Chii chinonzi Semantic Scholar: AI-powered academic search engine

Kana huwandu hwezvinyorwa zvesainzi huchikura nguva dzose, kuwana chinyorwa chakakosha kunogona kuve odyssey. Apa ndipo panopinda Semantic Scholar, injini yemahara yekutsvaga inoshanda njere yekufunga kuwana nekunzwisisa tsvagiridzo nekukurumidza uye nerunyoro rushoma pane zvechinyakare injini.

Kupfuura rondedzero yakapfava yemhedzisiro, sevhisi iyi inosanganisira zviono zvemhando yepamusoro zvakaita sehuwandu hwemakotesheni, mamiriro ezvakatorwa, uye pfupiso-pfupi yemutsara mumwe. Nekuda kwekudzidza kwemichina, kugadzira mutauro wechisikigo, uye maitiro ekuona komputa, inokwanisa kutora hukama hune musoro pakati pemabasa, vanyori, uye misorokuita kuti kubhurawuza mabhuku kunyanyobudirira.

Chii chinonzi Semantic Scholar uye chinoshandiswa chii?

Semantic Scholar isayenzi yekutsvaga nekutsvaga, muenzaniso pakati pe mhando dzeinjini dzekutsvaga nyanzvi, inofambiswa neAllen Institute yeAI (AI2). Chinangwa chayo ndechekusimudzira kufambira mberi kwezivo nekubatsira vaongorori, vadzidzisi, uye vadzidzi kutsvaga nekunzwisisa basa rakakodzera. Zvemahara; unogona kuishandisa nekunyoresa neakaundi. Google kana masangano, uye muna 2020 zvakapfuura mamiriyoni manomwe vanoshandisa pamwedzi, chiratidzo chokufarira kwainoita munharaunda.

Iyi puratifomu inoita sebhiriji pakati pako uye ruzivo rwakakodzera zvechokwadi: inobvumidza iwe kusefa nekunyora, kuwana PDF, nzvimbo yeruzivo kana rudzi rwekuburitswa, uye inopa mazano ekuverenga ane hukama zvichienderana nezvaunofarira. Zvese izvi nechinangwa chekudzikisa ruzivo rwakawandisa uye koshesa mabasa ane simba guru pamusoro wega wegakwete chete iyo inonyanya kutaurwa nenzira isina kunaka.

Kuti uite izvi, inosimudzira musanganiswa wekudzidza muchina, NLP, uye kuona komputa. Nehunyanzvi uhu, inoburitsa pfupiso yemutsara mumwe ichishandisa nzira isinganzwisisike, uye zvakare inozivisa masangano (semuenzaniso, makomboni, zvipenyu, kana akakosha pfungwa) uye zvinoonekwa mukati mezvinyorwa. Mune mamwe mazwi, inowedzera semantic layer inobvumira kuti iite nzwisisa zvinorehwa nezviri mukati kwete mashoko chete.

Rekodhi yega yega mudhatabhesi yavo ine yakasarudzika identifier inonzi S2CID (Semantic Scholar Corpus ID). Ichi chiziviso chinogonesa kurevera, kuteedzera shanduro, uye kubatanidza kune mamwe dhatabhesi. Saka, kana iwe ukawana rimwe basa, une tag isina kujeka yekutaura kana kuitora, iyo inobatsira Dzivisa kusanzwisisika pakati pezvinyorwa zvine mazita akafanana.

Kuenzaniswa neGoogle Scholar kana PubMed, mutsauko wemaitiro wakajeka: pamusoro pekuverenga zvinotapurwa uye kuongorora temu zvinoitika pamwe chete, Semantic Scholar inoburitsa zvinonyanya kukosha zvenzvimbo yega yega uye inokwevera hukama pakati pezvinyorwa uchishandisa algorithms inofunga mamiriro. Nenzira iyi, inopa mhedzisiro iyo Vanoisa pamberi kukosha uye pesvedzero chaiyo. mukati mehurukuro yesainzi.

Iyo inoshanda sei: masaini, makotesheni, uye mhando zviratidzo

Paunenge uchitsvaga uye ukavhura rekodhi, uchaona kuti nhamba yezvinyorwa zvinowanzoratidzwa zvakajeka. Chinhu chinobatsira ndechekuti kutenderedza mbeva yako pamusoro peiyi nhamba kunoratidza mafambiro epagore ekuita mugirafu. Ichi chiito chinokurumidza chinoratidza nhoroondo yechinyorwa nekufamba kwenguva, zvichikubvumidza kuti... kuona peaks of interest kana nguva dzekugadzikana.

Kana iwe ukasimudza chitubu chako pamusoro pemabhawa mugirafu, kukosha kwegore rega rega kunoonekwa. Izvi zvinobatsira kupindura mibvunzo yakadai seyokuti: Zvichiri kutaurwa here? Yakava nepesvedzero huru pakutanga ichibva yadzikira, kana kuti inoramba ichifarira? Icho chokwadi chekuti basa rinoramba richitaurwa nhasi chiratidzo chakanaka chekukosha kwaro uye chinogona kupokana murondedzero yekuongorora se. uchapupu hwokuti mipiro yavo inoramba ichibatsira.

Kudzvanya pachinyorwa chinokupa iwe kuwana kune rumwe ruzivo rwakadzama: pfupiso, zviripo zvinongedzo (semuenzaniso, kushanduro dzePDF kana muparidzi), zvinyorwa zvakadudzwa, uye zvinyorwa zvinoenderana. Iri pani rinoshanda sehwaro hwekuwedzera kuverenga uye, nekungodzvanya kaviri, kuvaka cheni yakasimba yemareferensi, zvese mukati meiyo interface yakagadzirirwa nyore kushandisa. kuderedza el tiempo tsvaga uye wedzera kukosha.

Mukona yepamusoro yekurudyi, bhuroka rine data rakapfuma rinowanzoonekwa. Pakati peaya, mashoko ane simba guru anobuda pachena—kureva kuti, kutaura nezvemabasa umo chinyorwa chakava netapuro huru. Uyezve, inoratidza panowanikwa chinyorwa mukati mezvinyorwa zvinodudzwa (semuenzaniso, zvikamu zvakaita seKumashure kana Maitiro), chinhu chakakosha chekunzwisisa kana chinyorwa chiri kushandiswa senzvimbo. theoretical framework, methodology kana mhedzisiro yakakosha.

Aya masaini emhando yepamusoro anopedzisa huwandu hwese hwemataundi nechirevo. Kuziva kuti chidzidzo chinodzokororwazve muchikamu chemitoo hakuna kufanana nekunongedzerwa muchikamu chekumashure chete. Naizvozvo, kana tichitsanangura kunaka kwemupiro, zvinokurudzirwa kutaura zvese huwandu uye mamiriro ezvinyorwa izvi, tichibatanidza iyi data kuita rondedzero yakajeka yekukanganisa uye kukosha.

Kukosheswa kwemhedzisiro kunoenderana nemamodheru anonzwisisa zvirimo pane semantic level. Ivo havangoverenge mazwi chete, asi zvakare vanoongorora hukama pakati pepfungwa, kuona masangano, uye kuziva manhamba. Nenzira iyi, kubatana kunobuda pakati pemitsara yekutsvagisa, vanyori, uye majenari, zvichibvumira kuwanikwa kwedzimwe nzira dzekuverenga uye. bhiriji zvinyorwa pakati pe subfields.

Kuvhara kweCorpus uye shanduko yeprojekiti

Semantic Scholar yakavambwa muna 2015 kubva kuAllen Institute yeAI, iine tarisiro yekutanga pasainzi yekombuta. Kubva ipapo, kufukidzwa kwayo kwakaramba kuchikura uye kusiyanisa, kuve nzira yekuenda kune avo vanotsvaga nzvimbo dzinokurumidza uye dzine hungwaru dzemabhuku akakosha, nekuyedza kuenderera mberi. kuwedzera minda uye kuvandudza ruzivo rwemushandisi.

Muna 2017, timu yakazivisa kuwedzera kukuru mubiomedicine, ichiwedzera mabasa angangoita 26 miriyoni ezvekurapa kumamiriyoni gumi nemaviri aakatovhara kubva kune dzimwe nzvimbo. Iyi vhezheni yakakwenenzverwa yaisanganisira yakakwenenzverwa chimiro, dingindira rematemu, uye kucherechedzwa kwemisoro ine hukama kana iri kuitika. Mutungamiriri wepurojekiti panguva iyoyo, Marie Hagman, akasimbisa kuti chinangwa chaive chekufambisa kufamba nemusoro uye nemusoro. kuwana miganhu inomuka mukutsvaga.

Pakazosvika Ndira 2018, iyo corpus yakapfuura mamirioni makumi mana zvinyorwa pakombiyuta sainzi uye biomedicine. Nguva pfupi yapfuura, munaKurume wegore rimwe chetero, Doug Raymond-ane basa rematanho ekudzidza muchina pachikuva cheAlexa-akabatana kutungamira chirongwa. Uku kukwidziridzwa kwesangano kwakasimbisa kutarisa kwekushandiswa kweAI kuitira kunatsiridza kukosha uye scalability yehurongwa.

Kukura kwakawedzera mu2019 nekuwedzera kwemarekodhi kubva kuMicrosoft Academic. Muna Nyamavhuvhu wegore iro, nhamba yezvinyorwa yakapfuura 173 miriyoni, kuwanda kwakawanda kwakasimbisa chinzvimbo cheSemantic Scholar seimwe yeanotungamira zviwanikwa zvepamhepo zvemunda. akakura dhatabhesi ane zviri nani semantic chiratidzo inowanikwa kunharaunda yesainzi.

Mukufanana, chikuva chaifanira kufambisa dambudziko rekukura kwemabhuku: zvinyorwa zvinopfuura mamiriyoni matatu zvinobudiswa gore negore mumakumi ezviuru zvemagazini. Iri vhoriyamu rinoita kuti kuchengetedza kuome, ndosaka mishoni yekuisa pamberi uye yekubatanidza zvidimbu zvakakosha zvakakosha, nekuti. inochengetedza nguva uye inoderedza ruzha muongororo yezvinyorwa.

Maturusi ekutsvaga anobatsira nemasefa

Kukwenenzvera mhedzisiro, mafirita akakosha. Iwe unogona kudzikisira nekubatana-kunyora, kuwanikwa kwePDF, chirango, rudzi rwekushambadzira, kana zuva, pakati pemamwe maitiro. Kuzvishandisa pamwe chete zvinokutendera kuti uvake mibvunzo chaiyo, semuenzaniso: vhura zvinyorwa zvekupinda, mukati megore rakati renji, uye zvakanyoreswa nechimwe chikwata. Uku kusanganiswa kwemasefa, kana kwashandiswa nemazvo, ndeye Iine simba lever kuti uwane izvo zvaunoda chaizvo.

Iyi puratifomu inopawo mazano evanyori vane hukama uye zvinyorwa zvinoenderana nenhoroondo yako yekutsvaga. Kurudziro idzi hadzisi generic rondedzero: dzakabva pamasemantic mapatani uye citation network, saka vanowanzo kufumura tambo dzaungave usina kufunga. Mukuita, aya mazano anokutendera kuti uteedzere nzira yekuverenga ine zvibereko uye wedzera nzvimbo yekuongorora kwakarongeka.

Rimwe remasimba eSemantic Scholar maonerwo aanoita network yekududza uye zvinyorwa zvinobatanidza mabasa akasiyana. Unogona kuona node dzine simba guru, kuona zvikoro zvepfungwa, uye, neimwe tsika, unongedza zvidimbu zvinoshanda semahinji pakati pemakambani akasiyana. Izvi zvinoita kuti zvive nyore kuwana ruzivo rwakakodzera. seminal mapepa uye transversal research nzira.

Iwe unofarira here chinyorwa chisina PDF inowanikwa pachikuva? Usazvinetse: unogona kuzvitsvaga pawebhusaiti yemuparidzi, munzvimbo dzinochengeterwa masangano, kana, kana ukashanda neraibhurari yeyunivhesiti, bvunza vashandi vereferensi kuti vakupe nhungamiro yekuwana zvinyorwa zvizere. Kubatanidza Semantic Scholar neaya chiteshi inzira inoshanda yeku vhara loop uye uwane zviri mukati.

Zano rinobatsira: Paunenge uchiongorora musoro mutsva, batanidza kuongorora kwekutanga nemasefa akafara wobva wakwenenzvera nemamiriro ekudzora (semuenzaniso, zvinyorwa zvemethodheki chete kana wongororo). Iyi nzira yekudzokorora, pamwe chete nemasaini epesvedzero uye kuteedzera kuteedzera, inokubatsira iwe kuvaka yemhando yebhaibheri uye Kudzikama kudzika nekuvhara.

Misiyano neGoogle Scholar uye PubMed

Google Scholar nePubMed mbiru dze ecosystem, asi pfungwa dzavo dzakagara dzichitsamira pakuverenga zvirevo, zvinyorwa chaizvo, uye izwi co-kushandisa. Semantic Scholar inosuma imwe layer: AI-powered mamiriro ekuverenga anoedza kunzwisisa zvinorehwa negwaro uye kubatana. Shanduko iyi inobvumira rongedzazve mibairo yakananga kune anesimba zvakanyanya muhurukuro yega yegakwete kungoenda kumisoro inogara ichitaurwa.

Chimwe chakanakira chiratidzo chemhando maererano nekushandiswa kwechinyorwa mumabasa anodudza. Kuziva kana basa rinosanganisirwa semashure kana senzira inopa nuances isingawanzo kutorwa neachinyakare kutsvaga injini. Yakasanganiswa nepfupiso yemutsara mumwe uye kudhirowa kwemasangano nenhamba, izvi zvinopa tarisiro yekukurumidza iyo inomhanyisa kuongorora kwekutanga kukosha.

Nekudaro, iyo inonyanya kushanda nzira ndeyekuvashandisa nenzira inopindirana: Google Scholar kune yakakura kuvharika kwayo, PubMed yekutsvaga biomedical ine terminology control, uye Semantic Scholar kukoshesa pesvedzero chaiyo uye semantic kubatana. Nekuvasanganisa, iwe unowedzera mukana wekusapotsa chero chinhu chakakosha uye che svika pazvinhu zvinounza mutsauko kutanga.

Common use kesi

Kana iwe uchitanga mutsara mutsva wekutsvagisa, unogona kushandisa chirevo-chimwe-pfupiso yekukurumidza kuravira kwekutanga. Zvadaro, uchishandisa citation metrics uye ma tag ekupesvedzera, unonatsa sarudzo yako kusvika wasara neseti yezvinyorwa zvakakosha. Iyi mafambiro ebasa inopa kukurumidza track ku kuenda kubva ku zero kuenda kumepu yepfungwa yemunda mumaawa mashoma.

Kuti urambe uripo, girafu-ne-gore rinobatsira kuona mapepa anoramba achitaurwa kakawanda. Kana bepa rikaramba rakagadzikana (kana kunyange kumusoro) curve, chiratidzo chekuti rinoramba rakakodzera uye rakakodzera nzvimbo pane yako yekutanga kuverenga runyorwa. Iyi nguva-yekuverenga kuverenga inobatsira kusiyanisa fadhi dzinopfuura kubva pamipiro yenguva refu.

Mukunyora kwepurojekiti kana mishumo, 'apo akadudzwa' ma tag akakosha: anopembedza kuti nzira yakanyatsogadziriswa kana chinyorwa chinowanzo taurwa muzvikamu zvemethodheji, kana kuti dzidziso yakavakirwa zvakanaka kana ichitonga ruzivo rwekumashure. Kududza mukati mechirevo chechinyorwa chino kunopa rondedzero inomanikidza pamusoro simba uye mari yeuchapupu.

Mukudzidzisa, aya maficha anobatsira kuvaka kuverenga kunotungamirwa: unogona kusimbisa zvinyorwa zvakadudzwa senheyo dzedzidziso uye zvimwe zvinoshandiswa pahunyanzvi hwavo. Uyezve, nekuratidza hukama pakati pemabasa, zviri nyore kugadzira nzira dzekudzidza kutsanangura kuti pfungwa inoshanduka sei munzvimbo diki. Izvi zvinoita Semantic Scholar a chishandiso chekudzidzisa chinobatsira sebhuku racho pacharo.

Semantic Scholar inosanganisa masaini ehuwandu uye emhando, anobvisa zvinoreva neAI, uye anoronga mabhuku ekufambisa achitenderedza uye mamiriro. Paunenge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge uchinge nguva, kuona maitiro chaiwo, uye kuvaka yakanyatsofungwa bhaibheri, iyi puratifomu inova mubatsiri akakosha. Inoderedza ruzha uye inotarisa pane zvakakosha..