-

THE ROLE OF RECOMMENDERS IN CULTURAL HERITAGE

George Pavlidis

0 0 Research Director, Athena Research Centre

Recommenders are, typically, systems that exploit knowledge regarding preferences of users on a set of items, in order to create user recommendations for unknown items. Recommenders are meant to create meaningful recommendations, enhancing the content personalisation and reducing the information overload. Applications of this type of technology have already appeared in the domain of cultural heritage, mainly in the form of museum and tourism recommenders. This paper reviews and explores the role of recommenders in cultural heritage and briefly discusses the main concepts, limitations, challenges and future directions.

- artificial intelligence cultural heritage museum guide recommendation recommender

Tapestry (Goldberg, Nichols, Oki, & Terry, 1992) and GroupLens (Resnick, Iacovou, Suchak, Bergstrom, & Riedl, 1994) , are considered to be the fathers of modern recommenders. It was then that a highly persistent approach in this field, collaborative filtering, was introduced, formulating the problem as an optimisation, seen either as a minimisation of a cost of an inaccurate prediction, or as a maximisation of a user’s utility or satisfaction (Good et al., 1999; Iaquinta, de Gemmis, Lops, Semeraro, & Molino, 2010) . In the bibliography in this domain three approaches are identified, the contentbased systems, the collaborative filtering systems, and the hybrid systems (Adomavicius & Tuzhilin, 2005; Aggarwal, 2016b; Anand & Mobasher, 2005; Bobadilla, Ortega, Hernando, & Gutiérrez, 2013; Jannach, Zanker, Felfernig, & Friedrich, 2011; Kaminskas & Ricci, 2012; Konstan, 2004; Lü et al., 2012; Ricci, Rokach, Shapira, & Kantor, 2011) . Various evaluation methods applicable to recommenders and an in- depth discussion can be found in Aggarwal (2016c).

This paper focuses on their recommender system applications in cultural heritage, where they appeared, mainly, as techniques to enhance museum visits and tourism applications. The most significant advances are being listed and a brief discussion concludes on the main concepts, the limitations, the challenges and possible future directions.

Recommenders in Cultural Heritage

There is a high volume of published works related to recommenders in the cultural heritage domain. An identification of the most influential and interesting works in this field leads to a list of around two innovations per year since 1999. This section lists some of those works in a chronological order to convey the essence of the progress in this domain.

In 1999 the Hippie guide (Oppermann & Specht, 1999) was developed as an electronic guide for adaptive exhibition guidance. The innovation included exploiting awareness of visitor location and user modelling.

In 2002, the Sotto Voce (Aoki et al., 2002) was developed as a PDA audio guide focused on social aspects of museum visits, by supporting a mediated sharing of audio content (termed eavesdropping), and providing location-based recommendations.

Rocchi, Stock, Zancanaro, Kruppa, & Krüger (2004) developed a mobile system focused on cinematic techniques to enhance engagement, using also user localisation.

Chou, Hsieh, Gandon, & Sadeh (2005 ) developed a collection of PDA applications that adapted the recommendations to visitor profiles and visitor behaviours, focusing of context awareness using a number of sensing technologies, based on approaches by Miller et al. (2004) .

The ARCHIE mobile guide (Luyten et al., 2006) focused on social awareness, influenced by studies like (Falk & Dierking, 2000) , in which Wi-Fi-based visitor localisation was used.

Grieser, Baldwin, & Bird (2007 ), presented a recommender based on user modelling and item features, extracted from textual descriptions, using the typical f − approach for the text- based similarity estimation and a probabilistic approach to assess the likelihood of a path.

Basile et al. (2008) , under the framework of the CHAT project, developed a content-based recommender capable of learning user profiles from static and usergenerated content, as a type of extension of the ITem Recommender (Degemmis, Lops, & Semeraro, 2007) .

Luh & Yang (2008) focused on recommendations based on visitor lifestyles based on collaborative filtering and a set of lifestyle factors proposed by the authors.

Between 2007 and 2010 several versions of a museum recommender were developed under the framework of project CHIP (Rijksmuseum). Initially, Pechenizkiy & Calders (2007) developed a content-based personalisation framework. Wang et al. (2008) proposed the creation of recommendations based on semantically-enriched museums collections adopting, again, a content-based approach for PDAs and sensor based localisation. A 2009 version appeared in a student research competition, focused on a mobile implementation for on-site museum visits (Roes, Stash, Wang, & Aroyo, 2009) . Van Hage, Stash, Wang, & Aroyo (2010 ) presented a more advanced version of the system, equipped with routing functionalities based on localisation information.

Huang, Liu, Lee, & Huang (2012) developed a personalised guide, focusing on museum learning settings, based on a rule-based recommender. The interesting in this work was that it seems to be among the first to present an experimental evaluation design that targeted user satisfaction factors, as defined in Ong, Day, & Hsu (2009 ).

Maehara, Yatsugi, Kim, & Ushiama (2012 ) developed a recommender that relies on a semantic network on museum exhibits based on item relations and user preferences, taking into account the limited timeframe of a visit.

Benouaret & Lenne (2015) proposed a combination of semantics (content-based) and collaborative filtering to create personalised museum tours, on smart mobile devices. The researchers used relevance, contextual information, time limitations, localisation, even weather information, to provide accurate context-aware recommendations.

Keller & Viennet (2015) presented a recommender within the AMMICO project focused on enhanced audio guidance in museum tours, claiming to tackle the challenges of the cold-start, the data sparsity, and an inherent over-specialisation as expressed in (Ardissono, Kuflik, & Petrelli, 2012).

Rossi, Barile, Improta, & Russo (2016 ) developed a collaborative filtering-based system to increase both individual and group visitor satisfaction, adopting matrix factorisation, along with localisation aspects.

Tavcar, Antonya, & Butila (2016 ) designed a hybrid recommender system within the eHERITAGE project that is based upon strong mash-up approach influences, combining technologies such as intelligent virtual assistants, Google Street View and recommenders.

Hashemi & Kamps (2017) developed a hybrid recommender within project meSch, adopting the free-roaming museum visit model, thus using localisation, online and onsite user behaviours, and content- and context-awareness.

Cardoso, Rodrigues, Pereira, & Sardo (2017) developed an association rule- based approach, within project M5SAR, clearly a hybrid method for museum visit recommendation, capable of supporting multiple visitors and multiple museums and sites, using the Apriori algorithm (Agrawal, Srikant, et al., 1994) to learn the rules, and utilised data from the open dataset of MoMA (Robot, 2018).

Kovavisaruch, Sanpechuda, Chinda, Sornlertlamvanich, & Kamolvej (2017) developed a probabilistic approach for a system capable of evaluating visitor paths in order to assess the effectiveness of a given museum exhibition organisation. Although the system naturally supports museum curators and exhibition designers, a very simplistic approach was described to exploit the model learned for visit recommendations.

Discussion

Most of the works on recommenders in cultural heritage, still conceptualise the museum as a gallery-like institution with linear narratives, which reveals a trend in the assumed visitor models or motivations by the involved researchers. This is a conception that needs to be revised since in the recent years the museums are transforming, mainly due to sustainability issues, adopting a different role closely related to education, study and enjoyment.

It is indicative that the “New toolkit for museum and heritage education” by ICOMCECA proposed eighteen methods a museum can use to enhance its educational services (ICOM- CECA. (2017, Oct.). In this report it is evident that the social and participatory factors are distributed among all the types of experiences, although there are strong requirements for support from the stakeholder, as these approaches need interesting storytelling, careful storyline organisation and fascinating narratives, meaningful and illustrative content and contextual structuring.

Social engagement and visitor participation aspects have already been considered, but the stakeholder’s role has not been properly defined and included in the loop, by being described as a type of repository curator. Although a modern view of an institution like a museum includes heavy investment on storytelling, history and narratives, aesthetics and education this has new view has not been considered yet.

From a technical point of view, hybrid recommender approaches have proven their strength in the cultural heritage domain, matching the complexity inherent in this domain. Since this is a highly dynamic domain with a large variance in tastes and various biases, online methods that consider context awareness, temporal dynamics and biased behaviours can be considered as most appropriate. Modern approaches using methods like reinforcement learning and agent- based techniques have not appeared yet.

In addition, semantic data and linked open data approaches need to be more seriously included in the technology in the domain, as more and more repositories and collections move towards international standards for data interoperability.

Location and context awareness can be easily integrated into cultural heritage recommenders as the Internet of Everything becomes more and more pervasive with easily applicable solutions, along with always connected and low-cost high-power ubiquitous computing, even in small-form devices.

The fast-developing intelligent virtual assistants technology is another important addition to cultural heritage recommenders, and there are already available cultural applications which incorporate intelligent guides.

Overall, there is room for further development both in the conceptualisation of the role of cultural institutions and the motivation of the visitors, and in the technologies that support an intelligent recommender. Impressive new developments are expected to appear in the near future.

Conclusion

Recommenders are artificial intelligence systems that have already been proven efficient in tackling information overload and personalisation in various contexts. Recommenders have appeared in the cultural heritage domain over the past decade to tackle personalisation in museum visits and cultural tourism applications. This paper reviewed works focused on cultural heritage applications of recommenders, a rather complex domain, in which basically hybrid approaches have been the most successful, although with limitations and assumptions. Challenges and benefits have been identified and a critical discussion on the reviewed approaches highlighted the foreseen future developments.

Acknowledgments

The present work was supported by the project “Computational Science and Technologies: Data, Content and Interaction/Technologies for Content Analysis in Culture”, MIS code 5002437, co-financed by Greece and European Union in the framework of the Operational Programme “Competitiveness, Entrepreneurship and Innovation” 2014-2020.

Adomavicius , G. , & Tuzhilin , A. ( 2005 ). Toward the next generation of recommender systems: A survey of the state-of-the-art and possible extensions . IEEE transactions on knowledge and data engineering , 17 ( 6 ), 734 - 749 .

Aggarwal , C. C. ( 2016b ). An introduction to recommender systems . In Recommender systems (pp. 1 - 28 ). Springer.

Aggarwal , C. C. ( 2016c ). Recommender systems: The textbook . Springer.

Agrawal , R. , Srikant , R. , et al. ( 1994 ). Fast algorithms for mining association rules . In Proc. 20th int. conf. very large data bases , vldb (Vol. 1215 , pp. 487 - 499 ). Aiolli, F. ( 2013 ). Efficient top-n recommendation for very large scale binary rated datasets . In Proceedings of the 7th acm conference on recommender systems (pp. 273 - 280 ).

Anand , S. S. , & Mobasher , B. ( 2005 ). Intelligent techniques for web personalization . In Intelligent techniques for web personalization (pp. 1 - 36 ). Springer.

Aoki , P. M. , Grinter , R. E. , Hurst , A. , Szymanski , M. H. , Thornton , J. D. , & Woodruff , A. ( 2002 ). Sotto voce: exploring the interplay of conversation and mobile audio spaces . In Proceedings of the sigchi conference on human factors in computing systems (pp. 431 - 438 ). Ardissono, L. , Kuflik , T. , & Petrelli , D. ( 2012 ). Personalization in cultural heritage: the road travelled and the one ahead. User modeling and user-adapted interaction , 22 ( 1-2 ), 73 - 99 .

Asanov , D. , et al. ( 2011 ). Algorithms and methods in recommender systems . Berlin Institute of Technology, Berlin, Germany.

Basile , P. , Calefato , F., de Gemmis , M. , Lops , P. , Semeraro , G. , Bux , M. , Cataldo , M. , Narducci , F. ( 2008 ). Augmenting a content-based recommender system with tags for cultural heritage personalization . Personalized Access to Cultural Heritage PATCH 2008 , 25 .

Benouaret , I. ,& Lenne , D. ( 2015 ). Combiningsemanticandcollaborativerecommendations to generate personalized museum tours . In East european conference on advances in databases and information systems (pp. 477 - 487 ).

Bobadilla , J. , Ortega , F. , Hernando , A. , & Gutiérrez , A. ( 2013 ). Recommender systems survey . Knowledge-based systems , 46 , 109 - 132 .

Cardoso , P. J. , Rodrigues , J. M. , Pereira , J. A. , & Sardo , J. D. ( 2017 ). An object visit recommender supported in multiple visitors and museums . In International conference on universal access in human-computer interaction (pp. 301 - 312 ).

Chou , S.-C. , Hsieh , W.-T., Gandon , F. L. , & Sadeh , N. M. ( 2005 ). Semantic web technologies for context-aware museum tour guide applications . In Advanced information networking and applications , 2005 . aina 2005 . 19th international conference on (Vol. 2 , pp. 709 - 714 ).

Degemmis , M. , Lops , P. , & Semeraro , G. ( 2007 ). A content-collaborative recommender that exploits wordnet-based user profiles for neighborhood formation . User Modeling and User- Adapted Interaction , 17 ( 3 ), 217 - 255 .

Falk , J. H. , & Dierking , L. D. ( 2000 ). Learning from museums: Visitor experiences and the making of meaning . Altamira Press.

Goldberg , D. , Nichols , D. , Oki , B. M. , & Terry , D. ( 1992 ). Using collaborative filtering to weave an information tapestry . Communications of the ACM , 35 ( 12 ), 61 - 70 .

Good , N. , Schafer , J. B. , Konstan , J. A. , Borchers , A. , Sarwar , B. , Herlocker , J. , & Riedl , J. ( 1999 ). Combining collaborative filtering with personal agents for better recommendations . In Aaai/iaai (pp. 439 - 446 ).

Grieser , K. , Baldwin , T. , & Bird , S. ( 2007 ). Dynamic path prediction and recommendation in a museum environment . In Proceedings of the workshop on language technology for cultural heritage data (latech 2007 ). (pp. 49 - 56 ).

Guttman , R. H. , Moukas , A. G. , & Maes , P. ( 1998 ). Agent-mediated electronic commerce: A survey . The Knowledge Engineering Review , 13 ( 2 ), 147 - 159 .

Hashemi , S. H. , & Kamps , J. ( 2017 ). Where to go next?: Exploiting behavioral user models in smart environments . In Proceedings of the 25th conference on user modeling, adaptation and personalization (pp. 50 - 58 ).

Huang , Y.-M. , Liu , C.-H. , Lee , C.-Y., & Huang , Y.-M. ( 2012 ). Designing a personalized guide recommendation system to mitigate information overload in museum learning . Journal of Educational Technology & Society , 15 ( 4 ). Iaquinta, L., de Gemmis , M. , Lops , P. , Semeraro , G. , & Molino , P. ( 2010 ). Can a recommender system induce serendipitous encounters? In E-commerce . InTech.

Jannach , D. , Zanker , M. , Felfernig , A. , & Friedrich , G. ( 2011 ). An introduction to recommender systems . New York: Cambridge.

Kaminskas , M. , & Ricci , F. ( 2012 ). Contextual music information retrieval and recommendation: State of the art and challenges . Computer Science Review , 6 ( 2- 3 ), 89 - 119 . Keller, I., & Viennet , E. ( 2015 ). Recommender systems for museums: Evaluation on a real dataset . In Fifth international conference on advances in information mining and management . Konstan, J. A. ( 2004 ). Introduction to recommender systems: Algorithms and evaluation . ACM Transactions on Information Systems (TOIS) , 22 ( 1 ), 1 - 4 .

Kovavisaruch , L.-o., Sanpechuda , T. , Chinda , K. , Sornlertlamvanich , V. , & Kamolvej , P. ( 2017 ). Museum layout evaluation based on visitor behavior and visiting suggestion under time constrain . International Journal of Business, Humanities and Technology , 7 ( 3 ), 5 - 18 .

Lü , L. , Medo , M. , Yeung , C. H. , Zhang, Y.-C., Zhang , Z.-K. , & Zhou , T. ( 2012 ). Recommender systems . Physics Reports , 519 ( 1 ), 1 - 49 .

Luh , D. , & Yang , T. ( 2008 ). Museum recommendation system based on lifestyles . In Computer-aided industrial design and conceptual design , 2008 . caid/cd 2008 . 9th international conference on (pp. 884 - 889 ).

Luyten , K. , Van Loon , H. , Teunkens , D. , Gabriëls , K. , Coninx , K. , & Manshoven , E. ( 2006 ). Archie: disclosing a museum by a socially-aware mobile guide . In 7th international symposium on virtual reality, archaeology and cultural heritage

Maehara , C. , Yatsugi , K. , Kim , D. , & Ushiama , T. ( 2012 ). An exhibit recommendation system based on semantic networks for museum . In Innovations in intelligent machines-2 (pp. 131 - 141 ). Springer.

Melville , P. , & Sindhwani , V. ( 2011 ). Recommender systems . In Encyclopedia of machine learning (pp. 829 - 838 ). Springer.

Miller , N. , Judd , G. , Hengartner , U. , Gandon , F. , Steenkiste , P. , Meng , I.-H. , Feng, MingWhei, Sadeh, N. ( 2004 ). Context-aware computing using a shared contextual information service . In Advances in pervasive computing. a collection of contributions presented at the 2nd international conference on pervasive computing (pervasive 2004 ).

Ong , C.-S. , Day , M.- Y. , & Hsu , W.-L. ( 2009 ). The measurement of user satisfaction with question answering systems . Information & Management , 46 ( 7 ), 397 - 403 .

Pechenizkiy , M. , & Calders , T. ( 2007 ). A framework for guiding the museum tours personalization . In Proceedings of the workshop on personalised access to cultural heritage (patch07) (pp. 11 - 28 ).

Resnick , P. , Iacovou , N. , Suchak , M. , Bergstrom , P. , & Riedl , J. ( 1994 ). Grouplens: an open architecture for collaborative filtering of netnews . In Proceedings of the 1994 acm conference on computer supported cooperative work (pp. 175 - 186 ).

Ricci , F. , Rokach , L. , Shapira , B. , & Kantor , P. B. ( 2011 ). Recommender systems handbook .

Rocchi , C. , Stock , O. , Zancanaro , M. , Kruppa , M. , & Krüger , A. ( 2004 ). The museum visit: generating seamless personalized presentations on multiple devices . In Proceedings of the 9th international conference on intelligent user interfaces (pp. 316 - 318 ).

Roes , I. , Stash , N. , Wang , Y. , & Aroyo , L. ( 2009 ). A personalized walk through the museum: The chip interactive tour guide . In Chi'09 extended abstracts on human factors in computing systems (pp. 3317 - 3322 ).

Rossi , S. , Barile , F. , Improta , D. , & Russo , L. ( 2016 ). Towards a collaborative filtering framework for recommendation in museums: from preference elicitation to group's visits . Procedia Computer Science , 98 , 431 - 436 .

Tavcar , A. , Antonya , C. , & Butila , E. V. ( 2016 ). Recommender system for virtual assistant supported museum tours . Informatica , 40 ( 3 ), 279 .

Van Hage , W. R. , Stash , N. , Wang , Y. , & Aroyo , L. ( 2010 ). Finding your way through the rijksmuseum with an adaptive mobile museum guide . In Extended semantic web conference (pp. 46 - 59 ).

Wang , Y. , Stash , N. , Aroyo , L. , Gorgels , P. , Rutledge , L. , & Schreiber , G. ( 2008 ). Recommendations based on semantically enriched museum collections . Web Semantics: Science, Services and Agents on the World Wide Web , 6 ( 4 ), 283 - 290 .