Early Detection of Pairs of Visitors by Using a Museum Triage

Tsvi Kuflik, Israel, Eyal Dim, Israel


We report on a visitors' study aimed at automatic detection of pair-behavior in the early stages of the museum visit. For that, we used sensors located in the "start of visit area" we called “triage.” In addition, behavior patterns were detected also by sensors located throughout the museum, since the initial pair-behavior patterns detected in the triage may change over time.

Keywords: Social signal processing, Visitor's study, Social synchronization, Pair behavior patterns

1. Introduction

It is well accepted in the user modeling community that a person has different needs in different contexts (Byun & Chevest, 2001). In the case of a museum visit, visitors have a personal identity (Falk, 2009) that is set by the visitor him/herself, the visit context, and the visited museum. In addition, providing personalized services to visitors requires that their personal and contextual aspects be identified (Zhu et al., 2006). An appropriately equipped museum may allow a system to monitor visitors’ behavior, reason about them, and detect such contexts. This, in turn, may allow it to provide a valuable personalized service throughout the visit (Kuflik et al., 2011), for example, by suggesting interesting exhibits to a bored visitor, or not disturbing a visitor immersed in the exploration of exhibits.

Providing personalized services in a museum poses two challenges: first, a User Model (UM) for a new user has to be bootstrapped, data about the visitor need to be learned and an initial UM built (alternatively, assuming that a UM already exists, it can be imported at the onset of the visit, as suggested by Kuflik and Poteriaykina (2009)); second, the UM has to be rapidly adapted to the current context. Thus, it seems that in order to provide meaningful service at the onset of the visit, special attention should be paid to the “start of visit” area in the museum. This area, where the visitor first encounters the museum, may be used as an area for intensive data collection about the user. We called this concept the triage, following that of the hospital triage, where physicians decide on the initial course of treatment.

This paper presents our work on exploring the idea of the museum triage. We gathered initial information about the visitors in an area populated with sensors, and tried to assess how consistent the visitors’ behavior was throughout the rest of the visit. We focused on identification of the social context, which plays an important role in the behavior of museum visitors (Packer & Ballantyne, 2005), by detecting and analyzing the behavior of pairs of visitors. Pair-behavior in the museum may vary from one pair to another (the word pair throughout this paper refers to dyads, and not necessarily to couples). Identification of such a behavior pattern depends on classifying behavior into types. A method for detecting pair-behavior types may allow future research to offer personalized services to pair members. This is a first step toward the development of a pairs’ behavior typology and behavior detection methodology based on instrumentation at the museum.

2. Related work

Social context in museums

The social context has a major impact on a museum visit. A visitor may be a single visitor, part of a small group (two to eight persons), or part of a large group of visitors. Falk (2009) claims that “only 60% of the visitor’s attention is spent on looking at the exhibits, while the other 40% of the visitor’s attention is directed elsewhere; mostly on conversation with other members of her/his social group or general observation of the setting.” Bitgood (2002) posits that “the overwhelming percentage of visitors come in groups—usually with families or friends unless they are part of a school group.” He sees the social contact as a significant aspect of informal learning settings, and sometimes as the most important part of the experience. McManus (1991) describes the social context of museum visitors as it pertains to small groups of children, couples, and adults. McManus also describes the deeper involvement generated by the social intimacy of visitors’ groups that exhibited good social relationships, moved around the museum in closed units, and were more likely to engage in behaviors that would improve their visit experience.

The contribution of groups to the enhancement of learning and the visit experience is emphasized by Falk and Dierking (2000) and supported by Leinhardt and Knutson (2004) in their work about museum conversations. Packer and Ballantyne (2005), who analyzed pairs’ learning, posit that learning through a group visit to museums is different from individual solitary learning. Group behavior of museum visitors was also analyzed by Dim and Kuflik (2009), showing that some group members tend to separate during a museum visit, while others may stay together but not interact. Finally there are groups that stay together throughout most of the visit and interact.

Hence, the social context seems to be an extremely important aspect of the visit and a challenge for technological detection and monitoring. Visitors display a variety of behaviors in the museum that may be a cue for understanding them and their needs. Examples of such behaviors may be found also in the work of Veron and Levasseur (1983), Dean (1994), and Hooper-Greenhill (1994). Finally, Falk (2009) suggests the “Identity-related Museum Visit Experience Model” that contains a typology of five visitor-identity prototypes: (i) “Explorer,” whose visit to the museum is motivated by curiosity or general interest in discovering more about the subject matter introduced by the museum; (ii) “Experience Seeker,” often a tourist, whose visit is typically motivated by the main attraction the museum is known to offer; (iii) “Professional/Hobbyist,” who is interested in specific topics out of the full collection of the museum visit; (iv) “Recharger,” who comes to the museum to reflect, rejuvenate, or relax and absorb the atmosphere; and finally, (v) “Facilitator,” who visits the museum to satisfy the needs and desires of someone she or he cares about rather than just her/himself. The identity of a “Facilitator” is actually a social identity that is manifested only within groups.

User modeling in museums

According to van Velsen et al. (2008), “an adaptive system tailors its output, using implicit inferences based on interaction with the user.” Adaptive systems use on-site dynamic updates of the UM for each person (Fantoni, 2002; Byun & Chevest, 2001). The data about the user are gathered in the UM and may include knowledge, interests, goals, tasks, background, context, and other data that represent the static characteristics and dynamic changes relevant to the user in the model (Brusilovsky & Millan, 2007). This information may be used by an intelligent system to support the user in a personalized manner, changing the behavior of the system according to data gathered about the user. User modeling and its application for computerized support of individuals in a variety of tasks have been researched for over 30 years (Kobsa & Wahlster, 1990).

Until recently, user modeling in general, and in cultural heritage environments in particular, has focused primarily on modeling individuals. Previous studies have focused primarily on exploring the feasibility of using novel technologies to support individuals visiting the museum, mainly by improving the methods of information delivery, as surveyed by Ardissono et al. (2012). Although the group model concept has not yet matured, several aspects of group-related studies, such as recommendation systems that recommend what may suit most of the small group members, are based on a variety of strategies (Jameson & Smyth, 2007; Kay & Niu, 2005). Some applications relate to cultural heritage, including: tabletop displays (Stock et al., 2009); Sotto Voce (Aoki et al., 2002; Szymanski et al., 2008); Traces (Stevens & Toro-Martell, 2003); ARCHIE (Luyten et al., 2006); PEACH (Stock et al., 2005); PIL (Kuflik et al., 2011); and AgentSalon (Sumi & Mase, 2004). These applications were aimed at providing collaborative tools, such as messaging, voice communication, leaving messages or video shots for future visitors, and eavesdropping during the visit, to allow and orient interaction and enhance sharing of information among visitors with the objective of supporting groups, not only individuals; however, these were specific prototypes that explored specific collaboration aspects.

3. Social behavior pattern

Following Veron and Levasseur (1983), who referred to individual visitor behaviors by identifying them with animals, we gave animal nicknames to selected pair-behavior types. These types represent the level of social synchronization between the pair members, as well as whether they pay attention to the museum exhibits. Figure 1 presents the six behavior types that we selected for the study, based on the analysis of video recordings: penguins, geese, meerkats, parrots, doves, and lone-wolves.

Figure 1: Selected visitor's behavior-typesFigure 1: Visitor’s  pair-behavior types

Penguins represent a pair of visitors who come together to the museum. They walk through the triage area, but do not pay attention to exhibits on either side. Their time in the triage area is short and they proceed to other parts of the museum. Geese represent a pair of visitors who advance together, but one visitor seems to lead, signaling to the other visitor that it is time to proceed. Meerkats represent a pair of visitors advancing from one exhibit to the other, together and side by side, paying a lot of attention to the exhibits in a synchronized manner. Parrots represent a pair of visitors advancing from one exhibit to the other, together, and turning halfway toward the other and halfway toward the exhibits, interacting while looking at the exhibits. Doves represent a pair of visitors standing face to face involved in conversation, ignoring exhibits. Finally, lone-wolves represent a pair of visitors that enter the museum together and then split. The split may be manifested as the pair members walking apart, or standing with their back toward each other.

Pairs seem to use a mix of behavior types, although there may be one or more dominant types. Such behaviors may be identified using measurements taken by simple sensors.  Such sensors and logic, installed in a museum, would allow an automatic system to sense and analyze visitors’ behavior. This analysis may allow the visitors’ experience to be enhanced by offering new personalized (or rather “groupalized”) services.

4. Exploring the idea of the museum triage

We focused on the identification of behavior patterns of pairs visiting the museum within the triage area and throughout the rest of the visit. We were interested in learning how simple sensors can help detect the behavior of a pair of visitors, what the contribution of the triage is, and whether the behavior is persistent along the visit. Due to space limitations, this paper focuses on the detection of the first two pair-behavior types, penguins and geese, using sensors in the triage area and throughout the visit.

5. Method


Thirty-five pairs of regular museum visitors took part in the study. They were recruited at the entrance to the museum. Each pair was manually identified and registered as a “group coming together.” They filled a demographic questionnaire as individuals (Table 1) and as a pair (Table 2).

Visitors (n=70)
Age group
Under 30 (min. age=14) 33%
30-59 31%
60 and over 36%
Male 36%
Female 64%
Years of education
Over 16 46%
15-16 27%
12-14 24%
Under 12 3%
Previous visit to the Hecht Museum
First visit 60%
Repeat visit 40%

Table 1: Individuals’ demographic data


Pairs (n=35)
Pair gender
Mixed female-male 54%
Female-female 37%
Male-male 9%
Age difference (years)
5 and less 63%
6 – 20 11%
Over 20 26%

Table 2: Pairs’ demographic data

The target exhibits

Research was conducted within the Hecht Museum (http://mushecht.haifa.ac.il/), an established archaeological and art museum located on the University of Haifa campus. The triage area was located in the first exhibition area of the museum through which visitors have to pass on their way to other parts of the museum. Figure 2 presents the museum layout and an image of the triage area (marked in pink on the map). The museum entrance leads to the triage, which is in the form of a corridor about 14 meters long and 4 meters wide, constituting about 3 percent of the total area of the museum’s archeology exhibition. The picture at the bottom left of figure 2 shows the triage area as seen from the museum entrance. Exhibits are along both sides of the triage. The corridor shape of the Hecht Museum triage area allows additional information to be collected. It has four main directions that allow visitors’ interest in exhibits on either side of the corridor to be distinguished, unless they just walk through the corridor.

Figure 2: The Hecht museum map and the triage image Figure 2: The museum map and the triage image

The instrumentation in the museum

As part of the PIL project (Kuflik et al., 2011), the museum is equipped with a wireless sensor network comprising a radio frequency (RF)-based indoor positioning system. It has three different types of component: small (matchbox size) mobile wearable RF tags called Blinds (figure 3a); stationary RF tags called Beacons (figure 3b); and RF to TCP Gateways, which relay signals from the Blinds and Beacons to a central computer for data collection and analysis. Beacons are statically located near selected locations of interest in the museum, having limited area coverage for proximity sensing (figure 3c).

Figure 3: Positioning system equipment and usage scenarioFigure 3: Positioning system equipment and usage scenario

A visitor receives a Blind at the museum entrance and wears it throughout the visit. Each Blind transmits a unique identification (ID) that allows the visitor wearing it to be identified. While walking through the museum, when the visitor approaches within a predefined distance of a Beacon, the Blind detects a signal transmitted by the Beacon containing the Beacon’s ID and reports it through the gateway and the communication network to the PIL server. At this point, the location of the visitor is determined, recorded, and may be used by services in real time, or saved for future analysis. In addition to the detection of a Blind’s proximity to a Beacon, the Blind measures and reports the following time-tagged signals: proximity to other Blinds (detecting proximity among visitors), compass data (measuring visitor’s orientation), and acceleration data through accelerometers. In general, the distance between adjacent Beacons is about 3 to 4 meters.

While providing a reasonable indoor positioning solution, the system has two major weaknesses. First, it knows only when a person wearing a Blind is in the proximity of a Beacon. Thus, it does not detect positioning in transition from one Beacon to another. Second, the specified detection range in the proximity of each Beacon or Blind is 2 meters. Thus, when we detect a user in the proximity of a Beacon, we can know only that she or he is within 2 meters of that Beacon, without knowing the exact position. All data are time-tagged, allowing time-related information, such as the visitor’s velocity, and time of arrival, and the holding power of an exhibit, to be gathered.


Data-collection sessions were conducted from December 2010 through December 2012, during regular museum opening hours. A research assistant located at the entrance to the museum asked visitor pairs, entering the museum together, to participate in the study. Those visitors who agreed to participate were asked to wear a Blind sensor during their visit. The research assistants neither instructed nor limited the visitors. The visitors were not given any visitors’ guide or any other device, in order not to distract their attention from the exhibits and their social companions. Data were collected using the instrumentation throughout the museum, including the triage area, during the entire visit.  A video camera was located within the triage area and was used to verify that the instrumentation worked properly and the data were reliable, as well as to observe pairs’ behavior. We did not use a post-visit questionnaire or review, and relied on the instrumentation’s data collection. Pairs who exited the museum before a cut-off time of 10 minutes were excluded from the sample, as were those who encountered an instrumentation malfunction.

6. Results and discussion

This study presents a first step toward automatic detection of social behavior in a museum. By exploring the possibility to reason about pair behavior, we identified several behavior types, presented above (i.e., penguins, geese,  meerkats, parrots, doves, and lone-wolves). Pairs of visitors were monitored by time-tagged reports that included proximity to Beacons, Blind IDs, and the orientation of each visitor, in addition to manual identification of members as being part of the same group, and the structure of the museum. In addition, we addressed the question of data persistency during the visit by considering the following three questions: Is there a dominant behavior? Can this dominant behavior be detected in the triage? Does it change during the museum visit? As already mentioned, due to space limitations, the results and discussion focus on two of the six behavior types: penguins and geese.

Penguins detection

Penguin behavior is exemplified by a group of two visitors entering the museum, walking in a relaxed manner through the triage area, and heading forward while ignoring most of the exhibits on either side. They proceed through the museum to a point of interest for them (e.g., in one case, we identified that the pair went directly to the art section of the Hecht Museum on the second floor, ignoring the archaeological exhibits on their way).

We used two measures to identify penguins in the triage area: walking velocity (walking faster through the triage may be indicative of penguin behavior) and orientation within the triage area. Of the 35 pairs, three exhibited clear penguin behavior, as detected by the video. The data were collected by identifying the time difference between the entry into and exit from the triage area, in addition to the prior knowledge of the triage’s corridor shape and the length of the triage area. In all three cases, the pair went through the triage faster than 0.25 meters/second (a velocity of 0.25 meters/second may be considered a relaxed walking pace). All other pairs walked more slowly through the triage area, because they stayed longer at the exhibits on either side of the triage. Penguin detection should be further investigated in the future, because three pairs constitute a very small sample for defining penguins’ velocity threshold.

Another parameter that may contribute to the identification of penguins is whether pair members are oriented toward exhibits on either side of the corridor or toward the walking path. In order to measure also orientation halfway toward exhibits (or pathways), we divided the orientation spectrum into five main categories: (i) heading forward in the corridor; (ii) heading halfway forward and halfway to exhibits; (iii) heading toward exhibits; (iv) heading halfway backward and halfway to exhibits; and (v) heading backward. In order to measure the mutual nature of their orientation, we counted the number of simultaneous occurrences of orientation types (i) through (v)  for both pair members (e.g., if both members are heading forward or halfway forward, there is a greater chance that they have penguin behavior).

Figure 4 presents two matrices, representing two pairs, each having a different dominant behavior of mutual orientation with regard to the walking path and exhibits in the triage area. The color code presented on the right of each matrix represents the percentage of messages reported for each type out of all messages transmitted in the triage area. Pair A seems to be heading forward most of the time, which indicates penguin behavior. Pair B is heading toward the exhibits most of the time, which may indicate non-penguin behavior. This measure may be expanded in future research to other exhibit halls, depending on their structure, to find whether the results of such an analysis are persistent throughout the whole visit.

Figure 4Figure 4: Mutual orientation in regards to exhibits and walking path in the triage area

Possible explanations for penguin behavior may be that the pair has a mutual social interest and the museum is only an embellishment; or, following Falk (2009), that it constitutes a pair of professionals, hobbyists, or attraction seekers going directly to a specific place of interest within the museum.

Geese detection

What is special about geese is that they follow each other: one leads and the other follows. Reports of time-tagged proximity to a Beacon allow recording of the time of arrival (TOA) at a position that the Beacon covers. Comparison of the TOA of both pair members at Beacon positions may reveal whether there is a major TOA difference between one member and the other (if yes, this may increase the chances that the pair is a pair of lone-wolves).

It is also interesting to detect whether a pair member is usually the first (dominant geese behavior), or whether the pair members alternate in taking the leading role. A summary of all 35 pairs is shown in figure 5. The X axis is the average time difference (in seconds) between both pair members’ TOA at positions covered by Beacons. The Y axis shows the percentage of cases where one visitor arrived at a position prior to the other out of the total visited positions that were covered by Beacons. Each blue diamond represents the value of one of the pairs. A high value on the Y axis represents a higher probability that one visitor is ahead of the other (geese behavior), whereas a low value  represents a higher probability that the two visitors alternate in taking the leading position. As for the X axis, the shorter the TOA difference between the two visitors, the greater the synchronization between them (i.e., they move together from one place to another). A high value on the X axis means that the two pair members are less coordinated (lone-wolves behavior).

Figure 5: Arrival order and average time difference – arrival at Beacon positionsFigure 5: Arrival order and average time difference – Arrival at Beacon positions

This data collection may start in the triage area (three Beacons) and can be accumulated during the visit. The data collected may show that sometimes information from the triage is not persistent during the rest of the visit, and does not predict the behavior throughout the visit. This means that the assessment should continue throughout the visit to identify whether there is a significant “first person,” and whether there is a change in this behavior during the rest of the visit.

Research limitations

The data were collected at the Hecht Museum and were not validated in other venues. The museum is located on a university campus and visited mainly by highly educated adults. The data collected are limited by the granularity, accuracy, and selected positioning of the sensors. While the sensor data are objective, we did not collect any subjective or introspective data through post-visit interviews. We focused on using simple measures, such as time and proximity. We did not use other complex measures, such as body gestures, eye gaze, and conversation analysis. As for the triage area, its special structure, a corridor, contributed additional data, but a different shape of the triage or a different setup of the exhibits within the triage might have affected the results.

7. Conclusions and future work

Pairs may exhibit multiple behavior patterns in a museum, and the proposed pair-behavior types classify some pair behaviors. Future work may validate the selection of pair-behavior types or update this selection. However, an instrumented museum that is capable of detecting visitors’ proximity to positions in the museum, general orientation, orientation with regard to the pathways, velocity, and order of arrival at selected positions in the museum may contribute to a better understanding of visitor behaviors, leading in the future to better adaptation to visitors’ needs. As for the question “What is the contribution of the triage?” we have seen that the triage does contribute to the identification of penguins, while geese behavior may change after the triage; hence, monitoring should continue. The advantage of the triage is that it is used when the visitors have just entered the museum, and the conclusions may lead to better adaptation to the pair members’ needs.

Once pairs’ typology is further validated, future study may reveal how museums could improve a visit-in-pair experience. For example, how to use the detection of the dominant role; or the significance of detecting that in a given pair the dominant role is switched from one visitor to the other along the visit; or what to suggest to the pair members in the case that they are not synchronized in their visit trajectory. It may also try to identify the relation between visitor identities and pair-behavior type (e.g., what happens if Falk’s (2009) “Explorer” comes to the museum with an “Attraction Seeker?”). In addition, the research can be expanded to the behavior of small groups of sizes greater than two. Last but not least, future research may suggest techniques for adapting and personalizing services to fit the pair-behavior type. Examples are avoiding interruption during social interaction, or suggesting a shared activity in order to regain social synchronization if a pair has separated during the visit. Detection of penguin behavior may lead for example, to identification of a “passing through” behavior (i.e., going fast through an exhibit area). In this case, future research may assess whether an adaptable smart museum guide would contribute more by refraining from suggesting presentations to the pair, allowing the pair members to proceed to a point of interest for them, or maybe  attracting their attention and increasing their interest in the current exhibit area.


The work was supported by the Israeli Science Foundation (ISF) grant 226/2010 and by the collaboration project between the Caesarea-Rothschild Institute at the University of Haifa and FBK-irst and FIRB project RBIN045PXH.


Aoki, P.M., R. E. Grinter, A. Hurst, M. H. Szymanski, J. D. Thornton, & A. Woodruff. (2002). “Sotto voce: Exploring the interplay of conversation and mobile audio spaces.” In CHI 02, Proceedings of SIGCHI Conference on Human Factors, in Computing Systems, New York, NY: ACM Press, 431–438.

Ardissono, L., T. Kuflik, & D. Petrelli. ( 2012). “Personalization in cultural heritage: The road travelled and the one ahead.” In User Modeling and User Adapted Interaction 22 (1–2). 73–99.

Bitgood, S. (2002). “Environmental psychology in museums, zoos, and other exhibition centers.” In R. Bechtel & A. Churchman (Eds.), Handbook of Environmental Psychology, John Wiley and Sons, 461–480.

Brusilovsky, P., & E. Millán. (2007). “User models for adaptive hypermedia and adaptive educational systems.” In The Adaptive Web, LNCS 4321, Berlin, Heidelberg, New York: Springer, 5–53.

Byun, H. E., & K. Cheverst. (2001). “Exploiting user models and context-awareness to support personal daily activities.” In UM2001 Workshop on User Modeling for Context-Aware Applications, Sonthofen.

Dean, D. (1994). Museum exhibition: Theory and practice. London: Routledge.

Dim, E., & T. Kuflik. (2009). “Group situational awareness: Being together.” In Proceedings of Workshop on Personalization, International Conference on User Modeling and Adaptive Hyper-media, UMAP ’09.

Falk, J. H. (2009). Identity and the museum visitor experience. Walnut Creek, CA: Left Coast Press Inc.

Falk, J. H., & L. Dierking. (2000). Learning from museums: Visitor experiences and the making of meaning. Walnut Creek, CA: AltaMira Press.

Fantoni, S. F. (2002). “Visiting with a “personal” touch: A guide to personalization.” In Museums. Maastricht McLuhan Institute – International Institute of Infonomics.

Hooper-Greenhill, E. (1994). Museums and their visitors. London, UK, and New York, NY: Routledge.

Jameson, A. & B. Smyth. (2007). “Recommendation to groups.” In The Adaptive Web, LNCS 4321, Berlin, Heidelberg, New York: Springer, 596–627.

Kay, J., & W. Niu. (2005). “Adapting information delivery to groups of people.” In Proceedings of the First International Workshop on New Technologies for Personalized Information Access at the Tenth International Conference on User Modeling, Edinburgh.

Kobsa, A., & W. Wahlster. (1990). User models in dialog systems. New York: Springer Verlag.

Kuflik, T., & K. Poteriaykina. (2009). “User model on a key.” In Proceedings of Hypertext 2009, Torino, Italy (June 29–July 1, 2009), 371–372.

Kuflik, T., O. Stock, M. Zancanaro, A. Gorfinkel, S. Jbara, S. Kats, J. Sheidin, & N. Kashtan. (2011). “A visitor’s guide in an ‘active museum’: Presentations, communications, and reflection.” In Journal on Computing and Cultural Heritage 3(3), ACM, 1–25.

Leinhardt, G., & K. Knutson. (2004). Listening in on museum conversations. Walnut Creek, CA: AltaMira Press.

Luyten, K., H. van Loon, D. Teunkëns, K. Gabriëls, K. Coninx, & E. Manshoven. (2006). “ARCHIE: Disclosing a museum by a socially-aware mobile guide.” In M. Ioannides, D. Arnold, F. Niccolucci, & K. Mania (Eds.), The 7th International Symposium on Virtual Reality, Archaeology and Cultural Heritage, VAST.

McManus, P. (1991). “Towards understanding the needs of museum visitors.,” In B. Lord & G. D. Lord (Eds.), Manual of Museum Planning, London: HMSO, 35–51.

Packer, J., & R. Ballantyne. (2005). “Solitary vs. shared: Exploring the social dimension of museum learning.” In Curator 48, 177–192.

Stevens, R., & S. Toro-Martell. (2003). “Leaving a trace: Supporting museum visitor interaction and interpretation with digital media annotation systems.” In Journal of Museum Education.

Stock, O., C. Rocchi, M. Zancanaro, & T. Kuflik. (2005). “Discussing groups in a mobile technology environment.” Proceedings of the Second Multiusers Intelligent Interactive Interfaces Workshop, IUI-2005, San Diego, CA.

Stock, O., M. Zancanaro, F. Pianesi, D. Tomasini, & C. Rocchi. (2009). “Formative evaluation of a tabletop display meant to orient casual conversation.” In Journal of Knowledge, Technology and Policy 22(1), 17–23.

Sumi, Y., & K. Mase. (2004). “Interface agents that facilitate knowledge interaction between community members.” In Springer Verlag (Cognitive Technologies Series), 405–427.

Szymanski, M. H., P. M. Aoki, R. E. Grinter, A. Hurst, J. D. Thornton, & A. Woodruff. (2008). “Sotto voce: Facilitating social learning in a historic house.” Computer Supported Cooperative Work 17, 5–34.

Van Velsen, L., T. van Der Geest, R. Klaassen, & M. Steehouder. (2008). “User-centered evaluation of adaptive and adaptable systems: A literature review.” In The Knowledge Engineering Review 23(3), Cambridge University Press, 261–281.

Veron, E., & M. Levasseur. (1983). Ethographie de l’exposition. Paris: Bibliotheque publique d’Information, Centre Georges Pompidou, 91–92.

Zhu, W., C. B. Owen, H. Li, & J. H. Lee. (2006). “Design of the PromoPad: An automated augmented reality shopping assistant.” In the 12th Americas Conference on Information Systems, Acapulco, Mexico.



Cite as:
T. Kuflik and E. Dim, Early Detection of Pairs of Visitors by Using a Museum Triage. In Museums and the Web 2013, N. Proctor & R. Cherry (eds). Silver Spring, MD: Museums and the Web. Published January 28, 2013. Consulted .

Leave a Reply