ÊÌ.RU íà ÐÎÌÈÏ-2008. Îïòèìèçàöèÿ ïàðàìåòðîâ ïîèñêîâîãî àëãîðèòìà


   www.romver.ru
/ Ïîëíûé ñïèñîê ñòàòåé / ÊÌ.RU íà ÐÎÌÈÏ-2008

Êàê çàêàçàòü ñàéò


ÀÁÐÀÊÀÄÀÁÐÀ (Òîæå ñàìîå íî â ÷èòàåìîì âèäå)

 cnfnmt jgbcfys vjlbabrfwbb fkujhbnvf byajhvfwbjyyjuj gjbcrf, ghtlcnfdktyyjuj ÊÌ.RU yf ctvbyfht ÐÎÌÈÏ-2007, b cbcntvf jgnbvbpfwbb rj'aabwbtynjd lkz jghtltktyyjuj yf,jhf gfhfvtnhjd, cjplfyyfz c wtkm. gjkextybz yfbkexib[ htpekmnfnjd d gjbcrt b hfy;bhjdfybb ljrevtynjd. Îcdtof.ncz htpekmnfns exfcnbz ghjtrnf «Ïjbcr ÊÌ.RU» d ctvbyfht ÐÎÌÈÏ-2008. Î,ce;lftncz lfkmytqibq genm hfpdbnbz ghjtrnf.
 1. Âdtltybt Ðf,jnf yfl ekexitybtv rfxtcndf gjbcrf b hfy;bhjdfybz ljrevtynjd j,sxyj cdzpfyf c gjzdktybtv yjds[ afrnjhjd, rjnjhst exbnsdf.ncz ghb hf,jnt c ljrevtynfvb d jndtn yf pfghjc. Ìs gjcnfhfkbcm cjplfnm lkz ct,z elj,ysq fdnjvfnbxtcrbq cgjcj, jghtltktybz pyfxbvjcnb jghtltktyys[ afrnjhjd ghb j,hf,jnrt pfghjcf, f nfr;t cgjcj, jgnbvbpfwbb gfhfvtnhjd yfituj gjbcrjdjuj fkujhbnvf. Â rfxtcndt ,fps lkz jgnbvbpfwbb vs bcgjkmpjdfkb 1) vfnthbfks ghjikjujlytuj ctvbyfhf ÐÎÌÈÏ: gfhs pfghjc-ljrevtyn b jwtyrb 'rcgthnjd rfr vths cjjndtncndbz ljrevtynf pfghjce, 2) ,fpe cj,cndtyys[ jwtyjr. Çflfxf, rjnjhfz cnjzkf gthtl yfvb d 'njv ujle: jghtltkbnm pyfxbvjcnm yjds[ afrnjhjd, f nfr;t dszcybnm, gjvjuftn kb jgnbvbpfwbz yfcnhjtr ekexibnm htpekmnfns hf,jns cbcntvs. Ïjlj,ye. hf,jne d 2005 ujle e;t ghjdjlbkb exfcnybrb ctvbyfhf ÐÎÌÈÏ-2005 [4], b b[ htpekmnfns ujdjhbkb j ghtbveotcndt jgnbvbpbhjdfyyjuj fkujhbnvf.

2. Õfhfrnthbcnbrf rjkktrwbq Ñvsck ljrevtynf vj;tn ,snm jwtyty gj nhtv cjcnfdkz.obv:

1. Çfujkjdjr (Title).

2. Òtrcn ljrevtynf.

3. Ñcskrb yf ljrevtyn.

Èc[jlz bp 'njuj, jgbitv cdjqcndf rjkktrwbq:

1. Legal – yfb,jkmifz cntgtym cnhernehbhjdfyyjcnb. Ôfrnbxtcrb, pfujkjdjr buhftn yfb,jktt ceotcndtyyjt pyfxtybt d jghtltktybb htktdfynyjcnb ljrevtynf. Ñcskrb ljcnjdthys.

2. KM.RU – cnhernehbhjdfyf, yj vtymit xtv legal. Äjrevtyns vjuen cjlth;fnm byajhvfwb., yt jnyjczoe.cz r jcyjdyjq ntvt ljrevtynf (ntreobt yjdjcnb b n.g.). Çfujkjdjr j,sxyj ljcnjdthty. Ñcskrb j,sxyj ljcnjdthys.

3. BY.WEB – yfbvtymifz cntgtym cnhernehbhjdfyyjcnb. Äjrevtyns vjuen cjlth;fnm byajhvfwb., yt jnyjczoe.cz r jcyjdyjq ntvt ljrevtynf. Çfujkjdjr vj;tn ,snm ytljcnjdthysv. Ñcskrb vjuen ,snm ytljcnjdthys.

Äkz rf;ljq cjcnfdkz.otq cvsckf ljrevtynf vj;yj ghjdtcnb hzl jgthfwbq, lf.ob[ yfb,jktt 'aatrnbdye. jwtyre dtcf ljrevtynf. Ïhbdtltv rkfccbabrfwb. cjdhtvtyys[ vtnjljd dscnfdktybz dtcf ljrevtyne.

1. Ïhjbpdtltybt xfcnjnf ckjdf bp pfghjcf d ljrevtynt – TF b dcnhtxftvjcnb ckjdf d rjkktrwbb – IDF. ×tv ,jkmit cnfnbcnbrf, ntv (d j,otv ckexft) jyf kexit.

2. Îwtyrf hfccnjzybz vt;le ckjdfvb pfghjcf. Ítrfz aeyrwbz jn hfccnjzybz – F (hfccnjzybt). Ïfccf;b (gjkyst, ytgjkyst, c extnjv gjhzlrf ckjd b ,tp, c hfccnjzybtv vt;le ckjdfvb b ,tp), gfhs ckjd (c njxysv gjhzlrjv ckjd, c j,hfnysv gjhzlrjv ckjd, cj ckjdfvb vt;le gfhjq ckjd).

3. Êjkbxtcndtyyst jwtyrb – dtc ljrevtynf b dtc ccskjr, hfccxbnfyysq gj ccskjxyjve hfyue.


Âct erfpfyyst vtnjls vjuen ,snm ghbvtytys lkz rjkbxtcndtyyjq jwtyrb 3-[ cvsckjds[ xfcntq ljrevtynf – pfujkjdrf, ntrcnf b ccskjr yf ljrevtyn. Ñjjndtncndtyyj, ghb nfrjv gjl[jlt, yfghbvth, vj;yj jnltkmyj jwtybdfnm gfccf;b d pfujkjdrf[, ntrcnt ljrevtynf b ccskrf[ yf ljrevtyn. Òfv ;t exbnsdfnm b gfhs ckjd. Órfpfyyst vtnjls ujhfplj kexit hf,jnf.n yf [jhjij cnhernehbhjdfyys[ rjkktrwbz[ b ytljcnfnjxyj [jhjij yf gkj[j cnhernehbhjdfyys[. 3. Ìjlbabrfwbz fkujhbnvf gjbcrf b hfy;bhjdfybz ljrevtynjd. Àdnjvfnbxtcrfz jgnbvbpfwbz rj'aabwbtynjd 3.1 Ñnhernehf fkujhbnvf f) ,fpjdfz ajhvekf; Çf jcyjde vs dpzkb ajhveke, hf,jnf rjnjhjq ,skf ghjltvjycnhbhjdfyf d ghjikjv ujle [2]. Ýnjn fkujhbnv c ytrjnjhsvb ekexitybzvb ctqxfc hf,jnftn yf gjhnfkt ÊÌ.RU. ,) vjlbabrfwbb ajhveks. Äj,fdktybt yjds[ afrnjhjd, fdnjvfnbxtcrfz jgnbvbpfwbz rj'aabwbtynjd. Äkz dsxbcktybz htktdfynyjcnb ljrevtynf pfghjce vs bcgjkmpjdfkb cktle.oe. pfdbcbvjcnm: W = k1*W1 + k2*W2 + k3*W3 + k4*W4(1), ult W – bnjujdjt pyfxtybt htktdfynyjcnb ljrevtynf. k1, k2, k3, k4 – rj'aabwtyns. Îcnfyjdbvcz gjlhj,ytt yf rf;ljv bp ckfuftvs[. W1 = TF*IDF(l) * F1(DocWeight), ult: TF*IDF(l) dsxbckztncz gj tfd(l) = freqd(l)/( freqd(l) + 0.5 + 1.5*dld/avg_dl)

freqD(l) - xfcnjnyjcnm ktvvs l d ljrevtynt, dlD – vthf lkbys ljrevtynf, avg_dl – chtlyzz lkbyf ljrevtynf (lkz rjkktrwbq


BY.WEB b KM.RU vs bcgjkmpjdfkb pyfxtybt avg_dl = 400, lkz rjkktrwbb Legal = 1500), IDF - “Inverse term frequency” - ajhvf inhfajdfybz xfcnj bcgjkmpetvs[ d rjkktrwbb ckjd: idf(l) = log((|c| + 0.5)/df(l))/log(|c| + 1) ult |c| - rjkbxtcndj ljrevtynjd d rjkktrwbb, df(l) - rjkbxtcndj ljrevtynjd, ult dcnhtnbkfcm ktvvf l. Â bnjujdjt pyfxtybt tf*idf d[jlzn, gjvbvj j,sxyjq dcnhtxftvjcnb ckjdf d ljrevtynt, yfl,fdrb pf ghbcencndbt ckjd d dsltktyys[ j,kfcnz[ (title, pfujkjdrb nbgf h1-h4 b n.g.). Â ghjikjv ujle lkz eghjotybz hf,jns vs exbnsdfkb njkmrj dcnhtxftvjcnm ckjd d title. Â 'njv ujle ddtkb jcnfkmyst gfhfvtnhs, jndtxf.obt pf ajhvfnbhjdfybt ckjd d ljrevtynt (h1-h4 b ghjx.). F1(DocWeight) – aeyrwbz jn dtcf ljrevtynf, dsxbcktyyjuj gj c[tvt, ghtlkj;tyyjq d [1]. Îcj,tyyjcnb aeyrwbb: f) F1, d njv xbckt, pfybvftncz ghbdtltybtv pyfxtybz DocWeight lj ye;yjuj lbfgfpjyf, afrnbxtcrb, yjhvbhjdrjq. Ätqcndbt aeyrwbb yf dtc ljrevtynf cbkmyj pfdbcbn jn cgjcj,f yjhvbhjdrb, xnj d bnjut ceotcndtyyj dkbztn yf gjhzljr ljrevtynjd d dslfxt; ,) ×fcnm ccskjr ghbpyf.ncz ytbyajhvfnbdysvb b d hfcxtnt yt exfcnde.n. Ôfrnbxtcrb W1 jndtxftn pf byajhvfwbjyye. pyfxbvjcnm ljrevtynf b tuj dtc gj jnyjityb. r lheubv ljrevtynfv, dsxbcktyysq gj c[tvt ccskjxyjuj hfy;bhjdfybz, jgbcfyyjq d [1]. W2 = Σ (TF*IDF(Link)* F2(LinkWeight)), ult: TF*IDF(Link) - TF*IDF ccskrb yf lfyysq ljrevtyn; F2(LinkWeight)) – aeyrwbz ghbdtltybz dtcjd ccskjr yf ljrevtyn. LinkWeight dsxbckztncz fyfkjubxyj DocWeight Ò.j. W2 jndtxftn pf byajhvfwbjyye. pyfxbvjcnm ccskjr yf lfyysq ljrevtyn b b[ dtcf.


W3 = F3(hfccnjzybt) – aeyrwbz, jndtxf.ofz pf extn hfccnjzybz vt;le ckjdfvb pfghjcf d ljrevtynt. Èvttn ytyektdjt pyfxtybt ghb ghj[j;ltybb rdjhevf. Äfktt j, 'njv gjlhj,yj. Â ghjikjv ujle d F3(hfccnjzybt) d[jlbk gfccf; bp pfghjcf, dcnhtxtyysq d ljrevtynt. Ñdjqcndf gfccf;f: 1) Ïfccf; vj;tn ,snm ytgjkysv. Îy dsxbckztncz gj rdjheve cevvs IDF d[jlzob[ d ytuj ckjd b gj rdjheve xbckf ckjd. Î,f gfhfvtnhf pflf.ncz d yfcnhjqrf[. Ífghbvth: gfccf;tv cxbnftncz njn, d rjnjhsq djikj 70% ckjd pfghjcf, b b[ dtc yt vtytt 60% jn cevvs IDF dct[ ckjd pfghjcf. Òjxyst pyfxtybz gjl,bhf.ncz c gjvjom. jgnbvbpfwbb gfhfvtnhjd.

2) Ïjhzljr ckjd d gfccf;t yt bvttn pyfxtybz. Ìs yfvthtyyj ddtkb 'ne jcj,tyyjcnm gfccf;f, ghbybvfz d hfcxtn cdjqcndf heccrjuj zpsrf, gj rjnjhjve cvsck xfcnj yt pfdbcbn jn gjhzlrf ckjd ([jnz b yt dctulf). Ðfcxtn bltn yf nj, xnj,s yt bcrk.xfnm lj,fdre lkz ljrevtynjd, ult ckjdf bp pfghjcf cktle.n d lheujv gjhzlrt. Ïk.c, xnj ceotcndtyyj lkz gjbcrf gj dt,e, ,jkmifz lj,fdrf pf ;tcnrbq gjhzljr ckjd lftn ,jkmijt gjkt ltzntkmyjcnb lkz cgfvf gjbcrjds[ vfiby.

f) Ïfccf; dsxbckztncz njkmrj lkz ckjd, d[jlzob[ d jlyj ghtlkj;tybt.

,) Ðfccnjzybt vt;le ckjdfvb bp pfghjcf d ljrevtynt yt ljk;yj ghtdsifnm vfrcbvfkmyjuj jryf. Ífghbvth, 10 bkb 15.

d) Â 'njv ujle vs ddtkb ljgjkybntkmyjt cdjqcndj gfccf;f: gfccf;tv cxbnftncz njn, lkz rjnjhjuj _dct_ ckjdf bp pfghjcf dcnhtnbkbcm d lfyyjv ghtlkj;tybb +/-N ghtlkj;tybq. Çf 'nj ljrevtyn gjkexftn ljgjkybntkmye. yfl,fdre. Â ghjujyf[ ÐÎÌÈÏ vs bcgjkmpjdfkb pyfxtybt N = 1. Ìj;yj bcgjkmpjdfnm ,Îkmitt pyfxtybt (yfghbvth, 2), yj d yfib[ 'rcgthbvtynf[ jyj lfkj [elibq htpekmnfn.

Äjgjkybntkmyj d 'njv ujle ,skb ddtlty cktle.obq gfhfvtnh, pfdbczobt jn hfccnjzybz:

4) Ïfhs ckjd. Âtc gj gfhfv ckjd dsxbckzkcz rfr jgbcfyj d [3]. Ïhb 'njv, gjybvfz, xnj d hzlt ckexftd njxysq gfccf; kexit jgbcsdftn cvsck ljrevtynf, vs gfhfvb ckjd exbnsdftv d njv xbckt b gjhzljr ckjd d


gfccf;t.  yfib[ 'rcgthbvtynf[ gfhfvtnh lfk ghb,fdre d rfxtcndt 5%. W4 - uheggf ljgjkybntkmys[ gfhfvtnhjd (ddtltyf d 'njv ujle): 1) Ákbpjcnm ckjd bp pfghjcf r yfxfke ghtlkj;tybz. Äj cb[ gjh vs yt dcnhtxfkb jgbcfybt 'njuj afrnjhf d hf,jnf[ gj byajhvfwbjyyjve gjbcre. Îcyjdfybt  heccrjv zpsrt ntvf (bpdtcnyfz byajhvfwbz) ghtlkj;tybz dshf;ftncz d njv xbckt b gjhzlrjv ckjd. Õjnz d heccrjv zpsrt ckjdf vjuen cnjznm d k.,jv gjhzlrt, ghzvjq gjhzljr ckjd zdkztncz ,jktt egjnht,bntkmysv. Òfrbv j,hfpjv, nj, j xtv bltn htxm (ntvf), - j,jpyfxtyj d yfxfkt ahfps. Ïhb j,hf,jnrt pfghjcf vs lfdfkb lj,fdre ljrevtyne, ult rk.xtdst ckjdf dcnhtxf.ncz d yfxfkt ghtlkj;tybz.  yfib[ 'rcgthbvtynf[ gfhfvtnh lfk ghb,fdre d rfxtcndt 5%; 2) Âcnhtxftvjcnm d ljrevtynt njxys[ ckjdjajhv bp pfghjcf. Ífb,jktt ceotcndtyysvb d gkfyt ekexitybz rfxtcndf jrfpfkbcm afrnjhs: 1. Ïfhs ckjd 2. Ákbpjcnm ckjd bp pfghjcf r yfxfke ghtlkj;tybz. Ïjcrjkmre gjzdktybt yjds[ afrnjhjd eckj;yztn ecnhjqcndj fkujhbnvf gjbcrf, vs pflfkbcm wtkm. cjplfnm ghjuhfvve fdnjvfnbxtcrjuj gjl,jhf gfhfvtnhjd b jghtltktybz df;yjcnb njuj bkb byjuj afrnjhf. 3.2 Îgnbvbpfwbz gfhfvtnhjd Ìs gjcnfdbkb gthtl cj,jq pflfxe cjplfnm ghjuhfvve, jgnbvbpbhe.oe. rj'aabwbtyns gthtl ckfuftvsvb d ajhvekt htktdfynyjcnb. Ïhjuhfvvf jcyjdsdftncz yf 'rcgthnys[ jwtyrf[ ljrevtynjd.

Áfpjq lkz yfcnhjqrb cbcntvs gjcke;bkb vfnthbfks ghjikjujlytuj ctvbyfhf ÐÎÌÈÏ: pfghjcs lkz ljhj;rb gjbcrf gj dt,-rjkktrwbb b ljrevtyns, dslfyyst cbcntvfvb-exfcnybwfvb yf lfyyst pfghjcs. Îhbtynbhfvb, ujdjhzobvb, rfrjq ljrevtyn «[jhjibq», f rfrjq «gkj[jq», cnfkb jwtyrb 'rcgthnjd ÐÎÌÈÏ.


Çflfxf ghjuhfvvs-jgnbvbpfnjhf – gjlj,hfnm gfhfvtnhs, rjnjhst j,tcgtxbdf.n yfbkexibt htpekmnfns jwtyrb dslfxb. a) ghbywbgs, kt;fobt d jcyjdt ghjuhfvvs-jgnbvbpfnjhf; 1) Îgnbvbpfwbz ltkftncz yf jcyjdt jwtyjr frctccjhjd; 2) Äkz jgnbvbpfwbb gfhfvtnhjd ljk;yf ceotcndjdfnm rjkbxtcn-dtyyfz vthf jwtyrb ljrevtynf – nj, xnj ujdjhbn, xtv jlby ljrevtyn kexit lheujuj. Ìs ghbvtyzkb vjlbabwbhjdfyyjt pyfxtybt gjrfpf-ntkz bpref-10. 3) Îgnbvbpfwbz ghj[jlbkf vtnjljv vjlbabwbhjdfyyjuj rjjh-lbyfnyjuj cgecrf. ,) ghj,ktvs, djpybribt d ghjwtcct cjplfybz; Îgnbvbpbhetvfz aeyrwbz j,kflftn cktle.obvb cdjqcndfvb: 1) Ôeyrwbz recjxyfz; 2) Ôeyrwbz ytvjyjnjyyfz; 3) Ôeyrwbz j,kflftn pfdtljvj ,Îkmibv xbckjv gfhfvtnhjd, xtv bcgjkmpetncz lkz tt dsxbcktybz. Èc[jlz bp 'nb[ cdjqcnd, yf[j;ltybt ukj,fkmyjuj vfrcbvevf zdkztncz pfdtljvj ytljcnb;bvsv. Â nfrjq cbnefwbb vj;yj ujdjhbnm njkmrj j ljcnfnjxyj [jhjitv kjrfkmyjv vfrcbvevt. Â 'nb[ eckjdbz[ vs ghbvtybkb ldf cgjcj,f jgnbvbpfwbb gfhfvtnhjd: 1) Ïfhfvtnhs jgnbvbpbhe.ncz dct chfpe; 2) Äkz yfb,jktt pyfxbvs[ c yfitq njxrb phtybz gfhfvtnhjd pflftncz ytcrjkmrj yfxfkmys[ pyfxtybq rj'aabwbtynf. Äkz rf;ljuj pyfxtybz rj'aabwbtynf ghjdjlbncz tuj jgnbvbpfwbz (jgnbvbpbhetncz njkmrj 'njn gfhfvtnh, jcnfkmyst ytbpvtyys). Èp ytcrjkmrb[ «jgnbvbpbhjdfyys[» pyfxtybq gfhfvtnhf ds,bhftncz njn, d rjnjhjv pyfxtybt aeyrwbb htktdfynyjcnb vfrcbvfkmyj. Ïjckt xtuj d aeyrwb. htktdfynyjcnb lj,fdkz.ncz lheubt gfhfvtnhs b jgnbvbpbhe.ncz nfrbv ;t cgjcj,jv. Î,f cgjcj,f: b 1, b 2 - gjcnhjtys yf rjjhlbyfnyjv cgecrt, hfpkbxbt d njv, xnj d 1 cgecr ltkftncz gj dctv gfhfvtnhfv «gj rheue», f d 2 rf;lsq gfhfvtnh jgnbvbpbhetncz lj vfrcbvfkmyjuj djpvj;yjuj pyfxtybz aeyrwbb, gjnjv yfqltyyjt jgnbvfkmyjt pyfxtybt gfhfvtnhf pfgbcsdftncz rfr rjycnfynf b lfktt ghjdjlbncz jgnbvbpfwbz cktle.otuj gfhfvtnhf.


d) htpekmnfns jgnbvbpfwbb gfhfvtnhjd  htpekmnfnt hf,jns cgjcj, 2 jrfpfkcz ,jktt 'aatrnbdysv ('aatrnbdyjcnm d lfyyjv ckexft jpyfxftn ,Îkmitt pyfxtybt aeyrwbb htktdfynyjcnb).  cdzpb c ceotcndtyyj hfpysvb cdjqcndfvb rjkktrwbq KM.RU, BY.WEB b Legal vs ghjdjlbkb jgnbvbpfwb. rj'aabwbtynjd lkz rf;ljq rjkktrwbb jnltkmyj. 4. Äjgjkybntkmyst djpvj;yjcnb gjbcrjdjuj vt[fybpvf Äkz ekexitybz rfxtcndf hf,jns vs ddtkb cktle.obt cdjqcndf:

1) Ïhbvtytybt ckjdfhz cjrhfotybq. Ïhbvth: gj pfghjce «ÐÔ» botncz «ÐÔ» b «Ðjccbqcrfz Ôtlthfwbz».

2) Ïhbvtytybt cgbcrf cnjg-ckjd.

5. Ìths gj cnhernehbhjdfyb. rjkktrwbb b pfghjcjd Èpdtcnyst cnfnbcnbxtcrbt vtnjls byajhvfwbjyyjuj gjbcrf kexit hf,jnf.n yf [jhjij cnhernehbhjdfyys[ lfyys[. Çflfxf cnhernehbpfwbb ghtlcnfdkztncz yfv cktle.otq. Äfyj:

1. Ítcnhernehbhjdfyyfz rjkktrwbz.

2. Ítcnhernehbhjdfyyst pfghjcs.

Ötkb:

1. Ñnhernehbhjdfnm rjkktrwb..

2. Ñnhernehbhjdfnm pfghjcs.

 htpekmnfnt nfrb[ ghtj,hfpjdfybq bp cdzpb |«Ítcnhernehbhjdfyyst pfghjcs» → «Ítcnhernehbhbhjdfyyfz rjkktrwbz»| gjkexftv cdzpm |«Ñnhernehbhjdfyyst pfghjcs» → «Ñnhernehbhbhjdfyyfz rjkktrwbz»|, hf,jnfnm c rjnjhjq ujhfplj ktuxt. Ñnhernehbhjdfybt rjkktrwbb

 eckjdbz[ gkj[j cnhernehbhjdfyyjq rjkktrwbb vs ghtlghbyzkb vths gj elfktyb. byajhvfwbjyyjuj ievf bp ljrevtynjd. Ïjl byajhvfwbjyysv ievjv vs gjybvftv nj, xnj yt jnyjcbncz r jcyjdyjve cjlth;fyb. ljrevtynjd: htrkfvye. byajhvfwb., ccskrb yf vfnthbfks yf lheub[ cfqnf[ b n.g. Äkz elfktybz vs ghbvtybkb vtnjl, gjpdjkz.obq yf[jlbnm jajhvktybt cnhfybw cfqnf [4]. Ìtnjl yt elfkztn dtcm vecjh, yj ljdjkmyj [jhjij cghfdkztncz c pflfxtq. Ñnhernehbhjdfybt pfghjcjd Ïjl cnhernehbhjdfybtv pfghjcjd vs gjybvftv cktle.obt ifub:

1) Ècghfdktybt jgtxfnjr;

2) Âsxktytybt cvsckf bp pfghjcjd;

3) Ðfcibhtybt pfghjcjd.

Ï.1 htfkbpetncz c gjvjom. cthdbcf bcghfdktybz jgtxfnjr. Ï.2 elftncz htfkbpjdfnm njkmrj ghb jghtltktyys[ eckjdbz[. Ífghbvth, bp pfghjcf «ghbynths vjcrdf» vj;yj gjkexbnm lfyyst lkz utjnfhutnbyuf, lkz pfghjcf «ntktajy rfat yf geirbycrjq» ghjdtcnb gjbcr gj ,fpt yjvthjd ntktajyjd b dslfnm cjjndtncnde.oe. byajhvfwb.. Íj xfcnj nfre. jgthfwb. ghjbpdtcnb yt ghtlcnfdkztncz djpvj;ysv bp-pf ytjghtltktyyjcnb pfghjcf. Ï.3 - lkz ghjujyjd ÐÎÌÈÏ vs htfkbpjdfkb ckjdfhm cjrhfotybq.

6. Óxfcnbt d ctvbyfht b gjkextyyst htpekmnfns.

6.1 Äjhj;rb, d rjnjhs[ vs ghbyzkb exfcnbt f) ljhj;rb gjbcrf  'njv ujle, rfr b d ghjikjv, vs exfcndjdfkb d ljhj;rt gjbcrf gj dt,-rjkktrwbb. Ïjbcr jceotcndkzkcz jnltkmyj gj yf,jhe ljrevtynjd ÊÌ.RU b jnltkmyj gj ljrevtynfv ,tkjheccrjuj Èynthytnf (BY.WEB). Ìs gjghj,jdfkb ct,z d yjds[ lkz yfc ljhj;rf[: gjbcrf gj yjhvfnbdyj-ghfdjdjq rjkktrwbb (Legal) b ljhj;rt gjbcrf gj cvtifyyjq rjkktrwbb, ult cj,hfys dvtcnt chfpe nhb yf,jhf ljrevtynjd: rjkktrwbz ÊÌ.RU, rjkktrwbz ,tkjheccrjuj Èynthytnf b yjhvfnbdyj-ghfdjdfz rjkktrwbz. ,) ghjujys Äkz ljhj;rb gjbcrf gj dt,-rjkktrwbb (b lkz rjkktrwbb ÊÌ.RU, b lkz rjkktrwbb BY.WEB) vs ltkfkb ldf ghjujyf.  gthdjv ghjujyt bcgjkmpjdfkcz fkujhbnv, rjnjhsq ctqxfc hf,jnftn yf gjbcrt gj gjhnfke ÊÌ.RU. Ânjhjq ghjujy jceotcndkzkcz c gjvjom. fkujhbnvf c yjdsvb gjbcrjdsvb afrnjhfvb b jgnbvbpbhjdfyysvb rj'aabwbtynfvb.


Äkz ljhj;rb gjbcrf gj yjhvfnbdyj-ghfdjdjq rjkktrwbb vs ltkfkb jlby 'rcgthbvtynfkmysq ghjujy. Ífv ,skj bynthtcyj, rfr gjbcrjdsq vt[fybpv, jgnbvbpbhjdfyysq lkz dt,f, cghfdbncz cj cgtwbabxtcrjq rjkktrwbtq. Äkz ljhj;rb gjbcrf gj cvtifyyjq rjkktrwbb vs ltkfkb jlby ghjujy. Ìs dszcyzkb, ljrevtyn bp rfrjq rjkktrwbb ,eltn yf 1-v vtcnt d dslfxt gj cjjndtncnde.otve pfghjce. Äheubt gjrfpfntkb vs d hfcxtn yt ghbybvfkb, n.r. pflfxf ,skf bvtyyj nfrjq. 6.2 Ïjkextyyst htpekmnfns, b[ fyfkbp Äkz ghbvtytybz yjds[ afrnjhjd vs cjplfkb yjdsq vt[fybpv gjbcrf b hfy;bhjdfybz ljrevtynjd. Ê cj;fktyb., xfcnm jgsnjd r chjrfv clfxb htpekmnfnjd ghjdtcnb yt ecgtkb. Òtv yt vtytt, ghjljk;bkb bccktljdfybz, htpekmnfns rjnjhs[, yfhzle c jabwbfkmysvb lfyysvb, ghtlcnfdkztv yb;t. Êjkktrwbz BY.WEB Îcyjdyjq egjh d 'njv ujle vs ltkfkb yf rfxtcndj gjbcrf gj rjkktrwbb BY.WEB Ó yfc gjkexbkbcm cktle.obt htpekmnfns: Ãhfabr TREC, jwtyrf OR lkz rjkktrwbb BY.WEB «KM» - htpekmnfns hf,jns fkujhbnvf, hf,jnf.otuj yf gjhnfkt KM.RU, «KM jgnbv» - htpekmnfns yjdjuj fkujhbnvf. 00,10,20,30,40,50,600,10,20,30,40,50,60,70,80,91xxxx-1xxxx-2xxxx-3xxxx-4xxxx-5xxxx-6KMxxxx-8xxxx-9xxxx-10KM - jgnbvxxxx-12xxxx-13xxxx-14xxxx-15


Ïhbdtltv htpekmnfns gj jcyjdysv gjrfpfntkzv:

Óxfcnybr

Prec(5)

Prec(10)

Bpref-10

Bpref

Recall

[[[[-1

0,3

0,26

0,24

0,19

0,37

[[[[-2

0,33

0,3

0,28

0,22

0,43

[[[[-3

0,24

0,2

0,17

0,14

0,24

[[[[-4

0,33

0,31

0,33

0,26

0,55

[[[[-5

0,31

0,26

0,24

0,2

0,38

[[[[-6

0,17

0,15

0,13

0,11

0,2

ÊÌ

0,31

0,28

0,26

0,22

0,38

[[[[-8

0,25

0,21

0,17

0,15

0,25

[[[[-9

0,29

0,27

0,22

0,19

0,31

[[[[-10

0,34

0,32

0,33

0,26

0,55

ÊÌ-jgnbv

0,34

0,3

0,27

0,23

0,4

[[[[-12

0,31

0,26

0,24

0,2

0,39

[[[[-13

0,29

0,26

0,24

0,19

0,38

[[[[-14

0,25

0,21

0,17

0,15

0,25

[[[[-15

0,3

0,27

0,24

0,19

0,39

Ïj gjrfpfntk. Precision(5) e yfc htpekmnfn yf ehjdyt kblthf (ghjujy xxxx-10), xnj cjjndtncndetn yfibv wtkzv. Ïj Precision(10), Bpref b Bpref-10 - ljcnfnjxyj [jhjibt pyfxtybz. Ïj gjrfpfntk. Recall yf[jlbvcz d jcyjdyjq uheggt. Încnfdfybt d Recall jn kblthjd ghtlgjkj;bntkmyj jcyjdsdftncz yf ytbcgjkmpjdfybb yfvb d ghjujyf[ gjbcrf gj rdjheve. Ïjckt clfxb htpekmnfnjd d ÐÎÌÈÏ vs ghjljk;bkb bccktljdfybz. Ïhtlghbybvfkbcm cktle.obt ifub: 1. Ó,hfkb bcgjkmpjdfybt cnjg-ckjd. Èynthtcyj, xnj, ghjdthbd 'njn ht;bv yf hfpys[ pfghjcf[ b rjkktrwbz[ (BY, KM.RU c pfghjcfvb 2007 b 2008 ujljd) dtplt vs gjkexbkb htpekmnfn kexibq, xtv ghb bcrk.xtybb cnjg-ckjd bp gjbcrf. 2. Ârk.xbkb ht;bv bcghfdktybz jgtxfnjr b bcgjkmpjdfybz rdjhevf. Ïhb 'njv rdjhev gj evjkxfyb. yt hf,jnfk, f gjlrk.xfkcz njkmrj ghb cktle.ob[ eckjdbz[: f) jncencndbt ljrevtynjd d dslfxt; bkb ,) jncencndbt gfccf;f d ljrevtynf[ dslfxb.


Ðtpekmnfns yfitq hf,jns ghtlcnfdktys yf uhfabrt yb;t. Ãhfabr TREC gj BY.WEB gjckt jnhf,jnrb vjlektq bcghfdktybz jgtxfnjr b gjbcrf gj rdjheve (ltkfkcz dyt ÐÎÌÈÏ gjckt clfxb jcyjdys[ htpekmnfnjd) Ïjckt ghbvtytybz 'rcgthbvtynfkmys[ vjlektq vs ekexiftv uhfabr, rjnjhsq gthdst itcnm njxtr bltn yfhfdyt c kblthjv, gjnjv cnfyjdbncz ytcrjkmrj [e;t, yj kexit uhfabrf «ÊÌ – jgnbv». Íjdst vjlekb ye;lf.ncz d jgnbvbpfwbb, b j,rfnrf b[ yf lfyys[ ÐÎÌÈÏ, gecnm b d ytjabwbfkmyjv ht;bvt, gjpdjkbkf yfv edbltnm b[ gthcgtrnbdyjcnm. Èpexfz gjdtltybt uhfabrf TREC b gjrfpfntktq Precision, vs cltkfkb ytcrjkmrj dsdjljd jnyjcbntkmyj hf,jns yfituj fkujhbnvf yf pfghjcf[ 2007 b 2008 ujlf r rjkktrwbb BY.WEB: 00,10,20,30,40,50,60,71234567891011ÊÌ - rdjhevÊÌ - jgnbvxxxx-10

1) Ïjrfpfntkb ghfrnbxtcrb yt pfdbczn jn ghbvtytybz bkb ytn ccskjxyjuj hfy;bhjdfybz (rfr dtcjd ljrevtynjd, nfr b ntrcnjd ccskjr).

2) Ïhbvtytybt d ccskrf[ gfh ckjd b gfccf;tq yt lftn ekexitybz rfxtcndf.

3) Ïhbvtytybt gfccf;f d pfujkjdrt (Title) yt lftn ekexitybz rfxtcndf.


Êjkktrwbz KM.RU Ó yfc gjkexbkbcm cktle.obt htpekmnfns: Ãhfabr TREC, jwtyrf OR lkz rjkktrwbb KM.RU «KM» - htpekmnfns hf,jns fkujhbnvf, hf,jnf.otuj yf KM.RU, «KM jgnbv» - htpekmnfns yjdjuj fkujhbnvf. Âblyj, xnj uhfabrb exfcnybrjd hfpltkbkbcm yf ldt uheggs. Ó gthdjq uheggs uhfabrjd htprj pflhfyj yfxfkj, cjjndtncndtyyj, gthdst njxrb e yb[ dsit. Ânjhfz uheggf uhfabrjd – ujhbpjynfkmyfz. Ífxfkmyst njxrb yf b[ uhfabrf[ yb;t. Ãheggs gthtctrf.ncz gjckt dnjhjq njxrb, b lfktt uhfabrb gthdjq uheggs cnhtvbntkmyj gflf.n, f dnjhjq gkfdyj cyb;f.ncz. Äevftv, xnj ltktybt yf uheggs cdzpfyj c jcj,tyyjcnzvb fkujhbnvjd exfcnybrjd. Ïhbdtltv htpekmnfns gj jcyjdysv gjrfpfntkzv: 00,10,20,30,40,50,60,70,81234567891011[[[[-1[[[[-2[[[[-3ÊÌ-jgnbv[[[[-5[[[[-6[[[[-7[[[[-8[[[[-9ÊÌ[[[[-11[[[[-12[[[[-13[[[[-14[[[[-15

¹

Pr(5)

Pr(10)

Bpref-10

Bpref

Recall

[[[[-1

0,54

0,43

0,29

0,27

0,42

[[[[-2

0,54

0,45

0,28

0,27

0,33

[[[[-3

0,40

0,33

0,20

0,19

0,28

ÊÌ-jgnbv

0,53

0,45

0,36

0,34

0,43


90

[[[[-5

0,55

0,46

0,31

0,29

0,45

[[[[-6

0,42

0,40

0,34

0,32

0,50

[[[[-7

0,50

0,43

0,27

0,26

0,31

[[[[-8

0,49

0,45

0,39

0,37

0,58

[[[[-9

0,52

0,48

0,37

0,34

0,46

ÊÌ

0,49

0,44

0,36

0,34

0,43

[[[[-11

0,60

0,46

0,29

0,27

0,40

[[[[-12

0,48

0,44

0,34

0,30

0,40

[[[[-13

0,56

0,44

0,31

0,29

0,39

[[[[-14

0,51

0,46

0,27

0,26

0,33

[[[[-15

0,50

0,45

0,30

0,28

0,41

Ïjckt clfxb htpekmnfnjd vs bcgsnfkb yf rjkktrwbb KM.RU gjbcr gj rdjheve c cj,k.ltybtv jgbcfyys[ dsit eckjdbq. Âblyj, xnj [fhfrnth uhfabrf yt bpvtybkcz, yj cfv uhfabr gkfdyj gjlyzkcz ddth[. Â wtkjv uhfabr TREC jrfpfkcz jxtym xedcndbntkty r cbnefwbzv, rjulf cbcntvf yt yf[jlbn ljrevtynjd. Ïhbvtybd rdjhev r itcnb pfghjcfv bp itcnbltcznb, pflfyys[ r rjkktrwbb KM.RU, vs gjkexbkb ceotcndtyyj kexibq uhfabr. Â cbnefwbz[, rjulf rdjhev yt nht,etncz, vjlbabrfwbz cbcntvs, ghtlcnfdktyyfz yf ÐÎÌÈÏ, jnhf,fnsdftn [jhjij. Ãhfabr TREC c vt[fybpvjv rdjhevf lkz rjkktrwbb KM.RU (ltkfkcz dyt ÐÎÌÈÏ gjckt clfxb jcyjdys[ htpekmnfnjd) 00,10,20,30,40,50,60,70,81234567891011ÊÌ - rdjhevÊÌ - jgnbvxxxx-5


Êjkktrwbz Legal Ó yfc gjkexbkbcm cktle.obt htpekmnfns: Ãhfabr TREC, jwtyrf OR lkz rjkktrwbb Legal Ïhbdtltv htpekmnfns gj jcyjdysv gjrfpfntkzv: 00,10,20,30,40,50,60,70,80,900,10,20,30,40,50,60,70,80,91xxxx-1xxxx-2xxxx-3ÊÌxxxx-5xxxx-6

¹

Pr(5)

Pr(10)

Bpref-10

Bpref

Recall

[[[[-1

0,4

0,36

0,27

0,29

0,44

[[[[-2

0,53

0,48

0,37

0,42

0,63

[[[[-3

0,5

0,47

0,33

0,37

0,55

ÊÌ

0,5

0,42

0,28

0,33

0,51

[[[[-5

0,52

0,5

0,36

0,41

0,62

[[[[-6

0,38

0,34

0,2

0,24

0,34

Ìs cltkfkb jlby 'rcgthbvtynfkmysq ghjujy. Ífv ,skj bynthtcyj, rfr gjbcrjdsq vt[fybpv, jgnbvbpbhjdfyysq lkz dt,f, cghfdbncz cj cgtwbabxtcrjq rjkktrwbtq. Ðtpekmnfn gjrfpfk, xnj yfit yfghfdktybt – kexibq htpekmnfn yf vfrcbvfkmyj dscjrjq gjpbwbb – dslth;bdftncz. 1-z njxrf yf uhfabrt TREC yf[jlbncz ljcnfnjxyj dscjrj. Ïjrfpfntkm precision(5) yf[jlbncz yf [jhjitv ehjdyt gj chfdytyb. c lheubvb exfcnybrfvb. Äfktt vs bltv dcktl pf jcyjdyjq uheggjq. Ïjcrjkmre gjbcr gj cgtwbabxysv rjkktrwbzv yt zdkztncz yfibv ghbjhbntnysv yfghfdktybtv, vs d wtkjv ljdjkmys htpekmnfnjv. Òtv yt vtytt, vs htibkb dszcybnm ghbxbye gfltybz uhfabrf b djpvj;yjcnm ekexitybz rfxtcndf gjbcrf gj ghjbpdjkmyjq rjkktrwbb (d lfyyjv ckexft, yjhvfnbdys[ ljrevtynjd). Ítcvjnhz yf nj, xnj d rjkktrwbb ghbcencndetn ,jkmijt xbckj ccskjr, htibkb dszcybnm, rfr gjbcr jnhf,jnftn ,tp ccskjxyjuj hfy;bhjdfybz. Òtcns gjrfpfkb, xnj d 'njv ckexft htpekmnfns gjkexf.ncz kexit. Âjpvj;yj, 'nj jpyfxftn, xnj r gjlj,yjq rjkktrwbb ytghbvtybv nhflbwbjyysq gjl[jl, bcgjkmpe.obq ccskjxysq uhfa. Äjgjkybntkmyj vs dszcybkb, xnj rfxtcndj hfy;bhjdfybz ceotcndtyyj pfdbcbn jn extnf gfccf;tq d pfujkjdrt ljrevtynf. Äfkmytqitt gjdsitybt rfxtcndf gjbcrf vj;tn ,snm cdzpfyj c bcgjkmpjdfybtv rdjhevf b ckjdfhz cgtwbabxtcrb[ lkz rjkktrwbb cjrhfotybq. Ãhfabr TREC ,tp ccskjxyjuj hfy;bhjdfybz b c gfccf;fvb d pfujkjdrf[ lkz rjkktrwbb Legal (ltkfkcz dyt ÐÎÌÈÏ gjckt clfxb jcyjdys[ htpekmnfnjd) Âsdjls Ýrcgthbvtyns yf ÐÎÌÈÏ gjrfpfkb, xnj jgnbvbpfwbz rj'aabwbtynjd d ajhvekt htktdfynyjcnb lftn ghbhjcn rfxtcndf gjbcrf. Íjdst gfhfvtnhs - gfhs ckjd b ,kbpjcnm ckjd r yfxfke ghtlkj;tybq – nj;t pfhtrjvtyljdfkb ct,z [jhjij. Ýnj jcj,tyyj pfvtnyj d gjbcrt gj dt,-rjkktrwbb. 00,10,20,30,40,50,60,70,81234567891011ÊÌ ,tp ccskjxyjujhfy;bhjdfybz b cgfccf;fvb dpfujkjdrf[ÊÌ cj ccskjxysvhfy;bhjdfybtv

VII. Âjpvj;yst genb lfkmytqituj hfpdbnbz Îcyjdyst vtnjls ekexitybz rfxtcndf dslfxb gjbcrjds[ vfiby d Èynthytnt lj cb[ gjh ,skb cdzpfys c fyfkbpjv ljrevtynjd d dt,-rjkktrwbb, f fyfkbpe pfghjcjd gjkmpjdfntktq yt eltkzkjcm ljk;yjuj dybvfybz. Îlyfrj ctqxfc dct ,jkmit bccktljdfntktq j,hfoftncz r 'njq ntvt (gjcrjkmre yfv bynthtcty byajhvfwbjyysq gjbcr d ctnb Èynthytn, vs ujdjhbv j, bccktljdfybb pfghjcjd b ntrcnjd ljrevtynjd d hfvrf[ dt,f). Ê fyfkbpe pfghjcf vj;yj jnytcnb fyfkbp tuj cbynfrcbxtcrjuj ecnhjqcndf, vjhajkjubxtcrb[ jcj,tyyjcntq ckjd d pfghjct, f nfr;t b[ ctvfynbxtcrjuj pyfxtybz. Ñbynfrcbxtcrbq fyfkbp pfghjcf jnxfcnb ghtlcnfdkztncz nhelyjghjbpdjlbvsv, gjcrjkmre ajhvekbhjdrf vyjub[ dt,-pfghjcjd yt cjjndtncndetn cnhjtyb. ahfps d tcntcndtyyjv zpsrt (ghbvth: Ìjcrdf ghbynths pfrfpfnm). Ñbynfrcbxtcrbq fyfkbp pfghjcf vj;tn ,snm bynthtcty lkz cjplfybz djghjcyj-jndtnyjuj gjbcrf (yf djghjc: Â rfrjv ujle hjlbkcz Ïeirby – botv jndtn: Ïeirby hjlbkcz d Õ ujle, ult Õ – cjcnfdyjt xbckbntkmyjt bkb xtnsht[pyfxyjt xbckj). Îlyfrj jhufybpfwbz nfrjuj dblf gjbcrf c cj,cndtyyjq ,fpjq ghjbyltrcbhjdfyys[ ljrevtynjd nht,etn yfvyjuj ,jkmit ecbkbq gj chfdytyb. c «nhflbwbjyyjq» gjbcrjdjq vfibyjq, gjcrjkmre djghjcyj-jndtnysq gjbcr ghtlgjkfuftn yfgbcfybt ,jkmijuj rjkbxtcndf ghfdbk, jnhf,jnrf rjnjhs[ ,eltn pfvtlkznm ghjwtcc j,hf,jnrb pfghjcf, b gjbcrjdjuj byltrcf bp hfpvtxtyys[ dt,-ljrevtynjd, rjnjhsq ,eltn pyfxbntkmyj ghtdsifnm gj j,]tve byltrc nhflbwbjyyjq gjbcrjdjq vfibys, xnj vj;tn gjpdjkbnm ct,t yt rf;lfz rjvfylf hfphf,jnxbrjd. Ïj'njve ctqxfc yf hsyrt ytn djghjcyj-jndtnyjq cbcntvs, cgjcj,yjq yfhfdyt rjyrehbhjdfnm c ghjvsiktyysvb cbcntvfvb byajhvfwbjyyjuj gjbcrf d Èynthytnt. Óxtn vjhajkjubxtcrb[ jcj,tyyjcntq ckjd d pfghjct gjnht,etn fyfkbpf vjhajkjubxtcrb[ jcj,tyyjcntq ckjd b d dt,-ljrevtynf[, xnj, jxtdblyj, pyfxbntkmyj edtkbxbn j,]tv gjbcrjdjuj byltrcf. Ïhjwtcc, crjhtt dctuj, ,eltn ntcyj cdzpfy c cbynfrcbxtcrbv fyfkbpjv ckjd d pfghjct (yfghbvth, bcgjkmpjdfybt byajhvfwbb j gflt;f[ lkz jghtltktybz ukfdyjuj ckjdf d ckjdjcjxtnfybz[ «,hfcktn bp pjkjnf», «rfhnf ujhjlf»).

Ífb,jktt ghbdktrfntkmyjq gjrf jcnftncz gthcgtrnbdf ghjdtltybz ctvfynbxtcrjuj fyfkbpf. Èyajhvfwb. j ctvfynbxtcrb[ pyfxtybz[ vj;yj bcgjkmpjdfnm lkz jghtltktybz ntvfnbrb pfghjcf, gjl,jhf cbyjybvjd lkz ckjd bp pfghjcf, cjplfybz ,fps cdzpfyys[ gjyznbq b lkz gthtajhvekbhjdrb pfghjcf lheubvb ckjdfvb. Ó;t ctqxfc gjbcrjdst cbcntvs evt.n hfcibahjdsdfnm cjrhfotybz (ÌÃÓ = Ìjcrjdcrbq ujcelfhcndtyysq eybdthcbntn) b cjgjcnfdkznm heccrbq b fyukbqcrbq dfhbfyn yfgbcfybz ckjd (ÁÌÂ = BMW). Äkz yfc ghtlcnfdkztncz bynthtcysv gjcnhjtybt ,fps cdzpfyys[ gjyznbq lkz ckjd heccrjuj zpsrf, b vs yfxfkb 'rcgthbvtynbhjdfnm d 'njv yfghfdktybb. Ýrcgthbvtyns gjrfpfkb, xnj gjcnhjtybt cdzptq vt;le ckjdfvb lkz dsltktyyjq ntvfnbxtcrjq j,kfcnb - 'nj dgjkyt htiftvfz pflfxtq, jlyfrj bcgjkmpjdfybt ds,hfyys[ yfvb vtnjljd dhzl kb gjpdjkbn cjplfnm ctnm cdzpfyys[ gjyznbq lkz dct[ ntvfnbxtcrb[ j,kfcntq chfpe. Ïj'njve yf ,kb;fqitt ,eleott vs cnfdbv gthtl cj,jq ldt pflfxb: 1) yfexbnmcz ecnfyfdkbdfnm cdzpb vt;le ckjdfvb lkz ghjbpdjkmyjq ds,jhrb ntrcnjd b pf ghbtvktvjt dhtvz; 2) yfexbnmcz rjhhtrnyj bcgjkmpjdfnm ,fpe cdzpfyys[ gjyznbq lkz gthtajhvekbhjdrb dt,-pfghjcf. Òjulf rfr ghb ghjcns[ gthtajhvekbhjdrf[ pfghjcf (hfcibahjdrf cjrhfotybq, pfvtyf heccrjuj yfgbcfybz yf fyukbqcrjt) vj;yj vfrcbvfkmyj juhflbnm ct,z jn jib,jr, nj ghb ,jktt ckj;ys[ dfhbfynf[ gthtajhvekbhjdrb ghb[jlbncz cnfkrbdfnmcz c hzljv nhelyjcntq. Äj cb[ gjh jcyjdysv 'nfgjv j,hf,jnrb pfghjcf Èynthytn-gjkmpjdfntkz zdkztncz gjbcr ckjd bp pfghjcf d dt,-ljrevtynf[, b ghb pfvtyt ckjd bp pfghjcf yf cbyjybvs, rjnjhst ghtlcnfdkz.ncz yfv jxtdblysvb, gjbcr ,eltn ghjbpdjlbnmcz e;t gj lheubv ckjdfv, xnj vj;tn e[elibnm rfxtcndj dslfxb. Ífghbvth, d yfcnjzott dhtvz cdzpm gjyznbq Ìtldtltd b ghtpbltyn Ðjccbb lkz yfc ,tpeckjdyf, b tckb yf pfghjc Ìtldtltd vs dslflbv cfqn ghtpbltynf Ðjccbb, nj 'njn ljrevtyn gjkmpjdfntkm, crjhtt dctuj, cjxntn [jhjibv rfr jlby bp dfhbfynjd jndtnf, jlyfrj vs yt vj;tv dslfxe lkz pfghjcf Ìtldtltd gjkyjcnm. gjlvtybnm dslfxtq lkz pfghjcf ghtpbltyn Ðjccbb, n.r. d 'njv ckexft gjkmpjdfntkm, djpvj;yj, yt gjkexbn ljrevtyns c ,bjuhfabtq Ìtldtltdf, c tuj bynthdm. b ghjx., b 'nj tot yt cfvjt gkj[jt, [e;t, tckb ghb nfrjq gjlvtyt dslfxb vs jnlflbv ljrevtyns ghj ghtpbltynf Ïenbyf bkb ghtpbltynf Åkmwbyf, njulf rfr gjkmpjdfntkm, vj;tn ,snm, [jntk epyfnm kbxyst afrns ,bjuhfabb ghtpbltynf Ìtldtltdf, b tuj cjdctv yt bynthtcjdfk gjkbnbxtcrbq fcgtrn.

Òtv yt vtytt, ytcvjnhz yf vyj;tcndj nhelyjcntq, cdzpfyys[ c ghjdtltybtv ctvfynbxtcrjuj fyfkbpf pfghjcf, 'njn genm hfpdbnbz ghtlcnfdkztncz ljcnfnjxyj gthcgtrnbdysv, b vs gkfybhetv ghjdjlbnm 'rcgthbvtyns d 'njv yfghfdktybb. Åckb nfrbt cgjcj,s jwtyrb htktdfynyjcnb ljrevtynf, rfr hfcxtn tf*idf, dtcf ljrevtynf, hfccnjzybz vt;le rk.xtdsvb ckjdfvb, vj;yj cxbnfnm rjkbxtcndtyysvb gjrfpfntkzvb, nj fyfkbp ctvfynbxtcrjuj pyfxtybz pfghjcf – 'nj jwtyrf e;t lheujuj gkfyf. Ítcjvytyyj, vs yt [jnbv jnrfpsdfnmcz jn extnf rjkbxtcndtyys[ gjrfpfntktq ghb j,hf,jnrt pfghjcf, yj yflttvcz, xnj bcgjkmpjdfybt yjdjq byajhvfwbb gjvj;tn yfv cltkfnm rfxtcndtyysq ifu dgthtl. Ëbnthfnehf

[1]. «The anatomy of a large-scale hypertextual Web search engine» S. Brin, L. Page. - http://infolab.stanford.edu/~backrub/google.html

[2]. «KM.RU yf ÐÎÌÈÏ-2007», Ñ.Òfntdjczy, Í.Áhspufkjdf - http://romip.ru/romip2007/romip2007_KM.RU.pdf

[3]. «Àkujhbnv ntrcnjdjuj hfy;bhjdfybz ßyltrcf yf ÐÎÌÈÏ-2006» Àylhtq Ãekby, Ìb[fbk Ìfckjd, Èkmz Ñtufkjdbx - http://download.yandex.ru/company/03_yandex.pdf

[4]. «Èpdktxtybt pyfxbvjq byajhvfwbb bp web-cnhfybw lkz pflfx byajhvfwbjyyjuj gjbcrf» Ì.Ñ. Àuttd, È.Â. Âthibyybrjd, Á.Â. Äj,hjd - http://www.cir.ru/docs/ips/publications/2005_yandex_obraml.pdf

[5]. «Îgnbvbpfwbz gfhfvtnhjd fkujhbnvf gjbcrf yf jcyjdt fyfkbpf jwtyjr 'rcgthnjd», Ì.Ñ. Àuttd, Á.Â. Äj,hjd - http://romip.ru/romip2005/07_uirussia.pdf

KM.RU at RIRES-2008. Parameter optimization S. Tatevosyan, N. Bryzgalova The paper describes a new modification of the information retrieval algorithm, introduced by KM.RU at RIRES-2007, we also talk about an optimizing system aimed at getting the most efficient coefficients for algorithm’s parameters for obtaining better results in information retrieval and document ranking. The article reports on results of KM.RU at RIRES-2008 and our future plans.

© Ñthutq Òfntdjczy, Ífnfkmz Áhspufkjdf «ÊÌ jykfqy» {tatevosyan, bryzgalova}@post.km.ru

3
Ñîçäàíèå ýêñêëþçèâíûõ ñàéòîâ, þçèáèëèòè àíàëèç è áåñïëàòíûé àíàëèç ïîä çàïðîñû îñíîâíûõ ïîèñêîâûõ ìàøèí
Êîíòàêòíàÿ èíôîðìàöèÿ :
òåë. +7(98I) 7608865

Íàïèñàòü ïèñüìî íà e-mail
icq 415547094  romverðåéòèíã íà mail.ru ñàéòà romverinbox.ru
© 1997 - 2024 romver.ru

Ïîëíàÿ êàðòà ñàéòà Display Pagerank