Skip to main content

Museums and the Web

An annual conference exploring the social, cultural, design, technological, economic, and organizational issues of culture, science and heritage on-line.

One-to-one: supporting artist-visitor dialogue

Silvia Filippini-Fantoni, Cogapp; Kirstie Beaven, Tate; and Ben Rubinstein, Cogapp, United Kingdom

http://aiweiwei.tate.org.uk/

Abstract

In this paper, we describe in detail the principles, philosophy and technology behind One-to-One with the artist. This project, developed by Tate in conjunction with Cogapp, consists of a series of booths which allow visitors to the Chinese artist Ai Weiwei’s Sunflower Seeds installation to record video questions and answers for the artist. These are then published online at www.tate.org.uk/go/aww, together with the artist’s responses. This paper will also report on the effective usage of the kiosks and website by both the visitors and the artist, and the challenges encountered in the development and implementation process, which was completed in less than eight weeks.

Keywords: User-generated content, videos, contemporary art, kiosk, artist-visitor dialogue

1. Introduction

Artists have been using the Internet since the early days of the Web, but no-one has done anything quite like Chinese artist Ai Weiwei who uses technology, Twitter in particular, as a way of sharing his artistic and political life with people all around the world.

1.1 The artist

Ai Weiwei is one of China’s most active and internationally-regarded Conceptual artists. He is probably best known outside the art world for his collaboration with architects Herzog and de Meuron on the “Bird’s Nest” Beijing National Stadium for the 2008 Olympics and Paralympics. Yet his status in China remains rather unstable. He has often remarked that he is a Chinese citizen with strong feelings for his country, and he does not criticise it without reason. However, his outspoken comments on Chinese society, his activism, protests and campaigns for human rights and freedom of expression in China have made him a controversial figure for the Chinese authorities, and perhaps contributed to his relatively scant recognition within the Chinese art system.

Ai Weiwei’s art and life are inextricably intertwined. His work encompasses huge sculptural installations, photography, performative interventions, hand-crafted art objects, and sharing his life on social networking sites with his Chinese and international followers. He has even gone as far as to call himself a “readymade”: referring to Marcel Duchamp’s notorious repurposing of found objects to become art objects.

Until it was shut down by the authorities in May 2009, Ai had a popular blog where he posted images of his life alongside social and political commentary. Now, his focus is more on Twitter, where he is a prolific tweeter. The 140-character limit allows for more in-depth conversations in Chinese than in European languages. Though the site is blocked in China, many circumnavigate the firewall to communicate with him and each other, and join Ai in his campaigns.

1.2 The Unilever series

Given the participatory nature of Ai Weiwei’s artistic practice and his understanding of architecture and scale, he is an ideal artist to conceive and create a work of art for the Unilever Series in Tate Modern's Turbine Hall.

The annual commission began with the opening of Tate Modern in 2000 with Louise Bourgeois’ extraordinary towers I do, I Undo, I Redo 2000; subsequent works have explored unexpected and innovative creative responses to this huge, half-public, half-gallery environment. Olafur Eliasson blurred the boundary between inside and outside space with his work The Weather Project 2004, while Doris Salcedo’s Shibboleth 2008 (the Crack, as it became popularly known) literally mined the museum, creating a fissure in the floor.

The thing that is crucial to (and unifies) all the work in the Unilever Series, however, is the public’s engagement with it. Works such as Carsten Holler’s Test Site, 2007 or Miroslaw Balka’s How It Is 2009, though entirely different in conception and intention, nonetheless require the public to bring the work and the space alive. The public’s expectation of the Turbine Hall has crept out of the traditional elitist gallery sphere and begun to populate the British popular imagination. Visitors to Tate Modern expect to be able to participate in the art, to join in, to become part of it. Commissioning an artist like Ai Weiwei for 2010’s programme fitted perfectly with this expectation.

1.3 The artwork: Sunflower Seeds

With his wide-ranging practice, Ai Weiwei’s Unilever Series commission could have taken many forms. For Sunflower Seeds 2010, he has carpeted the floor of the monumental space of the east end of the Turbine Hall with one hundred million individually crafted ceramic sunflower seeds.

Fig 1: Ai Weiwei’s Sunflower Seeds in the Turbine HallFig 1: Ai Weiwei’s Sunflower Seeds in the Turbine Hall

Each seed is a tiny individual piece of work, each apparently identical, but each actually unique. They were all individually sculpted and painted by ceramic specialists working in small-scale workshops in the Chinese city of Jingdezhen: China’s “Porcelain Capital”, where high quality porcelain has been produced for 1,700 years. Poured into the interior of the Turbine Hall’s vast industrial space, the seeds coalesce to form a landscape, until a close look focuses your attention on the individual seeds that go to make it up.

Sunflower Seeds invites us to look more closely at the ‘Made in China’ phenomenon, since far from being industrially produced, the seeds are the effort of hundreds of skilled hands. It also raises questions about the individual and society and plays with our sense of scale and perspective.

2. The project

2.1 The principles

As Ai Weiwei is an artist who regularly engages directly with his audience, we were keen to find a way to give visitors to the Turbine Hall a window into that part of his practice, alongside the physical work in the gallery. An online digital project is created for each Unilever Series commission and, just as each work in the gallery is unique, the online treatment for each artist and work has been bespoke.

Ai Weiwei’s show at the Haus der Kunst in Munchen was accompanied by a complementary blog; previous gallery talks and discussions with him had been streamed live online, and his tweets are translated into English by a team of translators, so the precedent was there for online engagement with a London (as well as global) audience. He was very supportive of finding a way to broaden his interaction with the visiting audience, and was committed to continuous discussion, throughout the run of the project, even knowing how time-consuming this might prove for him.

As it was already possible for a global and diverse audience to engage with Ai Weiwei online, we tried to find a way to engage the specific audience for this particular work by “catching” them right there in the gallery. We were excited about the possibility of opening up a direct channel of reciprocal communication between the gallery visitor and the artist. Again, Ai Weiwei was supportive and enthused by the project. This sort of project of course relies heavily on the interest, goodwill and input of the artist, and would not be possible without the artist’s full collaboration.

It was decided that a video system would facilitate a new way for visitors to interact with Ai Weiwei, rather than duplicating a channel of communication that already existed. We wanted visitors to be able to ask him a question about Sunflower Seeds in particular, or his practice in general. The system of recording in the gallery needed careful thought to encourage visitors to take part – asking an artist a question could potentially be a daunting activity. On the online side, we wanted it to be easy to browse the videos, so gallery visitors could find their own question and see if it had been answered, but also so that online visitors could watch videos submitted by others that were of particular interest and see the ones he had answered.

2.2 The implementation

Based on these principles and with the help of Cogapp, a Brighton-based digital agency specializing in working with museums, we conceived and implemented a communication system between the visitors and the artist, made up of two main components: a series of video recording kiosks at the museum, and a dedicated website where both visitors and non-visitors to the installation could view the recorded videos and the artist’s eventual responses. All this was delivered in two languages (English and Chinese) in an eight-week timeframe.

2.2.1 The kiosks

The eight video recording booths were installed in a space created under the Turbine Hall bridge, in proximity to the installation but not so close or prominent as to distract from the work itself. For the artist, in fact, it was very important that the visitor see the work of art first and then eventually, on the way out, stop to record a comment and watch a video about the making of the work (which was projected in another space nearby).

Fig 2: One-to-One with the artist kiosks in the Turbine HallFig 2: One-to-One with the artist kiosks in the Turbine Hall

The kiosks are each made up of a touch screen with a set of speakers and a camera with a built-in microphone. Two of the screens are positioned at a lower height to enable use by children and visitors in wheelchairs.

The timeout/attractor screen of the kiosks includes muted video clips of visitors’ messages and calls to action in both English and Chinese (see Fig. 3). Depending on whether the visitor touches the "Start" control, or the equivalent Chinese control, the remaining screens are presented entirely in that language. Having implicitly selected a language, the visitor is taken through a video welcome message by Ai Weiwei with basic instructions, followed by a selection screen where users are given the choice of either asking or answering a question about the work. If people opt to answer a question, they are given a list of five to choose from.

Fig 3: Kiosks’ timeout screenFig 3: Kiosks’ timeout screen

The idea of allowing visitors to either ask or answer a question was introduced in the early days of the development process because it strengthens the notions of a two-way communication between visitors and the artist, who is also curious to know what people think of his work. Also providing some questions is a way of facilitating the formulation of people’s thoughts regarding their feelings about the work, its meaning and the role of the artist in contemporary society.

Once the visitor has chosen whether to ask or answer a question, a live feed from the camera, positioned on top of the kiosks, appears on screen. The visitors can see themselves and adjust their position (or that of the camera) to make sure that their face is clearly visible.

Fig 4: Live feed from the cameraFig 4: Live feed from the camera

When ready, the visitor can click on the recording button, at which point the recording process starts. When finished, the visitor then clicks on the finish button available on the screen (see Fig. 4). At this point, if the visitor has opted to ask a question, they will be invited to tag their question by choosing from a list of five possible topics (see Fig. 5). The tagging is necessary to facilitate classification and retrieval of the video online. After careful consideration, we opted for a pre-defined list of tags, rather than allowing visitors to input the term using a virtual keypad, which would have been very difficult to implement in Chinese, as well as adding a layer of complexity to a process we preferred to keep as direct as possible.

Fig 5: Video tagging screenFig 5: Video tagging screen

The last three screens of the kiosk include a terms and conditions agreement, to which visitors must subscribe if they want Tate to post the video online (alternatively, the video will be discarded), a thank you message from Ai Weiwei, and a final screen summarizing the video details together with a message notifying them that it will be posted online within the next 72 hours.

2.2.2 The Website

Once created, the video goes onto a moderation queue, where Tate staff examine it and decide whether to discard it, publish it, or feature it as a recommended video and eventually send it to the artist to solicit a response.

If published, the video will appear on the One-to-One with the artist dedicated website (http://aiweiwei.tate.org.uk/), where people can access it in various ways.

Visitors who are looking for their own videos can do so by clicking on the “Find your video” option, where they will be asked to provide different information to help retrieve the video, such as the date and approximate time of creation, the language (Chinese or English), and whether they answered or asked a question, as well as the type of question or answer.

Based on the criteria provided, the system will display a set of matching video results, including a warning that videos can take up to 72 hours to be published and that some might have been removed for legal reasons.

Fig 6: HomepageFig 6: Homepage

Videos can also be explored by using the tag cloud available on all pages of the site. Here when a user selects one of the terms, a list of matching video results is provided in reverse chronological order, accompanied by a filtering system that allows them to be sorted by topic (tags), video type (answer or question), language (English or Chinese), or whether it has a response by the artist (yes/no).

From the home page, users can also access the most recently added and the most popular videos, as well as those to which Ai Weiwei has responded. A link to the dedicated twitter feed (#tateaww) for the project is also available to allow people in and outside the museum to join in the conversation.

Ai Weiwei’s responses are either in the form of a short text or a video, which he uploads on to the site using a slightly different version of the editor’s interface that is used by Tate staff for moderation. On the actual website the responses, when available (Ai Weiwei only answers selected videos), appear on the relevant video page, immediately after the video, where a link to the most asked question and relevant answers is also provided. On each video page, visitors can also choose to recommend or flag the video as inappropriate, share it with their friends and family on Twitter or Facebook, or even embed it in their own blogs or personal pages.

Links to similar videos as well as videos answered by Ai Weiwei are also available on this page to encourage users to further explore the content and see what other people have asked or said about the work.

Fig 7: Video pageFig 7: Video page

2.3 The technology

Given the limited time available for its implementation and the nature of the project, which was expected to be seen by an estimated two million mainly non-technology-savvy visitors, over a period of six months, the technical approach had to be rapid, scalable, robust and simple. As a result, Cogapp devised a system with four components, each of which can stand alone in the event of issues with any of the others (see Fig. 8).

Figure 8: Diagram of the system’s infrastructureFigure 8: Diagram of the system’s infrastructure

The kiosk was built in LiveCode for tight control and extremely rapid development. This was especially important as the short duration of the project meant we had to start building early, knowing that the specifications would likely evolve during the build phase.

The kiosk concentrates on capturing the visitors' comments using Logitech cameras with built in microphones, placed above the ELO touch screens. Each kiosk is a separate unit running on its own Mac mini; and it performs only basic compression (PhotoJPEG) on the fly during video recording, sufficient to substantially reduce the data to be written to disk and moved over the network, while minimising CPU load to ensure we don't drop frames or lose sync. All navigation and messages on the kiosk are in English and Simplified Chinese; video clips are tagged with the language of the interface in use when the video was recorded, as well as with the topic or question selected by the user.

The second component is the ‘local server’, which collects videos from all the kiosks, transcodes them to H264 for better compression, and uploads them. The local server is an Apple Xserve which is connected to all the kiosks by a private network, running an application which retrieves videos from the kiosks and manages them through a series of queues in which videos are transcoded to H264, uploaded to a content delivery network. Finally, details are sent via an API to the third component, the website. Each kiosk has sufficient disk space to run unattended over several days, in the event of network failure or problems upstream; but in practice, freshly recorded videos are swept off the kiosk drives several times an hour by the local server.

The application running on the local server is written in Python; it can manage multiple threads to maximise throughput, but can also throttle uploads; for example, to avoid flooding the connection to the Internet. We expect the local server to act as a buffer, if necessary catching up overnight or during quiet periods with video clips accumulated during peak times. It has sufficient storage to keep many days’ worth of recorded video in case of problems with the network, or upstream at the CDN or website.

The third component is the content delivery network. It was decided to opt for Brightcove with which Tate already has a contract. Although Brightcove is capable of video transcoding, the "local server" was used to carry out this function, leaving Brightcove to act purely as a content delivery network.

The fourth component, the website, is a customised installation of Drupal, with a front-end interface available both in English and Chinese. As far as the back end is concerned, a standard editing interface for Tate staff was created to support the video moderation process. A special interface was also configured for the artist to allow him to post video responses to visitor contributions directly from anywhere in the world.

2.4 The design

The kiosk and Web design use the same principles: elegant design and logical navigation to build a unique bridge between artist and audience. The dialogue is the focus of every aspect of the design. Video is the 'hero' here.

The kiosk allows simple, fluid panels to build across the screen to ensure an easy, quick guide to get to the video recording screen. In an attempt to move away from a static page-to-page experience with the kiosk, full screen transitions were used to ensure the user can glide forward or back at any point in the interactive, keeping everyone's experience quick and understandable but also beautiful at the same time.

Visually, the kiosk and website feel part of the exhibition identity. It was crucial that the kiosks fit with the signage and displays, especially the timeout screens, which are the first point of contact with the interactive. Installed sensitively in the hall, they act as a digital poster for the exhibition and a call to action for visitors.

The website had to allow users to easily see the latest and most popular videos, find their own, and watch the conversation grow over time. Whilst the site had to be beautifully designed, it also had to be simple, quick and clear. The ability to share and connect content via Facebook and Twitter was key, so these tools were clearly labelled and visible in the design.

2.5 The challenges

Like most technology-based projects, we have encountered a series of challenges during its implementation, some of which were known from the start, while others became evident only after the launch.

Certainly the biggest challenge for the project was the limited timeframe in which it was developed. This had an impact not only on the approach that was taken but also, and especially, on the software and hardware that were employed and the limited amount of testing that could be run on these before launch.

Another significant challenge came from the fact that two days after the show opened to the public, a health risk was identified with the dust created by people interacting with the ceramic seeds. Consequently, visitors were no longer permitted to walk on the work. Although this had no technical effect on the kiosks, it did considerably change the way people used the booths as a way to voice their unhappiness (see section Usage and evaluation for more detail).

2.5.1 Audio quality issues

We knew from the start that audio quality would be a challenge, given the nature of the space and the crowds expected. The choice of going for a camera with a built-in microphone was practical, as we wanted to avoid having to install too much equipment in the booths. This meant, however, that a high audio quality could not be guaranteed. We originally hoped that the physical installation could help us considerably, by adding isolation between the booths and audio insulation; in the event, however, the exhibition designers, having to juggle many constraints, implemented a much simpler area.

While initial tests in Cogapp’s office and in the Turbine Hall seemed to indicate that the audio quality was fit for purpose, immediately after the launch we realized that the background noise of the exhibition video playing in the space nearby and of the many visitors to the installation made it difficult to hear what people had to say. We looked into the possibility of retrofitting some kind of insulation between the booths, but doing so after launch proved to be quite complex and expensive. As a result, we decided to reduce the audio level of the video playing in the background, as well as of the welcome and thank you messages available on each kiosk. After appropriate testing, we also adjusted the microphone input levels. This helped to improve slightly the audio quality of the videos, which, although still not exceptional, is enough to distinguish what people are saying in the videos.

2.5.2 Camera dimming problems

While the problem with the audio quality was somewhat expected, issues with the camera came as a surprise. Immediately after the launch, we noticed that some of the cameras started to dim down in the early afternoon, making it impossible to distinguish people in the videos. Because of this problem we had to discard many videos at least until the problem was solved. Unfortunately, the problem was intermittent, and it took us over a month to get to the bottom of it. Cogapp tried without any success to replicate the phenomenon in their office in Brighton on a machine identically set; this made it very difficult to get a correct diagnosis.

Initially, we thought it might be caused by the fact that we had not used powered USB cables to connect the cameras and the Mac minis, which were installed a few meters away in an appropriate room, but after we replaced some of the cables, the problem still occurred. We swapped cameras, experimented with different settings, and even contacted the manufacturer to ask if they were aware of the issue, but without any success. At the end, almost a month after launch, we established that the issue was caused by the presence of a strong directional light opposite the camera. Being positioned in parallel, the camera of each kiosk would capture the light of the opposite booth, causing it to adjust its focus and dim. Adding an extra set of lights that illuminate the entire space seems to have solved the problem.

2.5.3 Moderation

The moderation process also proved to be quite challenging. To try to maintain the direct communication of visitor and artist videos, it was planned at first that all uploaded content would be post-moderated. We planned to have the videos published directly to the website and then allow users to flag the videos as inappropriate, in conjunction with Tate staff keeping an eye on the videos uploaded to weed out anything offensive. Anyone using the screens in the gallery had to agree to terms and conditions – giving us permission to publish their video in this way on Tate Online.

During the development project, however, in discussion with the legal team, it became a source of concern: first, that we might receive defamatory or offensive videos, and secondly, that under-16s could not give permission for us to publish their images. In the UK, only a parent or guardian can agree to the publishing of an image of a child. It was agreed that we would therefore have to move to a system of pre-moderation – manually checking every video before it was uploaded to the website.

For the first two weeks of the project, we watched every video uploaded to check the content and see how many children appeared without obvious visual consent of an adult. Content-wise, the public nature of the space tended to dissuade visitors from anything offensive, and the fact that they were (albeit only nominally) constrained to asking or answering a question left less room for defamatory comments. We agreed that for content, post-moderation with the ability to flag as inappropriate on the site was a good solution.

However, the age issue proved more problematic. As in many cases we could not ascertain whether permission had been given for a child to appear, we had to have a way to make sure we did not publish any images of those under 16. This meant that we had to switch permanently to pre-moderating all the videos. Any images of an unaccompanied minor were deleted from the system before publishing.

The needs of moderation put a lot of strain both on a system that was not designed with pre-moderation in mind and on staff time. On a busy weekend, up to 500 videos might be uploaded in a day. Without even watching all these videos, it could take about an hour each day just to get all the videos checked and published. With busy schedules, this un-anticipated time could not fit into anyone’s working day. In the end, we had to employ someone to make sure no images of minors were published and to watch a proportion of the rest, and send the most interesting questions to Ai Weiwei.

2.6 Usage and evaluation

Despite these initial issues, some of which were easily resolved while others required more time to be fully addressed, the project has enjoyed a certain degree of success with the public both online and onsite.

2.6.1 The kiosk

Over a period of 4 months (the exhibition opened October 11 th2011), more than 20,000 videos have been created in the kiosks, of which about 12,000 have made it online after moderation.

Fig 9: A visitor using the kioskFig 9: A visitor using the kiosk

Visitors seem to be interested in both asking and answering a question, with a clear preference for the former (66%). Given the nature of the work, the most popular questions evolve around the topics of production and process and meaning and inspiration (see Fig.10).

Fig 10: Type of questions answered by visitors between October 11 thand January 30 thFig 10: Type of questions answered by visitors between October 11 thand January 30 th

Frustrating for us, many of the most-asked questions were already answered actually in the gallery space. As previously mentioned, other questions reflected visitors’ disappointment at the fact that the artwork was closed to visitors walking on it early in the run of the show. Though visitors knew they could not walk on it, and knew why (the possible health risk of the dust made by the porcelain), they saw the video screens as a way to voice their unhappiness.

Other frequently-asked questions in English were about the importance of the sunflower seed – why had Ai Weiwei chosen to use this seed rather than any other? Sunflower seeds (especially in the un-processed form in which they are seen in the work) are rarely seen in the UK other than in pet food or health food shops. Chinese speakers, however, seemed to have a more direct response to the ubiquity of the sunflower seed in China – thus the question “why sunflower seeds?” was very rarely asked by Chinese speakers. Chinese speakers (although proportionally a very small number) also broadly tended to ask more questions about society and the artist’s and the work’s relationship with society, while English speakers were very focused on the process: how and why the work was made and conceived.

Many visitors found, once they had asked their question, that it was answered by watching the documentary film that was playing in the space nearby. It was not possible to design the space differently, but it is possible that the questions (or even the number of uploads) would have been different if visitors had all seen the film before being asked to pose a question to the artist.

While less popular than questions, answers were also recorded (44%), focusing mainly on the feelings inspired by the work, as well as its meaning (see Fig. 11).

Fig 11: Topic of visitors’ questions recorded between October 11 thand January 30 thFig 11: Topic of visitors’ questions recorded between October 11 thand January 30 th

2.6.2 The website

While the kiosks have been successful in capturing visitors’ interest and curiosity on site, the success of the website is even more evident.

In the period between October 2010 and the end of January 2011, the website received over 100,000 visits by 50,000 unique visitors, with 3.05 average page views and an average time spent on the site of about two and a half minutes.

Online visitors are mainly from the UK and Europe, with a limited percentage also from the US and China. The majority are repeat visitors (53% against 47% of new visitors), suggesting that people come back to the site after an initial contact. Whether this is due to the fact that they did not manage to find their video the first time around, or whether they wanted to see more, is not known at present and will require further investigation.

The homepage, find your video, search results, most asked questions, videos answered by Ai Weiwei and most recent videos are the most popular sections, suggesting that people come to the website primarily to look for their videos and eventual answers by the artist.

People seem to use both the flag and recommend functionalities available on the video page. It is not a coincidence that the most watched videos are those that have been recommended by other visitors.

To date Ai Weiwei has answered 250 videos, out of the roughly 12,000 that have been published. They range from in-depth answers about Chinese society to simple answers about his favourite colour or food. Since Weiwei can spend up to eight hours a day interacting with his audience, though the task is onerous, it fits within his ideal to interact freely with people.

Even though no research has been conducted yet on visitor satisfaction, the indications at this stage of the project are that visitors enjoy the opportunity to directly engage with an artist, relatively unmediated by the institution.

Within the organisation, the project has gained a lot of interest. It is the first large-scale digital interactive project we have done that has been a cross-departmental collaboration with an artist, with an in-gallery input and an online output.

3. Conclusions

One-to-one with the artist is one of the first successful attempts by a cultural organization to support a true and engaging dialogue between a contemporary artist and gallery visitors using online, video and social media technologies.

Attempts have been made before by museums to try to facilitate such communication on their websites (e.g. artist-in-residence blogs); however, this is the first time that this kind of initiative has met with such a positive response both from the artist and the audience, evidenced by both the high number of videos created and the high volume of online traffic.

The success of the project can be attributed to a number of factors:

  • the visitor expects to have an interactive experience in the Turbine Hall – it is not perceived as an intimidating or elite space. The simplicity and immediacy of the system for recording the videos removes many of the barriers to engagement seen before, while the very nature of the installation lends itself to raising questions and comments regarding the meaning and process of the work and the role of the artist in contemporary society.
  • the artist’s continuing interaction is likely to increase uptake – knowing that he does answer questions is very motivating, whilst also showing that the website is active and changing.
  • Other factors that have contributed to increase interest and participation are the recent events surrounding not only the artist’s personal life, but also the artwork itself, which gained a lot of press in the opening days (and was continued by the ongoing discussion of the health and safety concerns).

By supporting the dialogue between the artist and the visitors, One-to-One with the artist represents an important step in the ongoing process of turning museums from “temples” into “fora”, where points of views other than (though including) the traditional curator and artist voice are valued, published and shared with the wider community, both in the gallery and online.

Cite as:

Filippini-Fantoni, S., et al., One-to-one: supporting artist-visitor dialogue. In J. Trant and D. Bearman (eds). Museums and the Web 2011: Proceedings . Toronto: Archives & Museum Informatics. Published March 31, 2011. Consulted http://conference.archimuse.com/mw2011/papers/one_to_one_supporting_artist_visitor_dialogue