in-Car Entertainment via Group-wise Temporary Mobile Social

Networking

Mario G. C. A. Cimino

, Antonio Di Tecco

1,2 b

, Pierfrancesco Foglia

, Raffaele Giannessi

Jacopo Malvatani

, Cosimo A. Prete

and Giulio Rossolini

Department of Information Engineering, University of Pisa, Pisa, Italy

University of Florence, Florence, Italy

{g.rossolini, r.giannessi4, j.malvatani}@studenti.unipi.it

Keywords: Autonomous Car, in-Car Entertainment, Temporary Mobile Social Networking, Music Streaming Service.

Abstract: Next generation cars will increase the passengers’ time for fun and relax, as well as the number of unknown

passengers traveling together. A key functionality to improve the users’ experience is that of Temporary

Mobile Social Networking (TMSN): where passengers form, for a limited-time, a mobile social group with

common interests and activities, using their already available social network accounts. The goal of TMSN is

to automatically redesign the users’ profiles and interfaces into a group-wise passengers’ profile and a

common interface, by reducing isolation and enabling socialization. In this paper, a TMSN-inspired music

selection is proposed and developed via the Spotify music streaming service. Early results are promising and

encourage further developments towards the concept of in-car entertainment.

1 INTRODUCTION

Nowadays mobile technology guarantees seamless

network connectivity and application functionality to

travelers. As a consequence, car passengers can easily

enrich their traveling experience via mobile social

networks, providing various entertainment such as

music and video streaming, feed, stories, and so on

(Coppola, 2016; Bilius, 2020).

In the current social network market, some major

products of interest for in-car entertainment are

represented in Figure 1. Facebook is characterized by

focus on user’s feed and status; it allows to share

thoughts and emotional feedback. YouTube is the

most popular social video platform, on which users

can find and share video about their interests. Spotify

is leading the market of social music, where favorite

artists, genres, and playlists are managed. Finally,

Instagram is the repository of stories, with

photographs and short videos of friends and

influencers.

https://orcid.org/0000-0002-1031-1959

https://orcid.org/0000-0003-0126-8079

https://orcid.org/0000-0001-6432-4504

https://orcid.org/0000-0002-8467-8198

Given the role that Mobile Social Networking

(MSN) can play in the smart city paradigm and in the

next generation cars, a great development of MSN

industry is expected towards in-car infotainment,

enabled by the ongoing, high usable car’s interior

redesign (Coppola, 2016; Foglia, 2014) and the

increasing AI-based interaction (vocal interaction,

emotion recognition, and so on) (Rong, 2021).

Indeed, MSNs have already been introduced in cars

via Android Auto or Apple CarPlay, the major

platforms for smart phone interoperability with car's

dashboard information and entertainment unit.

Specifically, Figure 2 illustrates the autonomous

driving levels classified by the Society of Automotive

Engineers (SAE) (Coppola, 2016):

• level 0 – manual driving, i.e., a completely manual

vehicle, without electronic stability program, paring

assistance, or any kind of assistance system;

• level 1 – driver assisted but steering the vehicle

independently, i.e., cruise control, lane departure

warning, emergency brake assistance;

432

Cimino, M., Di Tecco, A., Foglia, P., Giannessi, R., Malvatani, J., Prete, C. and Rossolini, G.

in-Car Entertainment via Group-wise Temporary Mobile Social Networking.

DOI: 10.5220/0011096000003191

In Proceedings of the 8th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2022), pages 432-437

ISBN: 978-989-758-573-9; ISSN: 2184-495X

• level 2 – partial automation, i.e., the vehicle can

independently perform individual driving, such as

parking, navigating stop-and-go traffic, lane

departure warning, distance warning;

• level 3 – conditional automation, i.e., autonomous

driving under certain conditions;

• level 4 – high automation, i.e., vehicle controls

complete journeys on the highway, or in city traffic

predominantly independently;

• level 5 – full automation, i.e., neither driving ability

nor a driving license are required to use the vehicle.

Figure 1: Some major social network platforms of interest

for in-car entertainment.

Figure 2: Autonomous driving levels and MSN access.

At level 3, in which the car controls a significant

number of driving operations, the number of MSN

products can sensibly increase. In this trend, car

sharing is also expected to become popular. As a

consequence, next generation cars will increase the

passengers’ time for fun and relax, as well as the

number of unknown passengers traveling together.

A deeply discussed issues of MSN is the negative

effect of the prolonged use of mobile apps: many

studies show the orthopedic injuries and the

psychological diseases that can affect a social user,

resulting in isolation, anxiety, depression, and so on.

In this regard, ambient intelligence and next

generation car (Bilius, 2020) are expected to solve the

inherent limits of smart phones and of current MSN.

To further enrich the interaction of an MSN user

with the physical world, by reducing his isolation, an

important paradigm made possible by next generation

car is the group-to-many interaction, illustrated in

Figure 3. In the traditional one-to-many interaction,

the user is alone in his physical world and is

connected to the others via MSN. With the group-to-

many interaction, a group of users is temporarily

living in a common physical space, interact in person

and share their experience with the MSN as a whole

group.

(a)

(b)

Figure 3: One-to-many interaction paradigm vs. group-to-

many in the context of in-car MSN.

In the literature, the concept of Temporary MSN

(TMSN) has attracted attention as a conceptual

framework to be used at hotels, concerts, theme parks,

sports arenas, where people form a mobile social

group for a limited time, by having a common

physical interaction (Yin, 2019). People confined in a

specific place are allowed to join the TMSN using

their personal account, and acting via a group-wise

interaction with the others, improving the mobile

users’ experiences with such temporal friends. A

surrogate of this concept is Spotify Group Session, a

beta testing feature that allows up to six people to

share control over the music playing in the

background of a physical or virtual get-together

(Spotify GS, 2022). There is an expansion of Spotify

in-Car Entertainment via Group-wise Temporary Mobile Social Networking

433

Group Sessions available in Android Automotive OS,

an Android-based infotainment system that is built

into vehicles, such as the Polestar 2 (2022).

In the context of hotels, LobbyFriend was the first

TMSN, enabling hotels to stay connected to guests

throughout their stay, within the one hotel or across

multiple hotels within the vicinity. It was inspired by

loneliness while traveling often for business. When a

guest checks out of the hotel, all interactions in the

TMSN are erased (Yin, 2019).

TMSN is beyond sharing a common space and a

collaborative playlist (Spotify CL, 2022; Spotify FM,

2022), it concerns algorithms for the exploitation of

the user’s profile and the current environmental

context, enabling an augmented interaction via

proactive services such as recommendation (Cimino,

2011). Social network platforms are today strongly

managed by powerful analytics and algorithms (Lu,

2015; Zhang, 2019), also based on key techniques of

modern sociology. As a such, in this study TMSN is

an additional intelligent layer on an ecosystem of

services available on next generation cars.

In this paper a group-wise TMSN is proposed as a

design paradigm for in-car entertainment.

Specifically, in the context of social music, a

functional design is illustrated. A prototype has been

implemented, based on Spotify API (Spotify API,

2022), and experimented.

The paper is structured as follows. Section 2

illustrates the core concepts and functional design of

the proposed approach. Experimental studies are

covered by Section 3. Conclusions are drawn in

Section 4.

2 CORE CONCEPTS AND

FUNCTIONAL DESIGN

In the context of audio streaming, a TMSN based

playlist recommender exploits both ambient (car) and

MSN data. For interoperability reasons, the design

should be based on a standard music ontology (MO,

2022), which provides a vocabulary for managing

music-related data across multiple applications

(Ciaramella, 2010).

Figure 4 illustrates the fundamental concepts and

static relationships in the context of audio streaming

recommendation, as an ontology diagram. In figure,

each concept is enclosed in a rectangular shape.

Concepts are connected by relationships, represented

with labelled directed edges. Some concepts are also

characterized by properties, listed in lowercase

letters. The fundamental outcome of this ontology is

a recommendation of a track, based on the social

profiles of the passengers (containing listened tracks,

artists, and genres). Specifically, from the top-middle,

a Passenger is in a Car, is in a Mobile Social Net,

listens a Track. A Track is made by an Artist, and

belongs to a Genre. A Car plays a Track, and

manages a Temporary Mobile Social Net, which

generates a Passengers Profile. On the other side, a

Mobile Social Net builds a Social Profile, which is

made by listened tracks, artists and genres. A Social

Profile is merged in a Passengers Profile. Finally, the

Passenger Profile recommends a Track.

Figure 5 shows a protocol of an audio streaming

recommender based on TMSN and Spotify analytics,

in a standard graphical representation called Business

Process Model and Notation (BPMN). The protocol

is built on the ontology in Figure 4 and covers only

the essential aspects of the proposed approach, for the

sake of readability.

The BPMN is based on a solid mathematical

foundation, to enable the execution, simulation, and

automation of consistency checking (Cimino, 2017).

It is also suitable to standardize and facilitate

communication between all stakeholders. In BPMN,

a rectangular area represents a participant who takes

part in a protocol, via message exchange. In each

rectangular area, the protocol is managed via events,

activities, and decision/merge nodes, represented by

circles, rounded boxes, and diamonds, respectively.

Sequence flows and data flows are represented by

solid and dotted arrows, respectively. Finally, data

storages are represented by cylindrical shapes.

Figure 4: TMSN Ontology in the context of audio streaming

recommendation.

The protocol starts on the top left (white envelope

in a thin circle), when new passengers are detected by

the TMSN, and ends when a playlist is determined

(one of the black envelopes in thick circles). As a first

task, the TMSN asks the social profiles of all

VEHITS 2022 - 8th International Conference on Vehicle Technology and Intelligent Transport Systems

434

Figure 5: BPMN protocol of an audio streaming recommender based on TMSN and Spotify analytics.

passengers to Spotify Analytics. The gear icon for the

task means that is a service task, i.e., supported by

web services offered by Spotify. As defined by the

ontology, the collected social profiles contain listened

tracks, artists and genres of each passenger. Social

profiles are archived for the next steps of the protocol.

Then, a script task (i.e., a task internally developed,

characterized by a sheet icon) identifies the set of

common tracks between all passengers. If some

tracks are found, then the recommended play list is

generated. Otherwise, a script task identifies the

common artists. If some common artists are found, a

service task asks the top tracks for them to Spotify

Analytics and, if some are found, the recommended

play list is generated. Otherwise, a service task asks

to Spotify Analytics the recommended tracks for

common artists and, if some are found, the

recommended play list is generated. Otherwise, a

script task identifies the common genres in social

profiles and, if some are found, a service task asks the

recommended tracks for common genres to Spotify

Analytics, to generate the recommended playlist.

Finally, if the recommended tasks are not found, a

service task asks the recommended tracks for top

artist to generate the playlist.

3 IMPLEMENTATION AND

EXPERIMENTAL STUDIES

The protocol is purposely designed to heavily exploit

Spotify Analytics services. Indeed, it has been

implemented and experimented on both a desktop

computer and a Raspberry PI3b+, a small Chip

Multiprocessor (CMP) (Foglia, 2014) single-board

computer, equipped with WIFI for interfacing with

smartphone, and 4G capabilities for interfacing with

Spotify web services. The web application controller

has been developed on Apache2 web server: it

manages the authentication required by Spotify API,

via Javascript. The host access point daemon has been

implemented on Hostapd. The protocol logic has

been developed on Python3, using Spotipy, a third-

party library to access the Spotify APIs, and

DBConnection to manage Social Profiles on a local

MariaDB data base management system. The client-

server communication is managed via the OAuth2, an

industry-standard protocol for providing token-based

authorization flows for web applications, mobile

phones, and living room devices. A keep-alive

mechanism has been also implemented to detect

dynamic join/disjoin of passengers. Finally, a

persistent buffering of social profiles and playlists has

been implemented to speed up the recommendation

requests time.

Six people have been involved to carry out the

following sessions: 8 sessions made by 2 passengers,

5 sessions made by 3 passengers, and 5 sessions made

by 4 passengers.

At the end of a recommended track, each

passenger has provided a binary like/dislike rating. At

the end of the recommended play list, for each

passenger, the approval rate has been calculated as

follows:

𝑎𝑝𝑝𝑟𝑜𝑣𝑎𝑙 𝑟𝑎𝑡𝑒 =

| |

| |

× 100 (1)

As a result, Figure 6, Figure 7 and Figure 8 show

the approval rate histograms for sessions of 2, 3 and

4 passengers, respectively. It is clear that, for

increasing number of passengers, the approval rate

decreases. This can be ascribed to the smaller

common tracks, artists and genres between

passengers. Figure 9 clearly shows this trend, via the

mean approval rate against passengers’ number.

Finally, Figure 10 shows the generalization

ability of the recommender, i.e., to go beyond the

in-Car Entertainment via Group-wise Temporary Mobile Social Networking

435

passenger’s playlist (known tracks). It is calculated as

follows:

𝑔𝑒𝑛𝑒𝑟𝑎𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒 =

| |

| |

× 100 (2)

Assuming that the tracks in its own playlist are all

liked, the maximum value is 1 (i.e., no additional

liked tracks). For increasing unknown tracks that are

liked, the value decreases (i.e., better generalization).

Figure 6: Approval rate histogram for 2 passengers.

Figure 7: Approval rate histogram for 3 passengers.

Figure 8: Approval rate histograms for 4 passengers.

Figure 9: Mean approval rate against passengers’ number.

Figure 10: Mean generalization rate against passengers’

number.

Overall, for increasing passengers the

generalization ability increases, for the higher variety

of tracks, authors and genres available.

VEHITS 2022 - 8th International Conference on Vehicle Technology and Intelligent Transport Systems

436

4 CONCLUSIONS

In this paper, group-wise Temporary Mobile Social

Networking recommender is proposed as a design

paradigm for in-car entertainment. Specifically, in the

context of social music, a functional design is

illustrated. A prototype has been implemented, based

on Spotify Analytics and Raspberry PI3, and

experimented involving six people on various

sessions.

The achieved approval and generalization rates

show that, for increasing number of passengers, the

approval rate decreases, for the smaller common

tracks, artists and genres between passengers.

However, for increasing number of passengers, the

generalization ability increases, providing liked

tracks that are not already known.

Although a more in-depth exploration of the

techniques is needed, together with an enrichment of

the experiments, the early results are promising, and

show the potential of the proposed approach.

An extensive study in this direction can be a future

work to bring a stronger contribution in the field.

ACKNOWLEDGEMENTS

Work partially supported by the Italian Ministry of

Education and Research (MIUR) in the framework of

the CrossLab project (Departments of Excellence,

Lab Cloud Computing, Big Data & Cybersecurity,

Lab Augmented Reality, Lab Advanced

Manufacturing).

REFERENCES

Bilius, L. B., & Vatavu, R. D. (2020). A multistudy

investigation of drivers and passengers’ gesture and

voice input preferences for in-vehicle

interactions. Journal of Intelligent Transportation

Systems, 25(2), 197-220.

Castellano G., Cimino M.G.C.A., Fanelli A.M., Lazzerini

B., Marcelloni F., Torsello M.A. (2013). A

Collaborative Situation-Aware Scheme based on an

Emergent Paradigm for Mobile Resource

Recommenders, Journal of Ambient Intelligence and

Humanized Computing, 4:421-437, doi:

10.1007/s12652-012-0126-y

Ciaramella, A., Cimino, M.G.C.A., Marcelloni F., Straccia,

U. (2010). Combining Fuzzy Logic and Semantic Web

to Enable Situation-Awareness in Service

Recommendation", Lecture Notes in Computer

Science, 6261:31-45.

Cimino M.G.C.A., Lazzerini B., Marcelloni F., Castellano

G., Fanelli A.M., Torsello M.A. (2011). A

Collaborative Situation-Aware Scheme for Mobile

Service Recommendation, Proceedings of the 11th

International Conference on Intelligent Systems Design

and Applications, 130-135 (2011), doi:

10.1109/ISDA.2011.6121643

Cimino, M.G.C.A., Palumbo, F., Vaglini, G. et al. (2017)

Evaluating the impact of smart technologies on harbor’s

logistics via BPMN modeling and simulation.

Information Technology Management, 18, 223–239.

Coppola, R., & Morisio, M. (2016). Connected car:

technologies, issues, future trends. ACM Computing

Surveys (CSUR), 49(3), 1-36.

Foglia, P., & Solinas, M. (2014). Exploiting replication to

improve performances of NUCA-based CMP

systems. ACM transactions on embedded computing

systems (TECS), 13(3s), 1-23.

Foglia, P., Zanda, M., & Prete, C.A.:, I. (2014). Towards

relating physiological signals to usability metrics: a

case study with a web avatar. WSEAS Transactions on

Computers, 13, 624.

Lu, J., Wu, D., Mao, M., Wang, W., & Zhang, G. (2015).

Recommender system application developments: a

survey. Decision Support Systems, 74, 12-32.

MO, Music Ontology, musicontology.com, accessed 2022.

Polestar 2 electric car, www.polestar.com/us/polestar-2/,

accessed Jan 2022.

Rong, Y., Han, C., Hellert, C., Loyal, A., & Kasneci, E.

(2021). Artificial Intelligence Methods in In-Cabin Use

Cases: A Survey. IEEE Intelligent Transportation

Systems Magazine.

Spotify API, support.spotify.com/us/article/spotify-in-the-

car/, accessed Jan 2022.

Spotify CL, Collaborative Playlist, support.spotify.com/us/

article/collaborative-playlists/, accessed Jan 2022.

Spotify FM, Family Mix, support.spotify.com/us/article/

family-mix/, accessed Jan 2022.

Spotify GS, Group Session, support.spotify.com/us/article/

group-session/, accessed Jan 2022,

Yin, Y., Xia, J., Li, Y., Xu, W., & Yu, L. (2019). Group-

wise itinerary planning in temporary mobile social

network. IEEE Access, 7, 83682-83693.

Zhang, S., Yao, L., Sun, A., & Tay, Y. (2019). Deep

learning based recommender system: A survey and new

perspectives. ACM Computing Surveys

(CSUR), 52(1), 1-38.

in-Car Entertainment via Group-wise Temporary Mobile Social Networking

437