posts\/437\/revisions\/440"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/media\/439"}],"wp:attachment":[{"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/media?parent=437"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/categories?post=437"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.datuskola.lv\/index.php\/wp-json\/wp\/v2\/tags?post=437"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}