{"id":218,"date":"2014-12-27T23:46:28","date_gmt":"2014-12-27T22:46:28","guid":{"rendered":"http:\/\/www.mlguru.cz\/?p=218"},"modified":"2017-08-12T15:39:04","modified_gmt":"2017-08-12T13:39:04","slug":"analyza-prezidentskych-projevu","status":"publish","type":"post","link":"https:\/\/www.mlguru.com\/cs\/analyza-prezidentskych-projevu\/","title":{"rendered":"Anal\u00fdza prezidentsk\u00fdch projev\u016f"},"content":{"rendered":"

\u010cesk\u00fd rozhlas v\u010dera na sv\u00e9m webu<\/a>\u00a0zve\u0159ejnil p\u0159episy v\u00e1no\u010dn\u00edch a novoro\u010dn\u00edch projev\u016f v\u0161ech \u010deskoslovensk\u00fdch a \u010desk\u00fdch prezident\u016f od roku 1934 do sou\u010dasnosti. Bezesporu se jedn\u00e1 o velmi zaj\u00edmav\u00e1 data, kter\u00e1\u00a0si zaslou\u017e\u00ed anal\u00fdzu. Prvn\u00ed vla\u0161tovkou byl\u00a0Michal Dvo\u0159\u00e1k<\/a>, kter\u00fd si v\u0161iml, \u017ee n\u00e1\u0161 sou\u010dasn\u00fd prezident ve sv\u00e9m projevu lehce\u00a0nadu\u017e\u00edv\u00e1 prvn\u00ed osobu singul\u00e1ru, co\u017e demonstroval jednoduchou statistikou<\/a>. Frekven\u010dn\u00ed anal\u00fdza, kterou pou\u017eil, je jist\u011b mocn\u00fd n\u00e1stroj, ale nedok\u00e1\u017ee odhalit slo\u017eit\u011bj\u0161\u00ed z\u00e1konitosti. Proto jsem se nad korpusem projev\u016f rozhodl prov\u00e9st Latentn\u00ed S\u00e9mantickou Anal\u00fdzu<\/a>.<\/p>\n

Nejprve jsem dal dohromady v\u0161echny texty jednotliv\u00fdch prezident\u016f, ty rozd\u011blil na slova a\u00a0zlemmatizoval<\/a>\u00a0je pomoc\u00ed n\u00e1stroje MorphoDiTa<\/a>. T\u00edm vzniklo 11 dokument\u016f, ka\u017ed\u00fd reprezentuj\u00edc\u00ed texty jednoho prezidenta. Z\u00e1jemci si je m\u016f\u017eou st\u00e1hnout zde<\/a>.\u00a0Na ka\u017ed\u00fd dokument jsem d\u00e1le aplikovat frekven\u010dn\u00ed normalizaci TF-IDF<\/a> a v\u00fdsledek transformoval do dvoudimenzion\u00e1ln\u00edho prostoru\u00a0pomoc\u00ed latentn\u00ed s\u00e9mantick\u00e9 anal\u00fdzy (LSA).<\/p>\n

V\u00fdsledek je zn\u00e1zorn\u011bn na n\u00e1sleduj\u00edc\u00edm obr\u00e1zku.<\/p>\n

\"lsa\"<\/a><\/p>\n

Je t\u011b\u017ek\u00e9 smyslupln\u011b interpretovat jednotliv\u00e9 osy grafu, m\u016f\u017eeme se v\u0161ak d\u00edvat na vzd\u00e1lenosti prezident\u016f ve vznikl\u00e9m s\u00e9mantick\u00e9m prostoru. Z grafu je vid\u011bt, \u017ee algoritmus bezpe\u010dn\u011b odd\u011blil demokratick\u00e9 prezidenty od nedemokratick\u00fdch, kte\u0159\u00ed vytvo\u0159ili shluk v prav\u00e9 horn\u00ed \u010d\u00e1sti grafu. Mezi demokratick\u00fdmi prezidenty stoj\u00ed trochu stranou T. G. Masaryk. To m\u016f\u017ee b\u00fdt d\u00e1no t\u00edm, \u017ee byl u n\u011bho k dispozici pouze jeden velmi kr\u00e1tk\u00fd projev z roku 1934 a data jsou tak ovlivn\u011bna \u0161umem. Velmi zaj\u00edmav\u00e9 je postaven\u00ed M. Zemana a E. H\u00e1chy, kte\u0159\u00ed spolu tvo\u0159\u00ed samostatn\u00fd shluk. Stoj\u00edme tedy p\u0159ed ot\u00e1zkou, co maj\u00ed tito dva prezidenti spole\u010dn\u00e9ho. Na LSA je p\u011bkn\u00e9 to, \u017ee dok\u00e1\u017ee zachytit i vztahy, kter\u00e9 v textech v\u016fbec nejsou explicitn\u011b zm\u00edn\u011bn\u00e9, a d\u00e1v\u00e1 tak prostor pro zamy\u0161len\u00ed. Co je tedy spojuje?<\/p>\n","protected":false},"excerpt":{"rendered":"

\u010cesk\u00fd rozhlas v\u010dera na sv\u00e9m webu\u00a0zve\u0159ejnil p\u0159episy v\u00e1no\u010dn\u00edch a novoro\u010dn\u00edch projev\u016f v\u0161ech \u010deskoslovensk\u00fdch a \u010desk\u00fdch prezident\u016f od roku 1934 do sou\u010dasnosti. Bezesporu se jedn\u00e1 o velmi zaj\u00edmav\u00e1 data, kter\u00e1\u00a0si zaslou\u017e\u00ed anal\u00fdzu. Prvn\u00ed vla\u0161tovkou byl\u00a0Michal Dvo\u0159\u00e1k, kter\u00fd si v\u0161iml, \u017ee n\u00e1\u0161 sou\u010dasn\u00fd prezident ve sv\u00e9m projevu lehce\u00a0nadu\u017e\u00edv\u00e1 prvn\u00ed osobu singul\u00e1ru, co\u017e demonstroval jednoduchou statistikou. Frekven\u010dn\u00ed anal\u00fdza, […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_exactmetrics_skip_tracking":false,"_exactmetrics_sitenote_active":false,"_exactmetrics_sitenote_note":"","_exactmetrics_sitenote_category":0,"footnotes":""},"categories":[3],"tags":[],"class_list":["post-218","post","type-post","status-publish","format-standard","hentry","category-analyza-dat"],"_links":{"self":[{"href":"https:\/\/www.mlguru.com\/cs\/wp-json\/wp\/v2\/posts\/218","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.mlguru.com\/cs\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.mlguru.com\/cs\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.mlguru.com\/cs\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.mlguru.com\/cs\/wp-json\/wp\/v2\/comments?post=218"}],"version-history":[{"count":13,"href":"https:\/\/www.mlguru.com\/cs\/wp-json\/wp\/v2\/posts\/218\/revisions"}],"predecessor-version":[{"id":432,"href":"https:\/\/www.mlguru.com\/cs\/wp-json\/wp\/v2\/posts\/218\/revisions\/432"}],"wp:attachment":[{"href":"https:\/\/www.mlguru.com\/cs\/wp-json\/wp\/v2\/media?parent=218"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.mlguru.com\/cs\/wp-json\/wp\/v2\/categories?post=218"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.mlguru.com\/cs\/wp-json\/wp\/v2\/tags?post=218"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}