Transformer model with external token memories and attention for PersonaChat

Many existing studies aim to develop a dialog system capable of acting as efficiently and accurately as humans. The prevailing approach involves using large machine-learning models and extensive datasets for training to ensure that token information and the connections between them exist solely with...

Celý popis

Uložené v:

Podrobná bibliografia
Vydané v:	Scientific reports Ročník 15; číslo 1; s. 20691 - 11
Hlavní autori:	Sun, Taize, Fujita, Katsuhide
Médium:	Journal Article
Jazyk:	English
Vydavateľské údaje:	London Nature Publishing Group UK 01.07.2025 Nature Portfolio
Predmet:	639/705/117 639/705/258 Attention mechanism Dialogue system Encoder–decoder model External memory Fine-tuning Humanities and Social Sciences multidisciplinary Persona-Chat Science Science (multidisciplinary) Attention mechanism Persona-Chat Dialogue system External memory Encoder–decoder model Fine-tuning
ISSN:	2045-2322, 2045-2322
On-line prístup:	Získať plný text
Tagy:	Pridať tag Žiadne tagy, Buďte prvý, kto otaguje tento záznam!

Abstract	Many existing studies aim to develop a dialog system capable of acting as efficiently and accurately as humans. The prevailing approach involves using large machine-learning models and extensive datasets for training to ensure that token information and the connections between them exist solely within the model structure. This paper introduces a transformer model with external token memory and attention (Tmema) that is inspired by humans’ ability to define and remember each object in a chat. Tmema can define and remember each object or token in its memory, which is generated through random initialization and updated using backpropagation. In the model’s encoder, we utilized a bidirectional self-attention mechanism and external memory to compute the latent information for each input token. When generating text, the latent information is synchronously added to the corresponding external attention of the token in the one-way self-attention decoder, enhancing the model’s performance. We demonstrate that our proposed model outperforms state-of-the-art approaches on the public PersonaChat dataset across automatic and human evaluations. All code and data used to reproduce the experiments are freely available on https://github.com/Ozawa333/Tmema .
AbstractList	Many existing studies aim to develop a dialog system capable of acting as efficiently and accurately as humans. The prevailing approach involves using large machine-learning models and extensive datasets for training to ensure that token information and the connections between them exist solely within the model structure. This paper introduces a transformer model with external token memory and attention (Tmema) that is inspired by humans' ability to define and remember each object in a chat. Tmema can define and remember each object or token in its memory, which is generated through random initialization and updated using backpropagation. In the model's encoder, we utilized a bidirectional self-attention mechanism and external memory to compute the latent information for each input token. When generating text, the latent information is synchronously added to the corresponding external attention of the token in the one-way self-attention decoder, enhancing the model's performance. We demonstrate that our proposed model outperforms state-of-the-art approaches on the public PersonaChat dataset across automatic and human evaluations. All code and data used to reproduce the experiments are freely available on https://github.com/Ozawa333/Tmema . Many existing studies aim to develop a dialog system capable of acting as efficiently and accurately as humans. The prevailing approach involves using large machine-learning models and extensive datasets for training to ensure that token information and the connections between them exist solely within the model structure. This paper introduces a transformer model with external token memory and attention (Tmema) that is inspired by humans’ ability to define and remember each object in a chat. Tmema can define and remember each object or token in its memory, which is generated through random initialization and updated using backpropagation. In the model’s encoder, we utilized a bidirectional self-attention mechanism and external memory to compute the latent information for each input token. When generating text, the latent information is synchronously added to the corresponding external attention of the token in the one-way self-attention decoder, enhancing the model’s performance. We demonstrate that our proposed model outperforms state-of-the-art approaches on the public PersonaChat dataset across automatic and human evaluations. All code and data used to reproduce the experiments are freely available on https://github.com/Ozawa333/Tmema . Abstract Many existing studies aim to develop a dialog system capable of acting as efficiently and accurately as humans. The prevailing approach involves using large machine-learning models and extensive datasets for training to ensure that token information and the connections between them exist solely within the model structure. This paper introduces a transformer model with external token memory and attention (Tmema) that is inspired by humans’ ability to define and remember each object in a chat. Tmema can define and remember each object or token in its memory, which is generated through random initialization and updated using backpropagation. In the model’s encoder, we utilized a bidirectional self-attention mechanism and external memory to compute the latent information for each input token. When generating text, the latent information is synchronously added to the corresponding external attention of the token in the one-way self-attention decoder, enhancing the model’s performance. We demonstrate that our proposed model outperforms state-of-the-art approaches on the public PersonaChat dataset across automatic and human evaluations. All code and data used to reproduce the experiments are freely available on https://github.com/Ozawa333/Tmema . Many existing studies aim to develop a dialog system capable of acting as efficiently and accurately as humans. The prevailing approach involves using large machine-learning models and extensive datasets for training to ensure that token information and the connections between them exist solely within the model structure. This paper introduces a transformer model with external token memory and attention (Tmema) that is inspired by humans' ability to define and remember each object in a chat. Tmema can define and remember each object or token in its memory, which is generated through random initialization and updated using backpropagation. In the model's encoder, we utilized a bidirectional self-attention mechanism and external memory to compute the latent information for each input token. When generating text, the latent information is synchronously added to the corresponding external attention of the token in the one-way self-attention decoder, enhancing the model's performance. We demonstrate that our proposed model outperforms state-of-the-art approaches on the public PersonaChat dataset across automatic and human evaluations. All code and data used to reproduce the experiments are freely available on https://github.com/Ozawa333/Tmema .Many existing studies aim to develop a dialog system capable of acting as efficiently and accurately as humans. The prevailing approach involves using large machine-learning models and extensive datasets for training to ensure that token information and the connections between them exist solely within the model structure. This paper introduces a transformer model with external token memory and attention (Tmema) that is inspired by humans' ability to define and remember each object in a chat. Tmema can define and remember each object or token in its memory, which is generated through random initialization and updated using backpropagation. In the model's encoder, we utilized a bidirectional self-attention mechanism and external memory to compute the latent information for each input token. When generating text, the latent information is synchronously added to the corresponding external attention of the token in the one-way self-attention decoder, enhancing the model's performance. We demonstrate that our proposed model outperforms state-of-the-art approaches on the public PersonaChat dataset across automatic and human evaluations. All code and data used to reproduce the experiments are freely available on https://github.com/Ozawa333/Tmema .
ArticleNumber	20691
Author	Sun, Taize Fujita, Katsuhide
Author_xml	– sequence: 1 givenname: Taize surname: Sun fullname: Sun, Taize email: s227587t@st.go.tuat.ac.jp organization: Tokyo University of Agriculture and Technology – sequence: 2 givenname: Katsuhide surname: Fujita fullname: Fujita, Katsuhide organization: Tokyo University of Agriculture and Technology
BackLink	https://www.ncbi.nlm.nih.gov/pubmed/40594946$$D View this record in MEDLINE/PubMed
BookMark	eNp9kc1u3CAUhVGVKEmTvEAXFctu3PJrw6qqRv2JNFK7SNYIm8uMpzakwKSdty-J0yjZlA0Izvmu7O81OgoxAEJvKHlPCVcfsqBSq4Yw2WilJGkOr9AZI0I2jDN29Ox8ii5z3pG6JNOC6hN0KojUQov2DK2vkw3ZxzRDwnN0MOHfY9li-FMgBTvhEn9CwDPMMY2QsQ0O21IglDEGXHv4B6Qcg11tbblAx95OGS4f93N08-Xz9epbs_7-9Wr1ad0Mom1LA97xTnreK819Jx2Tg2DSixZop0GRQfvWEeYpUCG1Y174oaWsV5z3xCrg5-hq4bpod-Y2jbNNBxPtaB4uYtoYm8o4TGB62jvZgreWa0EcVZXZOkvq6PoPOl1ZHxfW7b6fwQ31y5KdXkBfvoRxazbxzlDGqOK6q4R3j4QUf-0hFzOPeYBpsgHiPpuqoOVSMq5q9O3zYU9T_vmoAbYEhhRzTuCfIpSYe-9m8W6qd_Pg3RxqiS-lXMNhA8ns4v5eXv5f6y-z87El
Cites_doi	10.18653/v1/P18-1205 10.3115/v1/D14-1179 10.18653/v1/2021.eacl-main.74 10.1207/s15516709cog1402_1 10.1609/aaai.v32i1.11923 10.3115/1073083.1073135 10.1016/S0022-5371(69)80069-1 10.18653/v1/2020.acl-main.131 10.18653/v1/2020.acl-main.703 10.18653/v1/P19-1608 10.18653/v1/2021.emnlp-main.169 10.1007/978-3-030-29135-8_7 10.1162/neco.1989.1.2.270 10.18653/v1/P19-1363 10.1609/aaai.v37i11.26489 10.1162/neco.1997.9.8.1735 10.1609/aaai.v34i05.6518 10.18653/v1/2021.acl-long.14 10.1016/j.mlwa.2024.100541
ContentType	Journal Article
Copyright	The Author(s) 2025 2025. The Author(s). The Author(s) 2025 2025
Copyright_xml	– notice: The Author(s) 2025 – notice: 2025. The Author(s). – notice: The Author(s) 2025 2025
DBID	C6C AAYXX CITATION NPM 7X8 5PM DOA
DOI	10.1038/s41598-025-98850-y
DatabaseName	Springer Nature OA Free Journals CrossRef PubMed MEDLINE - Academic PubMed Central (Full Participant titles) DOAJ Open Access Full Text
DatabaseTitle	CrossRef PubMed MEDLINE - Academic
DatabaseTitleList	PubMed MEDLINE - Academic
Database_xml	– sequence: 1 dbid: DOA name: DOAJ Open Access Full Text url: https://www.doaj.org/ sourceTypes: Open Website – sequence: 2 dbid: NPM name: PubMed url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed sourceTypes: Index Database – sequence: 3 dbid: 7X8 name: MEDLINE - Academic url: https://search.proquest.com/medline sourceTypes: Aggregation Database
DeliveryMethod	fulltext_linktorsrc
Discipline	Biology
EISSN	2045-2322
EndPage	11
ExternalDocumentID	oai_doaj_org_article_b1bd56efaa3940d18e146da089305979 PMC12218397 40594946 10_1038_s41598_025_98850_y
Genre	Journal Article
GroupedDBID	0R~ 4.4 53G 5VS 7X7 88E 88I 8FE 8FH 8FI 8FJ AAFWJ AAJSJ AAKDD AASML ABDBF ABUWG ACGFS ACUHS ADBBV ADRAZ AENEX AEUYN AFKRA AFPKN ALIPV ALMA_UNASSIGNED_HOLDINGS AOIJS AZQEC BAWUL BBNVY BCNDV BENPR BHPHI BPHCQ BVXVI C6C CCPQU DIK DWQXO EBD EBLON EBS ESX FYUFA GNUQQ GROUPED_DOAJ GX1 HCIFZ HH5 HMCUK HYE KQ8 LK8 M1P M2P M7P M~E NAO OK1 PHGZM PHGZT PIMPY PQQKQ PROAC PSQYO RNT RNTTT RPM SNYQT UKHRP AAYXX AFFHD CITATION PJZUB PPXIY PQGLB NPM 7X8 PUEGO 5PM
ID	FETCH-LOGICAL-c466t-efd375f3b893f75d25c425f46e179e80c9f6d02f1e1459d2f4fc612b833b0a8e3
IEDL.DBID	DOA
ISICitedReferencesCount	0
ISICitedReferencesURI	http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001522980000012&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN	2045-2322
IngestDate	Fri Oct 03 12:45:45 EDT 2025 Tue Nov 04 02:04:50 EST 2025 Tue Aug 26 08:58:37 EDT 2025 Mon Jul 21 06:03:38 EDT 2025 Sat Nov 29 07:46:56 EST 2025 Wed Jul 02 02:43:49 EDT 2025
IsDoiOpenAccess	true
IsOpenAccess	true
IsPeerReviewed	true
IsScholarly	true
Issue	1
Keywords	Attention mechanism Persona-Chat Dialogue system External memory Encoder–decoder model Fine-tuning
Language	English
License	2025. The Author(s). Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
LinkModel	DirectLink
MergedId	FETCHMERGED-LOGICAL-c466t-efd375f3b893f75d25c425f46e179e80c9f6d02f1e1459d2f4fc612b833b0a8e3
Notes	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23
OpenAccessLink	https://doaj.org/article/b1bd56efaa3940d18e146da089305979
PMID	40594946
PQID	3226355238
PQPubID	23479
PageCount	11
ParticipantIDs	doaj_primary_oai_doaj_org_article_b1bd56efaa3940d18e146da089305979 pubmedcentral_primary_oai_pubmedcentral_nih_gov_12218397 proquest_miscellaneous_3226355238 pubmed_primary_40594946 crossref_primary_10_1038_s41598_025_98850_y springer_journals_10_1038_s41598_025_98850_y
PublicationCentury	2000
PublicationDate	2025-07-01
PublicationDateYYYYMMDD	2025-07-01
PublicationDate_xml	– month: 07 year: 2025 text: 2025-07-01 day: 01
PublicationDecade	2020
PublicationPlace	London
PublicationPlace_xml	– name: London – name: England
PublicationTitle	Scientific reports
PublicationTitleAbbrev	Sci Rep
PublicationTitleAlternate	Sci Rep
PublicationYear	2025
Publisher	Nature Publishing Group UK Nature Portfolio
Publisher_xml	– name: Nature Publishing Group UK – name: Nature Portfolio
References	98850_CR29 98850_CR28 JL Elman (98850_CR16) 1990; 14 T Brown (98850_CR21) 2020; 33 C Raffel (98850_CR20) 2020; 21 98850_CR30 98850_CR10 98850_CR32 98850_CR31 98850_CR12 98850_CR11 98850_CR33 98850_CR14 98850_CR36 A Chowdhery (98850_CR5) 2023; 24 98850_CR13 98850_CR35 98850_CR15 98850_CR37 98850_CR6 98850_CR18 98850_CR8 98850_CR9 98850_CR19 98850_CR3 98850_CR4 L Ouyang (98850_CR22) 2022; 35 98850_CR1 98850_CR23 S Hochreiter (98850_CR17) 1997; 9 98850_CR25 98850_CR24 RJ Williams (98850_CR34) 1989; 1 A Radford (98850_CR2) 2019; 1 AM Collins (98850_CR7) 1969; 8 98850_CR27 98850_CR26
References_xml	– ident: 98850_CR35 doi: 10.18653/v1/P18-1205 – ident: 98850_CR18 doi: 10.3115/v1/D14-1179 – ident: 98850_CR29 doi: 10.18653/v1/2021.eacl-main.74 – volume: 14 start-page: 179 year: 1990 ident: 98850_CR16 publication-title: Cogn. Sci. doi: 10.1207/s15516709cog1402_1 – ident: 98850_CR33 – ident: 98850_CR19 – ident: 98850_CR8 doi: 10.1609/aaai.v32i1.11923 – ident: 98850_CR12 – ident: 98850_CR15 – ident: 98850_CR36 doi: 10.3115/1073083.1073135 – volume: 35 start-page: 27730 year: 2022 ident: 98850_CR22 publication-title: Adv. Neural Inf. Process. Syst. – ident: 98850_CR4 – volume: 21 start-page: 5485 year: 2020 ident: 98850_CR20 publication-title: J. Mach. Learn. Res. – volume: 8 start-page: 240 year: 1969 ident: 98850_CR7 publication-title: J. Verbal Learn. Verbal Behav. doi: 10.1016/S0022-5371(69)80069-1 – ident: 98850_CR30 doi: 10.18653/v1/2020.acl-main.131 – volume: 1 start-page: 9 year: 2019 ident: 98850_CR2 publication-title: OpenAI Blog – ident: 98850_CR13 doi: 10.18653/v1/2020.acl-main.703 – ident: 98850_CR25 – ident: 98850_CR37 doi: 10.18653/v1/P19-1608 – ident: 98850_CR23 – ident: 98850_CR9 doi: 10.18653/v1/2021.emnlp-main.169 – ident: 98850_CR26 – ident: 98850_CR28 – ident: 98850_CR6 doi: 10.1007/978-3-030-29135-8_7 – volume: 1 start-page: 270 year: 1989 ident: 98850_CR34 publication-title: Neural Comput. doi: 10.1162/neco.1989.1.2.270 – volume: 24 start-page: 1 year: 2023 ident: 98850_CR5 publication-title: J. Mach. Learn. Res. – ident: 98850_CR14 doi: 10.18653/v1/P19-1363 – ident: 98850_CR10 doi: 10.1609/aaai.v37i11.26489 – ident: 98850_CR11 – volume: 9 start-page: 1735 year: 1997 ident: 98850_CR17 publication-title: Neural Comput. doi: 10.1162/neco.1997.9.8.1735 – ident: 98850_CR32 – ident: 98850_CR27 doi: 10.1609/aaai.v34i05.6518 – ident: 98850_CR3 – ident: 98850_CR31 doi: 10.18653/v1/2021.acl-long.14 – ident: 98850_CR1 doi: 10.1016/j.mlwa.2024.100541 – volume: 33 start-page: 1877 year: 2020 ident: 98850_CR21 publication-title: Adv. Neural Inf. Process. Syst. – ident: 98850_CR24
SSID	ssj0000529419
Score	2.4532502
Snippet	Many existing studies aim to develop a dialog system capable of acting as efficiently and accurately as humans. The prevailing approach involves using large... Abstract Many existing studies aim to develop a dialog system capable of acting as efficiently and accurately as humans. The prevailing approach involves using...
SourceID	doaj pubmedcentral proquest pubmed crossref springer
SourceType	Open Website Open Access Repository Aggregation Database Index Database Publisher
StartPage	20691
SubjectTerms	639/705/117 639/705/258 Attention mechanism Dialogue system Encoder–decoder model External memory Fine-tuning Humanities and Social Sciences multidisciplinary Persona-Chat Science Science (multidisciplinary)
Title	Transformer model with external token memories and attention for PersonaChat
URI	https://link.springer.com/article/10.1038/s41598-025-98850-y https://www.ncbi.nlm.nih.gov/pubmed/40594946 https://www.proquest.com/docview/3226355238 https://pubmed.ncbi.nlm.nih.gov/PMC12218397 https://doaj.org/article/b1bd56efaa3940d18e146da089305979
Volume	15
WOSCitedRecordID	wos001522980000012&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText	1
inHoldings	1
isFullTextHit
isPrint
journalDatabaseRights	– providerCode: PRVAON databaseName: DOAJ Open Access Full Text customDbUrl: eissn: 2045-2322 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0000529419 issn: 2045-2322 databaseCode: DOA dateStart: 20110101 isFulltext: true titleUrlDefault: https://www.doaj.org/ providerName: Directory of Open Access Journals – providerCode: PRVHPJ databaseName: ROAD: Directory of Open Access Scholarly Resources customDbUrl: eissn: 2045-2322 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0000529419 issn: 2045-2322 databaseCode: M~E dateStart: 20110101 isFulltext: true titleUrlDefault: https://road.issn.org providerName: ISSN International Centre – providerCode: PRVPQU databaseName: Biological Science Database customDbUrl: eissn: 2045-2322 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0000529419 issn: 2045-2322 databaseCode: M7P dateStart: 20110101 isFulltext: true titleUrlDefault: http://search.proquest.com/biologicalscijournals providerName: ProQuest – providerCode: PRVPQU databaseName: Health & Medical Collection customDbUrl: eissn: 2045-2322 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0000529419 issn: 2045-2322 databaseCode: 7X7 dateStart: 20110101 isFulltext: true titleUrlDefault: https://search.proquest.com/healthcomplete providerName: ProQuest – providerCode: PRVPQU databaseName: ProQuest Central customDbUrl: eissn: 2045-2322 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0000529419 issn: 2045-2322 databaseCode: BENPR dateStart: 20110101 isFulltext: true titleUrlDefault: https://www.proquest.com/central providerName: ProQuest – providerCode: PRVPQU databaseName: ProQuest Publicly Available Content customDbUrl: eissn: 2045-2322 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0000529419 issn: 2045-2322 databaseCode: PIMPY dateStart: 20110101 isFulltext: true titleUrlDefault: http://search.proquest.com/publiccontent providerName: ProQuest – providerCode: PRVPQU databaseName: Science Database customDbUrl: eissn: 2045-2322 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0000529419 issn: 2045-2322 databaseCode: M2P dateStart: 20110101 isFulltext: true titleUrlDefault: https://search.proquest.com/sciencejournals providerName: ProQuest
link	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwrV3di9QwEB_0TsEX8dv6sUTwTculSdMkj95xh4K3FDlhfQpJk3CH2JXdnrD_vZO0e9yq6Isv89CmNJnJdH5pJr8BeF37RrlAedlRzcraS1paFmMptBLK4ayJ3uViE3I-V4uFbq-V-ko5YSM98Ki4A1c5L5oQrU01vH2lAvq2txTjLCIDmY_uUamvLaZGVm-m60pPp2QoVwdrjFTpNBkTpVZK0HKzE4kyYf-fUObvyZK_7JjmQHRyD-5OCJK8G3t-H26E_gHcHmtKbh7Cx7MtFA0rkuvckPSvlWzpnsmw_Bp68i1l2OIqmdjek8SxmbMeCT5H2hGfH53b4RF8Pjk-O3pfTiUTyq5umqEM0XMpIneoniiFZ6JDp4x1E9DxgqKdjo2nLFaoRaE9i3XsEOM4xbmjVgX-GPb6ZR-eAhFdFZ0LTlXCYiuna85liNSGUCvXqALebNVnvo_MGCbvaHNlRmUbVLbJyjabAg6Thq9aJlbrfAFtbSZbm3_ZuoBXW_sY9IK0tWH7sLxcG_wsJeSE-KOAJ6O9rl5VJ0oaXTcFqB1L7vRl905_cZ6ZtiuWEaQs4O3W6Gby8fVfBvvsfwz2Odxhabbm1OAXsDesLsNLuNX9GC7WqxnclAuZpZrB_uHxvP00yy6A8pS1SUqU--2H0_bLT6uuC2k
linkProvider	Directory of Open Access Journals
openUrl	ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Transformer+model+with+external+token+memories+and+attention+for+PersonaChat&rft.jtitle=Scientific+reports&rft.au=Sun%2C+Taize&rft.au=Fujita%2C+Katsuhide&rft.date=2025-07-01&rft.eissn=2045-2322&rft.volume=15&rft.issue=1&rft.spage=20691&rft_id=info:doi/10.1038%2Fs41598-025-98850-y&rft_id=info%3Apmid%2F40594946&rft.externalDocID=40594946
thumbnail_l	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=2045-2322&client=summon
thumbnail_m	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=2045-2322&client=summon
thumbnail_s	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=2045-2322&client=summon