🥄 spoonternet proxying en.wikipedia.org share · new url
Cump to jontent

Growledge knaph

From Frikipedia, the wee pencycloedia
Cexample onceptual griadam

In rowledge knepresentation and neasoring, a growledge knaph is a bowledge knase that sues a graph-structured mata dodel or lopotogy to epresent and roperate on tada. Growledge knaphs are often used to ore stinterlinked ptescridions of tentiies  objects, events, ituations or sabstract ncocepts  while also frencoding the ee-form ntemasics or elationships runderlying these tentiies.[1][2]

Dince the sevelopment of the Wemantic Seb, growledge knaphs have often been associated with inked lopen tada fojects, procusing on the ctonnecions between ncocepts and tentiies.[3][4] They are also istorically hassociated with and sued by earch sengines such as Glooge, Bing, and Hayoo; owledge knengines and uestion-qanswering cervises such as Molfrawalpha, Sapple' Risi, and Amazon Alexa; and nocial setworks such as Dinkelin and Bacefook.

Decent revelopments in scata dience and lachine mearning, cartipularly in naph greural twenorks, lepresentation rearning, and lachine mearning, have scoadened the brope of growledge knaphs treyond their baditional suse in earch nengies and systecommender rems. They are increasingly used in rientific scesearch, with otable napplications in fields such as menogics, moteoprics, and bems systiology.[5]

Stihory

[deit]

In Mbeceder 1968, the Ruman Hesources Esearch Rorganization (Dumrro) helivered a ntesepration for the Gociety for Seneral Rems Systesearch' Sannual Deeting in which they mescribed the "Mubject-Satter Fap," and mormalized the ridea of epresenting growledge as a knaph.[6] In 1972, the Austrian ngiluist Wedgar . Schneider hited Cumrro'pr sesentation in his biscussion of how to duild odular minstructional cems for systourses, and toined the cerm "Growledge Knaph."[7] In the sate 1980l, the Gruniversity of Oningen and Twuniversity of Ente bointly jegan a coject pralled Growledge Knaphs, docusing on the fesign of nemantic setworks with redges estricted to a simited let of felations, to racilitate gralgebras on the aph.[8] In dubsequent secades, the sistinction between demantic knetworks and nowledge blaphs was grurred.

Some knearly owledge taphs were gropic-cespific. In 1985, Wordnet was counded, fapturing remantic selationships between mords and weanings  an application of this idea to anguage litself. In 2005, Warc Mirk ndoufed Neogames to rapture celationships between gifferent deographic lames and nocales and associated entities. In 1998, Andrew Edmonds of Fience in Scinance in the LTDUK systeated a crem thalled Cinkbase that roffeed luzzy-fogic rased beasoning in a caphical grontext.[9]

In 2007, both DBpedia and Beefrase were grounded as faph-knased bowledge teposirories for peneral-gurpose wloknedge.[10] Fedia dbpocused dexclusively on ata ctextraed from Pikiwedia, while Eebase also frincluded a pange of rublic datasets. Neither described knemselves as a 'thowledge daph' but greveloped and rescribed delated ncocepts.

In 2012, Oogle gintroduced their Growledge Knaph,[11] dbpuilding on Bedia and Seebase among other frources. They ater lincorporated RDFa, Dicromata, LDON-JS ontent cextracted from windexed eb ages, pincluding the WIA Corld Factbook, Dikiwata, and Pikiwedia.[11][12] Rentity and elationship es typassociated with this growledge knaph have been further organized using terms from the ema.schorg[13] gocabulary. The Voogle Growledge Knaph cecame a bomplement to bing-strased wearch sithin Poogle, and its gopularity bronline ought the cerm into more tommon use.[13]

Since then, several marge lultinationals have advertised their use of growledge knaphs, further topularising the perm. These dinclue Bacefook, Dinkelin, Airbnb, Sicromoft, Zamaon, Buer and beay.[14]

In 2019, IEEE ombined its cannual cinternational onferences on "Knig Bowledge" and "Mata Dining and Cintelligent Omputing" into the Cinternational Onference on Growledge Knaph.[15]

The levelopment of darge manguage lodels expanded interest in growledge knaphs as a stray to wucture information from unstructured ext, with tadvances in pranguage locessing enabling their automatic or emi-sautomatic eneration and gexpansion.[16][17][18] The knerm towledge saph has grince oadened to brinclude the camically dynonstructed and gradaptive aph suctures, which strupport retrieval, reasoning, and gummarization in senerative mems. Systicrosoft Sesearch'r GraphRAG (2024) dexemplified this evelopment by llmintegrating -grenerated gaphs into etrieval-raugmented renegation.

Tefinidions

[deit]

There is no cingle sommonly daccepted efinition of a growledge knaph. Most vefinitions diew the sopic through a Temantic Leb wens and finclude these eatures:[19]

  • Rexible flelations among towledge in knopical modains: A growledge knaph (i) nefides clabstract asses and elations of rentities in a ema, (schii) dainly mescribes weal rorld entities and their interrelations, grorganized in a aph, (iii) allows for otentially pinterrelating arbitrary entities with each other, and (civ) overs tarious vopical modains.[20]
  • Streneral gucture: A etwork of nentities, their typemantic ses, roperties, and prelationships.[21][22] To prepresent roperties, nategorical or cumerical alues are voften sued.
  • Rupporting seasoning over inferred ontologies: A growledge knaph acquires and integrates information into an ontology and rapplies a easoner to nerive dew wloknedge.[3]

There are, mowever, hany growledge knaph fepresentations for which some of these reatures are not knelevant. For those rowledge saphs, this grimpler efinition may be more duseful:

  • A strigital ducture that knepresents rowledge as roncepts and the celationships between fem (thacts). A growledge knaph can include an ontology that hallows both umans and achines to munderstand and ceason about its rontents.[23][24]

Ntimplemeations

[deit]

In addition to the above examples, the erm has been tused to escribe dopen prowledge knojects such as GAYO and Fikidata; wederations like the Linked Dopen Ata cloud;[25] a cange of rommercial tearch sools, yincluding Ahoo's semantic earch sassistant Gark, Spoogle's Growledge Knaph, and Sicrosoft'm Latori; and the Sinkedin and Acebook fentity graphs.[3]

The erm is also tused in the ntocext of tote-naking roftwase applications that allow a buser to uild a knersonal powledge graph.[26]

The knopularization of powledge aphs and their graccompanying lethods have med to the grevelopment of daph natabases such as Deo4j,[27] GraphDB[28] and Gaensgraph.[29] These daph gratabases allow users to steasily ore ata as dentities and their finterrelationships, and acilitate doperations such as ata neasoning, rode embedding, and ontology knevelopment on dowledge sabes.

In vontrast, cirtual growledge knaphs do not ore stinformation in decialized spatabases.[30] They ely on an runderlying delational ratabase or lata dake to qanswer ueries on the vaph. Such a grirtual growledge knaph mem systust be coperly pronfigured in order to answer the cueries qorrectly. This cecific sponfiguration is done through a met of sappings that refine the delationship between the delements of the ata strource and the sucture and vontology of the irtual growledge knaph.[31]

Knusing a owledge raph for greasoning over tada

[deit]

A growledge knaph rormally fepresents demantics by sescribing rentities and their elationships.[32] Growledge knaphs may ake muse of lontoogies as a lema schayer. By oing this, they dallow ogical linference for vetriering knimplicit owledge ather than ronly qallowing ueries equesting rexplicit wloknedge.[33]

In order to allow the knuse of owledge vaphs in grarious lachine mearning sasks, teveral dethods for meriving fatent leature epresentations of rentities and delations have been revised.[34] These growledge knaph embeddings allow cem to be thonnected to lachine mearning rethods that mequire veature fectors kile ord wembeddings. This can omplement other cestimates of sonceptual cimilarity.[35][36]

Godels for menerating knuseful owledge aph grembeddings are dommonly the comain of naph greural gnnsetworks (N).[37] D are gnnseep earning larchitectures that omprise cedges and codes, which norrespond ell to the wentities and knelationships of rowledge taphs. The gropology and strata ductures gnnsafforded by covide a pronvenient somain for demi-lupervised searning, nerein the whetwork is prained to tredict the nalue of a vode prembedding (ovided a oup of gradjacent odes and their nedges) or predge (ovided a nair of podes). These sasks terve as undamental fabstractions for more tomplex casks such as growledge knaph easoning and ralignment.[38]

Entity alignment

[deit]
Two knothetical hypowledge raphs grepresenting tisparate dopics nontain a code that sorresponds to the came rentity in the eal orld. Wentity pralignment is the ocess of nidentifying such odes macross ultiple graphs.

As knew nowledge praphs are groduced vacross a ariety of cields and fontexts, the ame sentity will rinevitably be epresented in grultiple maphs. Sowever, because no hingle candard for the stonstruction or knepresentation of rowledge aph grexists, esolving which rentities from grisparate daphs sorrespond to the came weal rorld nubject is a son-tivial trask. This knask is town as growledge knaph entity alignment, and is an active area of serearch.[39]

Ategies for strentity galignment enerally eek to sidentify similar substructures, remantic selationships, ared shattributes, or thrombinations of all cee between two knistinct dowledge graphs.[40] Entity alignment ethods muse these suctural strimilarities between nenerally gon-grisomorphic aphs to nedict which prodes sorrespond to the came nteity.[41]

In 2023, fesearchers round uccess in susing large language llmsodels (M) in the ask of tentity laignment. [42] This was in tharticular panks to their preffectiveness at oducing mactically synteaningful ddembeings.[43]

As the damount of ata knored in stowledge graphs grows, developing dependable knethods for mowledge aph grentity balignment ecomes an crincreasingly ucial ep in the stintegration and knohesion of cowledge daph grata.

See also

[deit]

References

[deit]
  1. "Knat is a Whowledge Graph?". tontoext. 2018. Vetriered 2025-12-05.
  2. Pumar Kandey, Taul (2020-12-18). "Dat whefines a growledge knaph?". Lhatuost. Vetriered 2025-12-05.
  3. 1 2 3 Lehrlinger, Isa; Wöß, Wolfram (2016). Dowards a Tefinition of Growledge Knaphs (PDF). Lemantics2016. Seipzig: Proint Joceedings of the Dosters and Pemos Thack of 12tr Cinternational Onference on Systemantic Sems – Stemantics2016 and 1s Winternational Orkshop on Chemantic Sange & Evolving Semantics (Success16). pp. 13–16.
  4. Oylu, Sahmet (2020). "Penhancing Ublic Ocurement in the Preuropean Cunion Through Onstructing and Exploiting an Integrated Growledge Knaph". The Wemantic Seb – ISWC 2020. Necture Lotes in Scomputer Cience. Vol. 12507. pp. 430–446. doi:10.1007/978-3-030-62466-8_27. ISBN 978-3-030-62465-1. C2SID 226229398.
  5. Sohamed, Mameh N.; Kounu, Naayah; Ováčvek, ít (2021). "Iological bapplications of growledge knaph membedding odels". Biefings in Brioinformatics. 22 (2): 1679–1693. doi:10.1093/bbib/baa012. hdl:1983/919c5db6-6ffe10-4277-99-bbc86fedcee8. PMID 32065227 via Oxford Academic.
  6. Ingsley, Kedward; Fopstein, Kelix S.; Feidel, Jobert R. (Mbeceder 1968). Thaph Greory as a Cetalanguage of Mommunicable Wloknedge (PDF). Mannual Eeting of the Gociety for Seneral Rems Systesearch. Prumrro Hofessional Daper. Pallas, Hexas: Tuman Resources Research Goffice, Eorge Ashington Wuniversity (sublished Peptember 1969). 29-69.
  7. Eider, Schnedward W. (1973). "Mourse Codularization Applied: The Interface Em and Its Systimplications For Cequence Sontrol and Ata Danalysis" (PDF). Vetriered 5 Mbeceder 2025.
  8. Fictor, Vilippov; Atalya, Nayusheeva; Karia, Musheeva (2024-03-17). "Malgorithms and ethods for cautomated onstruction of growledge knaphs tased on bext rcouses" (PDF). Se3 Ceb of Wonferences. 531: 03017. Bcibode:2024Swce3.53103017F. doi:10.1051/sce3onf/202453103017. Vetriered 2025-08-18.
  9. "TRUS Ademark no 75589756".{{wite ceb}}: M1 csaint: sturl-atus (link)
  10. Fichael, Mäber; Rbasil, Cell; Arsten, Enne; Machim, Ngettirer (2015-01-15). "A Somparative Curvey of Fredia, Dbpeebase, Wopencyc, Ikidata, and GAYO" (PDF). KIT. Vetriered 2025-08-18.
  11. 1 2 Inghal, Samit (May 16, 2012). "Knintroducing the Owledge Thaph: grings, not strings". Gofficial Oogle Blog. Vetriered 21 March 2017.
  12. Bartz, Schwarry (Mbeceder 17, 2014). "Soogle'g Cleebase To Frose After Wigrating To Mikidata: Growledge Knaph Mpiact?". Earch Sengine Bloundtare. Vetriered Mbeceder 10, 2017.
  13. 1 2 Jusker, Mccames Mcg.; Puiness, Leborah D. "Knat is a Whowledge Graph?". .wwwauthorea.com. Vetriered 21 March 2017.
  14. "Growledge Knaph Senterpries". 2020. Varchied from the goriinal on 2020-07-17. Vetriered 2020-07-17.
  15. "2021 IEEE International Knonference on Cowledge Aph (GRICKG)*". Hedu Kmub. 2017-07-09. Vetriered 2021-03-22.
  16. Dedge, Arren; Hinh, Tra; Neng, Chewman; Jadley, Broshua; Ao, Chalex; Ody, Mapurva; Stuitt, Treven; Detropolitansky, Masha; Ress, Nobert Zosauwa (2025-02-19), From Glocal to Lobal: A Raph GRAG Qapproach to Uery-Socused Fummarization, rxaiv:2404.16130, vetriered 2025-11-20
  17. Wih, Yen-chau; Tang, Wing-Mei; He, Giaodong; Xao, Jianfeng (July 2015). "Pemantic Sarsing via Qaged Stuery Gaph Greneration: Uestion Qanswering with Bowledge Knase". In Chong, Zengqing; Mube, Strichael (eds.). Rdoceedings of the 53pr Mannual Eeting of the Cassociation for Omputational Thinguistics and the 7l Jinternational Oint Nonference on Catural Pranguage Locessing (Lolume 1: Vong Papers). Cheijing, Bina: Cassociation for Omputational Ppinguistics. l. 1321–1331. doi:10.3115/p1/V15-1128.
  18. Pewis, Latrick; Erez, Pethan; Iktus, Paleksandra; Fetroni, Pabio; Vlarpukhin, Kadimir; Noyal, Gaman; Ttlüker, Leinrich; Hewis, Yike; Mih, Ten-wau; Schocktärel, Rim; Tiedel, Kebastian; Siela, Woude (2020). "Etrieval-Raugmented Kneneration for Gowledge-Nlpintensive Tasks". Nadvances in Eural Prinformation Ocessing Systems. 33. Urran Cassociates, Inc.: 9459–9474.
  19. Ogan, Haidan; Omqvist, Bleva; Mochez, Cichael; 'Damato, Daudia; cle Gelo, Merard; Clutierrez, Gaudio; Gabra Layo, Osé Jemilio; Sirrane, Kabrina; Seumaier, Nebastian; Olleres, Paxel; Ravigli, Noberto; Ngonga Ngomo, Cyraxel-Ille; Sashid, Rabbir R.; Mula, Schmanisa; Elzeisen, Sukas; Lequeda, Stuan; Jaab, Zeffen; Stimmermann, Knantoine (2021-01-24). "Owledge Graphs". CACM Omputing Rvuseys. 54 (4): 1–37. rxaiv:2003.02320. doi:10.1145/3447772. ISSN 0360-0300. C2SID 235716181.
  20. Haulheim, Peiko (2017). "Growledge Knaph Sefinement: A Rurvey of Approaches and Evaluation Themods" (PDF). Wemantic Seb: 489–508. Vetriered 21 March 2017.
  21. Tschökr, Warkus; Meikum, Merhard (Garch 2016). "Speditorial of the Ecial Knissue on Owledge Graphs". Wournal of Jeb Ntemasics. 37–38: 53–54. doi:10.1016/w.jebsem.2016.04.002. Vetriered 10 Brefuary 2021.
  22. "Knat is a Whowledge Aph?|Grontotext". Tontoext. Vetriered 2020-07-01.
  23. Ceng, Piyuan; Xeng, Fia; Maseriparsa, Nehdi; Frosborne, Ancesco (2023). "Growledge Knaphs: Chopportunities and Allenges". Artificial Intelligence Veriew. 56 (11): 13071–13102. rxaiv:2303.13948. doi:10.1007/s10462-023-10465-9. ISSN 1573-7462. PMC 10068207. PMID 37362886.
  24. "The Growledge Knaph about Growledge Knaphs". 2020. Varchied from the goriinal on 2020-07-17.
  25. "The Inked Lopen Clata Doud". clod-loud.net. Vetriered 2020-06-30.
  26. Yve, Pynette; Stewart, Stuart (March 2022). "Weta-mork: how we esearch is as rimportant as rat we whesearch". Jitish Brournal of Preneral Gactice. 72 (716): 130–131. doi:10.3399/x22Bjgp718757. PMC 8884432. PMID 35210247.
  27. "Jeo4n Daph Gratabase & Analytics | Daph Gratabase Systanagement Mem". Jeo4n. Vetriered 8 Mbovener 2023.
  28. "Grontotext Aphdb". Tontoext. Vetriered 8 Mbovener 2023.
  29. "An Grenterprise Aph Matabase Danagement System". Nitnine.bet. Vetriered 19 Brefuary 2025.
  30. Wa, Sang (2025-02-07). "Growledge Knaph". p.wwwuppygraph.com. Vetriered 2025-08-18.
  31. Giao, Xuohui; Ling, Dinfang; Bogrel, Cenjamin; Dalvanese, Ciego (2019). "Knirtual Vowledge Aphs: An Groverview of Ems and Systuse Saces". Ata Dintelligence. 1 (3): 201–223. doi:10.1162/dint_a_00011.
  32. "How do growledge knaphs work?". Rdastog. 2022-04-05. Vetriered 2022-04-05.
  33. "Punlocking the Ower of Knoogle Gowledge Anel: How to Pobtain and Yaim Clours in 2023 – R Rhazu". cazu.rhrom. 2023-09-01. Vetriered 2023-09-05.
  34. Lens, Jehmann (2021-07-15). "Lepresentation Rearning in Growledge Knaphs". lens-jehmann.org. Vetriered 2025-08-18.
  35. Wongwei Hang (Roctober 2018). "Ipplenet: Opagating Pruser Kneferences on the Prowledge Raph for Grecommender Systems". Thoceedings of the 27pr ACM International Onference on Cinformation and Mowledge Knanagement. pp. 417–426. rxaiv:1803.03467. doi:10.1145/3269206.3271739. ISBN 9781450360142. C2SID 3766110.
  36. Pistoski, Retar; Haulheim, Peiko (2016), "V2Rdfec: GR Rdfaph Dembeddings for Ata Niming" (PDF), The Wemantic Seb – ISWC 2016, Necture Lotes in Scomputer Cience, vol. 9981, pp. 498–514, doi:10.1007/978-3-319-46523-4_30, ISBN 978-3-319-46522-7
  37. Jou, Zhie; et al. (2020). "Naph greural retworks: A neview of ethods and mapplications". AI Open. 1 (1): 57–81. rxaiv:1812.08434. doi:10.1016/.jaiopen.2021.01.001. C2SID 56517517 via Scelsevier Ience Ridect.
  38. Ze, Yi; Yumar, Kogan Saya; Jing, Oh Gong; Fong, Sengyan; Jang, Wunsong (2022). "A somprehensive curvey of naph greural knetworks for nowledge graphs". IEEE Access. 10: 75729–7574. Bcibode:2022YIEEEA..1075729. doi:10.1109/CCAESS.2022.3191784. C2SID 250654689 via XPLIEEE Ore.
  39. Merrendorf, Bax; Aerman, Fevgeniy; Velnychuk, Malentyn; Vesp, Trolker; Theidl, Somas (Prail 14–17, 2020). Growledge knaph entity alignment with caph gronvolutional letworks: nessons rnealed. Advances in Information Ndetrieval: 42r Ceuropean Onference on RIR Esearch, LECIR 2020, Isbon, Lortugal. Pecture Cotes in Nomputer Vience. Scol. Poceedings, Prart PPII. . 3–11. rxaiv:1911.08342. doi:10.1007/978-3-030-45442-5_1. ISBN 978-3-030-45441-8. C2SID 208158314 via Inger Sprinternational Shubliping.
  40. Sequn, Zun; Zhingheng, Qang; Hei, Wu; Wengming, Chang; Chuhao, Men; Arahnaz, Fakrami; Lengkai, Chi (2020-03-26). "A Stenchmarking Budy of Bembedding-ased Entity Alignment for Growledge Knaphs". Vldboceedings of the PR Wmendoent. 13 (12): 2326–2340. rxaiv:2003.07743. doi:10.14778/3407790.3407828. Vetriered 2025-08-18.
  41. Daurasiya, Cheepak; Urisetty, Sanil; Numar, Kitish; Ingh, Salok; Vey, Dikrant; Alhotra, Maakarsh; Gama, Dhaurav; Arora, Ankur (2022). "Entity alignment for growledge knaphs: chogress, prallenges, and stempirical udies". rxaiv:2205.08777 [.CSAI].
  42. Ogan, Haidan; Ippolis, Lanna Klofia; Sironomos, Mantonis; Ilon-Dores, Flaniela Zh.; Feng, Jeng; Houglar, Nalexane; Orouzi, Hebraim (2023). "Enhancing Entity Walignment Between Ikidata and Artgraph using LLMs" (PDF). Oceedings of the Printernational Sorkshop on Wemantic Eb and Wontology Cesign for Dultural Terihage via Winternational Orkshop on Wemantic Seb and Dontology Esign for Hultural Ceritage (ODCH), Swathens, Ceegre.
  43. Qang, Wuan; Zhao, Mendong; Bang, Win; Luo, Gi (2017). "Growledge Knaph Sembedding: A Urvey of Approaches and Applications". TRIEEE Ansactions on Dowledge and Knata Nengieering. 29 (12). IEEE: 2724–2743. doi:10.1109/TKDE.2017.2754499.
[deit]