In Japan, and especially in the urban areas, various kinds of events, such as musical and dramatic entertainment, are frequently held. Overcrowding around venue areas is always a serious issue when such events are held. Because many of these events sell tickets in pairs, it is common for two people to attend the event together, and it is necessary for those attending to meet up and exchange tickets before the event. Additionally, there are cases where several pairs gather and make up a medium-sized group, and such groups also try to gather around the event venue before and after the event. It takes a great amount of time for such groups to meet up before and after the event in such a crowded area, and as they do not move right away from the area after meeting up, the congestion around the event venue can last for a long time. While large-scale events may have exit control, this is not effective due to the overcrowding caused by the abovementioned groups trying to meet up.
Therefore, it is necessary to reduce the congestion and control the movements of those attending events in order to do so. By guiding them to a designated location away from the event venue, the congestion of meeting places for groups may be reduced. However, such meeting places must be easily accessible for each individual within the group. Additionally, as a condition for deciding on a meeting place for groups, there must be an adequate number of facilities in the area. Based on these conditions, limiting meeting places to a station (excluding the nearest station to the event venue) on the train line on which the nearest station to the event venue is located, the present study aims to design, develop, operate, and evaluate a recommendation system for meeting places targeting groups of two or more people. The system will focus on accessibility, which will be defined as the accessibility between two arbitrary stations accurately reflecting the actual usage situation.
2. Related Work
The present study is related to three study fields, including (1) studies related to travel support systems and methods, (2) studies related to recommendation systems and methods of locations, and (3) studies related to the accessibility between two points. In the first category of studies—studies related to travel support systems and methods—Ishizuka et al. (2007) [1
] proposed a similarity search method for the movement locus of tourists from the location information, as well as the related text information. Kurashima et al. (2011) [2
] proposed a travel route recommendation method using the geotags of photo-sharing sites. Kurata (2012) [3
] developed an automatic sightseeing route system using geographic information systems (Web-GIS) and genetic algorithms (GAs). Sasaki et al. (2013) [4
] gathered information concerning regional resources and developed a system that offers travel support for each user. Fujitsuka et al. (2014) [5
] used the pattern-mining method, which lists and extracts the chronological movement of those visiting sightseeing spots to develop an outing plan recommendation system. Ueda et al. (2015) [6
] generated posterior information from the movement of users while sightseeing and developed a sightseeing support system, which shares such information as prior information with other users. In order to support sightseeing activities under normal circumstances, as well as an evacuation during a disaster, Fujita et al. (2016) [7
] developed a navigation system using augmented reality (AR), Web-GIS, and social media. Mizushima et al. (2016) [8
] proposed a sightseeing tour recommendation system using data related to traveler demands (location, time, and purpose). Just targeting foreigners visiting Japan, Yamamoto (2018) [9
] developed a sightseeing navigation system using the two-dimensional and three-dimensional digital maps of Web-GIS with Fujita et al. (2016) [7
] as a reference.
The second category of studies—studies related to recommendation systems and methods of locations—are important for the studies of location-based social networks (LBSN). Canneyt et al. (2011) [10
] proposed a recommendation system for sightseeing spots. Batet el al. (2012) [11
] developed a recommendation system for sightseeing spots using a multi-agent system. Uehara et al. (2012) [12
] extracted sightseeing information from the web, calculated the similarities between sightseeing spots using several feature vectors, and developed a recommendation system for sightseeing spots. Shaw et al. (2012) [13
] took into consideration the location information and travel records of users and developed a system that presents users with a list of sightseeing spots located in the vicinity. Ikeda et al. (2014) [14
] integrated Web-GIS, social networking services (SNSs), and a recommendation system, accumulated sightseeing spot information, and developed a social recommendation GIS to recommend sightseeing spot information according to the preferences of each user. Okuzono et al. (2015) [15
] considered the preferences of several people using photos and proposed a recommendation system for sightseeing spots. Zhou et al. (2016) [16
] developed a sightseeing spot recommendation system using AR, Web-GIS, and SNSs.
Additionally, Mizutani et al. (2016) [17
] referred to Ikeda et al. (2014) [14
] and Okuzono et al. (2015) [15
] and developed a social recommendation GIS that considered the change in situation of several users. Xu et al. (2017) [18
] proposed a location recommendation system using the GPS locus of users, while Abe et al. (2017) [19
] developed a tourism information system with language-barrier-free interfaces, mainly targeting foreign visitors.
In the third category of studies—studies related to the accessibility between two points—Zhang et al. (2000) [20
] focused on the accessibility between two stations and evaluated the serviceability of railway lines in Japan and Korea. Additionally, there are systems that deal with accessibility besides those presented in research papers. For example, “Atsumaru-now” [21
] is a web system where users can search for meet-up stations from several departure stations. Moreover, Navitime [22
] and Yahoo! [23
] offer a web service to search for routes between two locations.
The first and second categories of studies mentioned above do not take into consideration the use of trains. Additionally, excluding Okuzono et al. (2015) [15
] and Mizutani et al. (2016) [17
], the preceding second category of studies only include individuals as recommendation targets and do not consider groups. Furthermore, though Okuzono et al. (2015) [15
] includes groups as recommendation targets, because user profiling is conducted by categorizing preference information using photos, there is a possibility that sightseeing spots that do not match the user’s preferences will be displayed. Moreover, Mizutani et al. (2016) [17
] only recommends sightseeing spots for groups and does not recommend meet-up locations prior to that. Regarding the existing system of the third category of studies, while a search can be conducted for meet-up stations focusing on time, distance, and the number of transfers, as only one item can be focused on, meet-ups before getting to the destination are not considered. Therefore, the present study intends to improve the accessibility calculation method proposed by Zhang et al. (2000) [20
] and demonstrate originality by recommending convenient meeting places for groups during events.
3. System Design
3.1. System Characteristics
As shown in Figure 1
, the system in the present study is composed based on the accessibility database, and the recommendation system is linked with Google Maps, as well as external SNSs (Twitter and LINE). The purpose of the system is to reduce congestion around event venues by recommending meet-up stations before and after the event for groups, as well as to recommend stations that are easily accessible for everyone within a group, based on the nearest station to the home of each individual within the group and the accessibility value calculated beforehand. On the page for the recommendation function for the meet-up stations, the top three stations with the highest accessibility values will be recommended. By displaying the total value of the travel time required and the number of transfers to the three stations for each individual within the group, users can make a final decision on the station at which to meet.
Further, by creating a link with Google Maps on the page for the recommendation function for the meet-up stations, the users can display digital maps, as well as search for any available facilities around the meet-up station and a route to the meet-up station from any point. Additionally, buttons linked with Twitter and LINE were installed on the page for the recommendation function for the meet-up stations, which enable the users to share information concerning the recommended meet-up stations within the group using their own account for the above SNSs. Accordingly, by recommending stations at which groups can easily meet up, the system encourages people to gather in places away from event venues. As mentioned in Section 1
, groups try to gather around the event venue before and after the event, and overcrowding around venue areas is always a serious issue.
3.2. Target Information Terminals
Though the system is meant to be used from PCs (Personal Computers) or mobile devices, as there is no difference in functions on different information terminals, the same function can be used from any device. No matter what device is used, the system is intended to support the final decision of meet-up stations for groups both before and after an event.
3.3. System Operation Environment
The system operates using the web server and database server. Figure 2
3.4. Details of the System Design
In the system, the suggested meet-up stations will include all stations (excluding the nearest station to the event venue) on the same railway line on which the nearest station to the event venue is located. The reason for this is that the purpose of the system is the reduction of congestion near the event venue, and the nearest station as the meet-up station is unsuitable. Additionally, when considering the accessibility for before and after the event, it is best to have a meet-up station that can be accessed without transfer from the nearest station.
In order to recommend a meet-up station, it is necessary to obtain information regarding the nearest station to the home of each individual within the group. As the system will be used by general people, the entry method for such information is designed to be easy and reliable. Specifically, the users can first select the line to which their nearest station belongs from a list of railway lines and then select their nearest station from a list of stations on that train line. Then, the top three stations with the highest accessibility value between the nearest station to the home of each individual within the group and the suggested meet-up stations will be recommended as the meet-up stations. Based on the reason mentioned in the previous paragraph, the nearest station to the event venue will be excluded from the suggested meet-up stations by the constraint in the algorithm of the process of the recommendation system in the backend.
shows a specific example of a meet-up station recommended to a group consisting of three individuals. In this case, Station X is easily accessible for A and B but is inconvenient for C. Either C will have to use this inconvenient route or travel separately. In order to prevent such a situation, Station Y with the highest accessibility value for all within the group will be recommended.
Moreover, in addition to the accessibility value used for the recommendations, the database is also developed with the travel time required, as well as the number of transfers. The reason for this is that the travel time required and the number of transfers to the three stations are displayed on the page for the recommendation function for the meet-up stations as reference indexes for the users when deciding upon a meet-up station after the three stations are recommended with the method mentioned above.
After the end of the operation, a web questionnaire survey and access analysis of the users’ log data were conducted in order to evaluate the system developed in the present study. It is important to consider that the users were biased by age when the system is evaluated in this section. However, this result reflects that younger people tend to frequently use various web services in their daily lives.
7.1. Evaluation Based on Web Questionnaire Survey
7.1.1. Overview of the Web Questionnaire Survey
Along with the aim of the present study, a web questionnaire survey was implemented in order to conduct (1) an evaluation concerning the use of the system and (2) an evaluation concerning the functions of the system. The web questionnaire survey was conducted for 1 week after the start of the operation. Table 2
also indicates an overview of the web questionnaire survey respondents. As shown in Table 2
, 30 out of 59 users submitted their web questionnaire survey, and the valid response rate was approximately 66%.
7.1.2. Evaluation Concerning the Use of the System
Evaluation Concerning the Compatibility with the Purpose of Using the System as well as User Tendencies
Regarding the viewing frequency of the website, 97% answered “everyday”. On the other hand, for the frequency of meeting up with people, 8% answered “everyday”, 21% answered “a few times a week”, 46% answered “a few times a month”, and 15% answered “almost never”. For meet-up locations, 23% answered “at the destination”, 46% answered “mostly at the destination”, 21% answered “mostly at locations other than the destination”, and 10% answered “locations other than the destination”. From this result, it is evident that most people tend to meet at destinations, and they try to gather around the event venue before and after the event. Therefore, by properly recommending meet-up stations to the users using the web system in the present study and encouraging them to meet up there, it is possible to reduce congestion in areas surrounding event venues.
Evaluation on the Use of the System
For the devices (multiple answers allowed) on which the system was used, 46% answered PC, 59% answered smartphone, and 5% answered tablet device. Therefore, the system is mainly used on PCs and smartphones, and more people use it on their smartphones. These results are appropriate for the purpose of using the system, which is to have the users use it any time before meeting up with their friends.
7.1.3. Evaluation Concerning the Functions of the System
describes the evaluation results for each function, as well as for the entire system.
Evaluation of the Entry Function for the Nearest Station to Home
Regarding the easiness of entry for the nearest station to home, 77% answered “I think so” or “I somewhat think so”, while 23% answered “I don’t think so” or “I don’t think so at all”. The first reason for the negative reviews is that the system did not have an interface optimized for smartphones, though 59% used the system from their smartphones, as indicated in Section 7.1.2
. The second reason is that there are many railway lines in the Tokyo metropolitan area, and it was difficult for the users to select the nearest station to home, due to the system design in which the users must first select a railway line and then a station, as mentioned in Section 3.4
Evaluation of the Recommendation Function for the Meet-Up Stations
Regarding the suitability of the calculation basis for the accessibility values, it was given an extremely high rating, with 95% answering “I think so” or “I somewhat think so”. Therefore, as mentioned in Section 4.2
, the suitability of the bases for calculation, which are the travel time and number of transfers to the destination, as well as the average number of passengers getting on and off at the arrival station per day, were made evident. Regarding the suitability of the suggested meet-ups stations, 46% answered “I think so” and 44% answered “I somewhat think so”, while 10% answered “I don’t think so”. From these results, it is evident that some of the users were not satisfied, though the recommendation results were suitable for the majority of the users. Regarding the suitability of the number of meet-up stations recommended (three stations), as 77% answered “I think so” and 13% answered “I somewhat think so”, it is clear that it was extremely well-received by the users. From these results, the users’ satisfaction can be increased by improving the calculation method for the accessibility values and the handling of such values in the system, without changing the calculation basis for the accessibility values and the number of meet-up stations recommended.
Regarding the convenience of the links with both Google Maps and external SNSs (Twitter and LINE), 92% answered “I think so” or “I somewhat think so”, while 8% answered “I don’t think” or “I don’t think so at all”. The latter was especially highly rated with 77% answered “I think so”. This shows that the users highly rated how the system was designed for the purpose of improving convenience. Additionally, the sharing function where users can share the meet-up stations with their group using external SNSs linked with the system was especially highly rated.
Evaluation Related to the Entire System
Regarding the suitability of the entire interface, 82% answered “I think so” or “I somewhat think so”, while 18% answered “I don’t think” or “I don’t think so at all”. One reason for the negative reviews was that the interface of the system was not optimized for smartphones as mentioned before.
Regarding the usefulness of the system when meeting up in groups, 90% answered “I think so” or “I somewhat think so”, while 10% answered “I don’t think”. Therefore, because the serviceability of the system is high when meeting up in groups, by making the improvements mentioned later in Section 7.3
, it can be anticipated that the users will be able to make more use of each function of the system.
7.2. Evaluation Based on Access Analysis
In the present study, an access analysis was conducted using the users’ log data during the operation period. This analysis was conducted using Google Analytics, which is a web access analysis service provided by Google. A PHP program with the analysis code made using Google Analytics was created, and for the target websites for the access analysis, the access log was obtained by scanning the PHP program made for the access analysis of the program in each page within that website.
shows the access methods to the system. With information terminals used as access methods, 40% were PCs and 60% were mobile devices, indicating that there was more access from the latter rather than the former. This is because many people have begun to use smartphones as a simple and convenient way to obtain information in recent years. The system design was effective in that it was made to be used regardless of the type of device in order to eliminate the difference in information that can be obtained depending on the device. However, as mentioned in Section 7.1.3
, as the system did not have an interface optimized for smartphones; thus, despite there being more access from smartphones than PCs, the users found it somewhat difficult to use the system from their smartphones.
shows the number of visits according to page (top 10). As it is made evident in Table 4
, there were 59 users using the system, and the number of visits to the page for the recommendation function for the meet-up stations was 119, which means each user used this function twice. This may be because the users attended each event with a different group and used the system with various groups. Additionally, the number of visits to the page for the explanation about the system was 121, suggesting that the users fully understood the system while using it. Therefore, it can be said that the users understood the system when evaluating each function in the questionnaire survey.
7.3. Extraction of Improvement Measures
The issues concerning the system were extracted based on the results of the web questionnaire survey, as well as the access analysis of the users’ log data and are summarized below.
The system developed in the present study displayed the screen for PCs on mobile devices including smartphones and tablets. Because of this, users found it difficult to use the system, particularly from their smartphones, especially when they selected the nearest stations to their homes. Therefore, in addition to implementing an interface for mobile devices, it is necessary to improve the page for the entry function for the nearest station to home.
The system recommends meet-up stations by referring only to the accessibility value from the nearest station to home to the suggested meet-up stations and does not consider the time, distance of the event venue, or the nearest station to the destination as the meet-up station. Therefore, it is necessary to take this into consideration and enhance the recommendation system.
Because the system operated with Nippon Budokan as the only destination, the system’s serviceability would be improved if the number of event venues as destinations increased.