=Paper=
{{Paper
|id=Vol-1803/paper8
|storemode=property
|title=Personalizing Support to Older Adults who Look
for a Job with the SpONSOR Platform
|pdfUrl=https://ceur-ws.org/Vol-1803/paper8.pdf
|volume=Vol-1803
|authors=Amedeo Cesta,Gabriella Cortellessa,Riccardo De Benedictis,Francesca Fracasso,Daniel Baumann,Stefano Cuomo,Julie Doyle,Adnan Imeri,Djamel Khadraoui,Pierre Rossel
|dblpUrl=https://dblp.org/rec/conf/aiia/CestaCBFBCDIKR16
}}
==Personalizing Support to Older Adults who Look
for a Job with the SpONSOR Platform==
https://ceur-ws.org/Vol-1803/paper8.pdf
Personalizing Support to Older Adults who Look
for a Job with the SpONSOR Platform ⋆
Amedeo Cesta1 , Gabriella Cortellessa1 , Riccardo De Benedictis1 ,
Francesca Fracasso1, Daniel Baumann2 , Stefano Cuomo3 , Julie Doyle4 ,
Adnan Imeri5 , Djamel Khadraoui6 , and Pierre Rossel7
1
CNR, Italian National Research Council, ISTC, Rome, Italy
2
FST, Fondation Suisse pour les Téléthèses, Switzerland
3
I+, Florence, Italy
4
Netwell, Netwell Centre and Casala, Dundalk Institute of Technology, Ireland
5
UNIGE, University of Geneva, Switzerland
6
LIST, Luxembourg Institute of Science and Technology, Luxembourg
7
CoSt, Coherent Streams, Switzerland
Abstract. Nowadays the loss of job of people close to retirement and
in general in old age is a reality shared by many European countries.
This condition causes serious difficulties for older employees when re-
turning to the job market and inevitably leads to a deterioration of their
emotional state and well-being. On a different perspective, following the
concept of active aging, the fact that older people continue to perform
volunteering activities exploiting their skill and abilities is extremely pos-
itive for their well being. Volunteering job is, however, known for being
very volatile and dynamic with people leaving activities, often, after a
relatively short period of time. The SpONSOR project tries to help this
particular segment of population by creating a platform that facilitates
the effective match between demand and offer of work integrated with
services particularly tailored for older people. A specific contribution is
related to the personalization of support to workers obtained by spe-
cializing plan-based technology to generate interactions over time able
to adapt to users and give them a sense of care. The paper contains an
overview of the project by presenting hints from the user requirements
analysis and an overview of the software architecture. Additional details
are given on a module called “Personalized Interaction over Time” (PIT)
dedicated to synthesize tailored suggestions for users during their long
lasting interactions with the software platform.
1 Introduction
The Active and Assisted Living (AAL) Joint Program is a European funding
initiative aiming at promoting the synthesis of new ICT solutions for supporting
⋆
Authors work is partially funded by the Ambient Assisted Living Joint Program
under the SpONSOR project (AAL-2013-6-118 – http://sponsor-aal.eu/). Send cor-
respondence to amedeo.cesta@istc.cnr.it
people to age well. The program (active since 2008 with the previous extended
name of Ambient Assisted Living) is now strongly focusing on the fact that older
adults should remain active as long as possible in order to preserve their role
in the society longer. The annual Call number 6 (2013) for project proposals
specifically asked for “the development of ICT-based solutions which enable
older adults to continue managing their occupation at work in an office, in a
factory and in any working environment; in a first or subsequent career, in paid
or voluntary occupation including local social activities while preserving health
and motivation to remain active”. Additionally, the specific call was looking for
solutions that promote, enhance and sustain:
– paid activities (including for example professional, entrepreneurial/small busi-
ness and self-employment)
– unpaid activity (e.g., volunteering, knowledge sharing, counseling).
Today’s older adults have a positive self-image and bright expectations for their
future. Some of them would like to continue working, preserving room for leisure
time; some intend to increase their voluntary engagement, while others wish to
change their field of activity and take up a completely new profession or career.
Yet, in many cases, the active older adult is still confronted with a “deficit
model of age” that – often unsubstantiated – assumes that occupational efficiency
and general learning ability decline with age. This implies that older adults are
less innovative, less productive and less able to work under pressure than their
younger colleagues.
The SpONSOR project is one of
the selected proposals for the above-
mentioned call. The project aims
at developing, testing and imple-
menting an ICT platform that fa-
cilitates the posting, browsing and
exchange of key information be-
tween competence-offering seniors
and search-based requests, from
competence-demanding organizations
from the public, private and volun-
tary sectors. SpONSOR aims at en-
Fig. 1. An initial view of where the SpON-
hancing senior persons’ access to a
SOR platform is expected to work.
wide range of occupational positions
meeting, in this way, the aims of the Call. The initial schematization created in
the project is given in Figure 1. The perspective taken is the one of supporting
Organizations that favour occupation (upper part of the figure) creating a soft-
ware platform that facilitates contact between people who offer their work and
people in need of support (lower part). Indeed during the development of the
project we have particularly focused on the relation between producers of work
(the Organizations) and the consumers (the older workers). The third group of
people (those in need for a work support) will be de-facto integrated inside the
role of the organizations.
Additionally, from the user requirement analysis it clearly emerged that loos-
ing job when at an higher age is really problematic. The older people feel aban-
doned, marginal in the society and at high risk of depression. One of the charac-
teristics that emerged from user requirement analysis is the need for functional-
ities able to give the sense of continuous assistance to people when they interact
over time. In the paper, after a general presentation of the project, we specif-
ically report on some personalization characteristics that we have added using
intelligent interactive technology based on a plan-based internal representation.
2 User Requirement Elicitation
The first step for user requirement elicitation envisaged a study of the state of
the art with the main aim to investigate the existence of any sort of previous
platform designed with the same intention of SpONSOR. Further efforts have
been spent in order to get in contact with users and associations involved in
seeking occupation opportunities for seniors in the countries partners of the
project (Italy, Luxembourg, Switzerland and Ireland).
State of the art
Within the European landscape, some realities exist whose aim is to foster the
exchange between demand and supply specifically among seniors. Some of them
did not have success, but all of them can provide valuable suggestions for SpON-
SOR. Some example are the following:
Competence-senior.com The company Competence seniors Sàrl, based in the
Canton of Geneva, created a website [2] in support of their mission to facil-
itate senior employment, providing space and profiling options for posting
offers (seniors proposing their competences or searching for job positions)
and demands (organizations proposing job positions), with some advisory
indications and a press release section as marketing extension. The whole
website was functioning as a profiling and match making mechanism for ad-
vertisements, which was supposed to facilitate, on both sides, successful job
search/job recruitment options. Unfortunately, this association shut down
the website in 2014 because of a low number of users.
RetiredButAble.com The website Retiredbutable.com provides for the UK
territory a profiling and posting service supporting senior work with a huge
list of service categories and preemptive geographical requests. It is also
“Facebook active”. The website [8] started very well, attracting a lot of
postings, but ending up in rather poor search results, and after a few months,
the number of postings gradually decreased.
SeniorsAvotreService.com This new portal [10] seems to be supported by a
lot of institutional and even political alliances and, as a matter of fact, shows
a lot of hits. It looks promising, slightly more appealing than Competence-
seniors.com while encompassing a broader set of options from micro-trading
options to job positions. It presents some stimulating posting examples (and
also a very “French-focused” job-related advisory check list).
From the experiences described above, we can assume that (1) some match-
making capabilities are necessary, but (2) they have to be produced within a
stimulating in-flow of senior- and occupation-relevant information, including
linkages with social networks, (3) they have to be open to a wide variety of
occupational forms, (4) they have to be mediated by organizations dedicated to
support these activities and the possibility for seniors to be involved, and (5)
they have to incorporate provisions for iterations, examinations of alternatives,
adaptations and updates, evaluation and mentoring regarding the output of the
types of match-making processes.
Other platforms do not perfectly fit with the SpONSOR aims, but are worth
being mentioned. Give&Take [5] is an AAL project which tries to facilitate
service exchange among seniors, with quite a complete match-making capacity
through a digital platform, creating new opportunities for senior citizens to con-
tribute to the society as volunteers and caregivers in their local communities.
In the same perspective, we can also mention a platform aimed at facilitating,
within the French- speaking constituency, forms of activities of many kinds for
seniors willing to share their interest with others: Quintonic [7]. These two
examples are not as ambitious, from the job focus perspective, as SpONSOR,
but they are nevertheless providing a match-making mechanism, with support-
ive services and, in the last case, a rather appealing interface. They have been,
therefore, a source of inspiration for SpONSOR. Finally, although they are not
strictly senior-oriented, there are two reality worth to be mentioned. Vicker [12]
is a new Italian service that is active since early 2016 and is growing up at na-
tional scale. It is a platform for Computers, Smartphones and Tablets that allows
people to find a service provider on one side, and to offer their own competence,
on the other. It specifically supports the exchange of one-shot jobs among citi-
zens by ensuring competence, and safety and fair payments. Service beneficiaries
can also provide feedback to their providers in order to foster their skills upon
the platform. Finally, TimeRepublik [11] is a platform, active at international
level, which fosters the exchange of services among people with the peculiarity
that the trading currency is represented by time (TimeCoin) instead of money.
User involvement
Beside the analysis of the state of the art in terms of existing solutions, an ad-
ditional effort has been carried out in order to get feedback from users. Focus
groups and semi-structured interviews have been organized around a guided dis-
cussion on post-retirement jobs, meant both as paid and volunteering activities,
that involve old people. Users who could possibly benefit from the SpONSOR
platform have been involved in order to get their opinion on the services that the
platform should provide and to collect the perspective from users on their needs
and what are the characteristics of the system. Additionally, also jobs for those
who are near retirement and have lost their job has been considered as a topic
addressed within the focus group since this kind of users could also represent
possible beneficiaries of the platform.
Participants have been
recruited among: (1) as-
sociations aiming at fos-
tering active aging for re-
tired persons; (2) associ-
ations with the aim to
support older adults in
finding jobs in order to
reach the retirement age;
and (3) elderly commu-
nities. In total, 50 mem- Fig. 2. Pictures from the focus group in Italy for the user
bers from 21 associations requirements elicitation.
among Italy (Figure 2),
Luxembourg, Ireland and Switzerland have been involved in this preliminary
investigation. Participants have been asked about the associations they belong,
so as to get a better understanding of their mission, and about the idea of post
retirement (or near too retirement as well) occupation, in order to understand
the difficulties and the related needs.
Moreover, participants have been exposed to the scope of the SpONSOR
project and encouraged to investigate possible ways to overcome difficulties and
the emerged needs through a digital platform.
Indications for developers
In order to produce effective indications for the developers, the transcriptions of
focus groups has been subject to a critical analysis with the aim to synthesize a
detailed list of user requirements. The whole amount of works carried out in order
to analyze the user requirements coming from focus group is deeper described in
[9]. This section attempts to summarize the most interesting findings. It emerged,
indeed, that the SpONSOR platform should necessarily address key requirements
in the following areas:
Profiling of users competences. User profiling beyond professional qualifi-
cation: SpONSOR is supposed to support user profiling by including more
than formal aspects as educational qualifications. Skills and competencies
acquired beyond the professional background have to be taken into account,
as well as social skills. Intelligent analysis of user’s competence: SpONSOR
should offer a service that supports a “human selector” to better analyze the
user’s competences in order to provide an optimized matching with vacant
activities/jobs by, for instance, showing different aspects of the persons that
could fit with a range of jobs.
Privacy and Ethical issues. Ensuring of not discriminatory announcements:
SpONSOR should rely its reasoning and criteria for the matchmaking on
individual skills and competences or any similar valuable characteristic linked
to the person and not on discriminatory factors like age, gender and race.
Privacy on sensitive data: SpONSOR must apply all the needed procedures
to keep sensitive data protected and disclosing its protection scheme to the
users. Security Clearance: SpONSOR should indicate whether police vetting
is required for a position and/or a project, and how to go about getting it.
Profiling of job opportunities. Efficient and effective categorization of roles
representing job opportunities and actual tasks to be performed: SpONSOR
should be able to classify required roles in order to facilitate matchmaking
with user profiles based not only on educational qualification, or profes-
sional background, but also on other skills and competencies of the person,
as well as on other personal characteristics. Assistance, guidelines to de-
fine profiles: SpONSOR should provide a service giving a methodology, tips,
pitfalls to avoid and reference cases for profiling issues. Help in orienting
the users seeking a job: The platform should be organized in a way that
the persons receive support to find the roles that better fits his/her compe-
tences/qualities/talents. Include an expression of interest form: SpONSOR
should allow people to express their interest in a particular job/advertisement
or declare (occupational) areas of interest, even if they have not been matched
to it.
Usability, accessibility and portability. Simple and easy terms: The SpON-
SOR platform should use simple terms since it has to be used by non expert
users. Whenever possible, technical terms have to be avoided, Portability:
SpONSOR should be available for use on different devices (smart phones,
tablets are seen as the best solution but also PC application versions should
be available). Accessibility: SpONSOR should make it possible for seniors to
easily read the text on web site. (i.e. proper font size).
Social Interaction. Implementation of a forum for exchanging information:
SpONSOR shall provide the possibility to the subscribers, individual or col-
lective, to interact with each others. A dedicated space within the platform
where users can exchange information, including experience, best practice
suggestions on specific topics and also areas of interest is needed. Provision
of services linked to communities: SpONSOR shall organize services on local
basis or facilitate ways of already existing local community support forms.
The territory is based not on political boundaries (i.e. municipalities), but
on social boundaries (i.e. communities, people networks).
Metrics and Validation. Assessment of the level of update for each job adver-
tisement: SpONSOR should allow monitoring how long each advertisement
is active on the platform and provide a clear feedback if the job is still va-
cant or not. Successful matchmaking rating visualization: SpONSOR should
provide the possibility to track the success of matchmaking between demand
and offer. Feedback to job respondents: SpONSOR should allow the organi-
zation to provide feedback to any person who has applied for a job at any
stage of the recruitment process, i.e. expression of interest, informal chat,
and interview.
The above mentioned requirements represent operative indications for the plat-
form developers in order to design an ICT solution by accomplishing with users’
needs and suggestions. Nevertheless, it is fundamental not to loose some in-
sights coming from the meetings with the senior-oriented associations. Actually,
in most cases the associations already rely on any sort of matchmaking solution,
digital or manual, in order to recruit people and, because of this, they have been
able to provide useful suggestions with regard to this function of SpONSOR,
in terms both of what can be useful and what should be avoided. Nevertheless,
interesting reflections emerged referring to a portion of population which results
to express peculiar needs beyond the mere service of matchmaking.
The personalization issue
Although the involved associations are engaged in different areas with retired or
near-to-retirement people, the manifested needs barely differ according to their
specificity, and a pervasive need of personalized services emerged in all cases. In
fact, the user requirements claimed for an ICT solution able to adapt, throughout
time, according to individual and contextual changes. A common element that
emerged from user requirement gathering is the fact that the SpONSOR platform
should provide personalized services according to each single user specificity and
dynamically interact with the users by adapting to different personalities and
preferences. This should represent the main characteristic which makes SpON-
SOR the successful solution beyond the state of the art. In fact, this platform
should be a means through which the senior is able to leverage on his/her indi-
vidual characteristics and aptitudes, beside skills and learned competences. On
this basis, SpONSOR should be able to provide personalized services to the per-
son by fostering the motivation of pursue a goal, namely finding an occupation.
In this perspective, it becomes crucial the possibility to provide SpONSOR with
proactive capabilities and the possibility to modulate through time its behavior
according to upcoming environmental events or changes in the users status and
preferences.
3 The SpONSOR Architecture
The analysis of the user requirement has led to the definition of a software ar-
chitecture able to supply the required services to the different users. Specifically,
the different components are glued together in an high level architecture that
supports the scenarios emerged from the elicitation of the user requirements.
A first distinction worth being captured concerns the different SpONSOR
users (see Figure 3). The intelligent environment offers services to different
targets. Specifically, the involved actors are: (i) Main Users (MU), senior-
supportive organizations which have all sorts of name. Such organizations include
public and private agencies, NGOs and local associations. In order for being qual-
ified as being part of this category, organizations must just involve some certain
level of institutionalization (formal membership or legal framing), a clear mis-
sion of support regarding seniors and some experience already in that respect on
which we can build. Such organizations may make use of senior occupation one
way or another: for example, they can be private firms looking for specific kinds of
employees.
(ii) Subsidiary Users (SU),
Seniors themselves, envisaged as
(not being “organization-related”)
individuals, who in the experi-
mental build-up of SpONSOR,
played a specific catalyzing role
for the progression of the project
and are of course, in spite of the
organization channel chosen to
develop SpONSOR, always legit- Fig. 3. The different users of the SpONSOR so-
lution.
imate for a direct access to the
platform. The seniors should always be the real beneficiaries of SpONSOR sup-
port, but in order to be able to fulfill that capacity, intermediate steps have to
be explored and experimented, involving all sorts of senior-supportive organiza-
tions.
In addition to the above mentioned users, some intermediaries have developed
quite sophisticated means to match user profiles for different demand-and-offer
configurations, and some of them might become useful “stakeholders”, however,
their status regarding “usage” has still to be defined (partners, advisors?). So
far, we have found intermediaries in the job mediation portal business, in the
consultancy arena and also in the umbrella company sector (or other forms of
work support mediations). Others may still come up during the duration of
the project. Among additional interesting users worth considering, let us high-
light those involved in co-working and micro-trading activities, to which senior-
supportive or senior-sensitive organizations, as well as non-affiliated individual
seniors may be related, thus allowing, provided we gradually integrate this cat-
egory of users and usages, covering quite a large spectrum of domains, talents
and services. These users are not the main target of the SpONSOR definition
and development phase, but they may come into the picture as the platform gets
more consistent.
Summarizing, although SpONSOR claims to be useful for many categories
of stakeholders, it is worth highlighting that Main Users (MU), namely, senior-
supportive organizations of all kinds, and Subsidiary Users (SU), namely, indi-
vidual seniors willing to use SpONSOR, are the main target actors of the project
and, therefore, of the SpONSOR platform and its services. Nonetheless, the ar-
chitecture is readily adaptable to other types of users as, for example, tiers 2
(formal carers) and tiers 3 of all kinds (including “non senior” individuals, asking
for services which seniors may possibly deliver, thus functioning as subsidiary
substitute of a formal job offering organization) may be useful and supportive,
even though they may be indirectly benefiting from SpONSOR platform services
(provided, of course, that they are concerned by senior occupation support one
way or another). Such users might be allowed to perform the same operations
as subsidiary users and, in addition, similarly to main users, they can manage
their associated users (members).
Fig. 4. The SpONSOR architecture.
Once described the possible users to which the solution is addressed, we deal
with the high-level architecture of the SpONSOR system. To facilitate reading,
in Figure 4 we identify two main subparts:
SpONSOR Front-end representing the parts that directly interact with hu-
man users (both main and subsidiary)
SpONSOR Back-end containing the intelligent modules of the system that
guarantee added-value services.
In the simplified representation of Figure 4 we see the main information streams
in SpONSOR: the Organizations post Opportunities while the potential Workers
(or Volunteers) post applications for certain Opportunities. These two basic in-
terventions become part of the data management services. This basic data store
service is then manipulated by specialized SpONSOR modules: (a) the Match-
making facilitates the Opportunities/Applications matching offering a first level
of suggestions to the Front-End functionalities; (b) the Personalized Interaction
over Time is a module responsible for enriching interaction with the user (at
present, mainly, the workers); (c) the Legal Services is a module that flexibly
supplies information of the legislations in different countries situated with the
particular job opportunity considered in an interaction; (d) the Video-Based Ser-
vice is a specialized module aiming at facilitating the CV entry by Workers as
well as the Organizations’ descriptions. For the sake of space, modules (c) and
(d) are not part of this paper description.
Addressing more technical issues, the main software interfaces, which allow
information interchange between the two sub-components, are (see Figure 5):
a BackEndAPI, allowing the software clients to create, read, update and delete
“content”, a ProfileAPI, allowing the software clients to update users’ profiles,
and a NotificationAPI, allowing the software clients to produce notifications
to be displayed, following a push strategy, on the SpONSOR website. In the
following, we detail the sub-components of these two main modules, linking
them to some usage scenarios and, consequently, to the user requirements.
From a technical point of view, it is worth
noticing that both the BackEndAPI and the
ProfileAPI can be accessed by the front-end side
either for retrieving data or for notifying updates
to data. For this purpose, these software inter-
faces have been implemented by means of web
services, allowing, among other things, a simple
extension to other media such as, for example,
smartphones and tablets. On the contrary, the
NotificationAPI is intended to produce events
that should be captured by the users through a
push strategy, that is, without the need for the
users to reload web pages. To that end, we opted Fig. 5. Basic front-/back-end
data exchange.
for a WebSocket based technology which, among
other things, allows real-time messaging capabilities.
SpONSOR Front-end
The main user interface (or front-end) for the different users is constituted by
the SpONSOR Website (top side of Figure 4). From an architectural point of
view, it is mainly composed of two different subcomponents representing two
distinct, although following an uniform schema, websites: (a) the Main User
Portal, dedicated to Main Users, and (b) the Subsidiary User Portal, ded-
icated to Subsidiary Users. The content of the two portals is retrieved by the
back-end, following a pull strategy, through the BackEndAPI. Specifically, when
a user accesses the SpONSOR front-end, the system makes requests through
the BackEndAPI for retrieving data stored by the back-end component. In addi-
tion, the BackEndAPI offers services for creating, deleting and updating content.
Among the content that can be managed by the SpONSOR system we find users,
messages, job offerings, training offerings, volunteering offerings, recreational ac-
tivity offerings, job interviews and many others, like legal advisory contents, for
instance.
When accessing the SpONSOR website, the web application retrieves initial,
relevant and common interest information by the back-end in order to encourage
visitors to subscribe to the system. Whenever a new user decides to subscribe
to the system, the BackEndAPI is invoked by the web application for retrieving
subscription options and, once the user has chosen his/her available roles, the
information is communicated to the back-end for the persistent storage of the
new user related information. In addition, the BackEndAPI also provides an in-
terface for basic communication, allowing the retrieval of stored messages as well
as sending new ones, for example, for mentoring purposes.
The BackEndAPI constitutes the main entry point for managing different
kinds of content. Through the SpONSOR front-end, the main users use the
BackEndAPI for creating different kinds of opportunities (e.g., job offerings, vol-
unteering offerings, training courses, recreational activities, etc.), for managing
applications and job interviews as well as for managing their associated users.
Similarly, through the front-end, subsidiary users might check for available op-
portunities (e.g., job offerings, volunteering offerings, training courses, recre-
ational activities, etc.), properly filtered and/or sorted by the back-end, and
manage their accepted applications.
In addition to the BackEndAPI, the back-end component offers a ProfileAPI
allowing subsidiary users to edit their profile. Since dynamic profiling capabil-
ities constitute an innovation point with respect to previous similar platforms,
we have chosen to separate this API from the BackEndAPI. Specifically, the
ProfileAPI allows users to retrieve their profile information and to update it in
terms of bio, personal info, work experience, interests etc. It is worth to empha-
size that, by dynamically updating profile information, reasoning capabilities of
the SpONSOR back-end might be triggered for possibly generating information
both for the subsidiary users, whose profile has changed, as well as for the or-
ganizational main users, who might take decisions according to new available
information. We will provide further examples of this type of information ex-
change in Section 4.
From the SpONSOR Website point of view, the NotificationAPI offers
a service for displaying in real-time notifications generated by the SpONSOR
back-end. Such notifications include newly received messages, relevant dynamic
changes of users profiles as well as profile update requests like questions, ques-
tionnaires or other forms of gamification strategies. The main role of this software
interface is to generate events by transmitting data in real-time. Just to provide
an illustrative example, a main user, by means of his/her dedicated portal, re-
trieves stored messages by means of the BackEndAPI. Again, by means of the
BackEndAPI sends a new message to a group of subsidiary users. These latter
users retrieve stored messages still by means of the BackEndAPI however, in re-
altime, they also receive a notification of the just sent message by the main user
through the NotificationAPI.
SpONSOR Back-end
The SpONSOR Back-end represents the container of all the persistent storage
services as well as intelligent services offered by the SpONSOR platform (bottom
side of Figure 4).
The data management module offers services for storing and retrieving stored
data for other sub-components and, consequently, to back-end clients (namely,
the SpONSOR web application). Stored data includes users information, ex-
changed communications, opportunities (including job offerings, volunteering
offerings, training courses, recreational activities, and other kinds of activities),
subsidiary users applications and relations between users (e.g., associated users
for associations). In addition to the above mentioned services, the data man-
agement module keeps track of mentoring messages, as well as work-flow infor-
mation and organization-related specific matchmaking parameters. Finally, the
data management module hides to other sub-components the connection with
external sources (including Facebook, LinkedIn, Google, Twitter, etc.) and on-
tologies, providing clients the required data according to a uniform and consistent
schema.
In order to provide services required by the end users, the matchmaking
service exposes interfaces for dynamically editing profiles in relation with the
different steps of the recruiting process and, according to stored information, for
performing match-making between subsidiary users and required activities. In
addition, in order to properly provide dynamic profiling capabilities, the match-
making service could possibly require temporal reasoning capabilities linked, for
instance, to organizational specific constraints, offered by the personalization
over time module. Specifically, the personalization over time module provides
capabilities for the management of temporal workflows and internal organization
processes as well as agenda and temporal notification services (e.g., reminders).
Finally, the personalization over time service could possibly (and autonomously)
produce requests to the matchmaking service resulting in changes to the user
profiles and consequent notifications to the interested users.
In other words, the matchmaking service and the personalization over time
module work in cooperation in order to provide dynamic profiling capabilities.
From their collaboration derives the temporal extension required by the match-
making services for guaranteeing the vision of “matchmaking as a process” and,
consequently, allowing the reproduction of the complex scenarios such as those
described in Section 4.
4 Personalized Interaction over Time (PIT)
The demand for more complex services, compared to a simple matchmaking
mechanism, requiring temporal reasoning aspects, has motivated the introduc-
tion of the Personalized Interaction over Time (PIT) module. Specifically, the
PIT module is responsible for providing personalized services, modulated with
respect to temporal aspects and user models. As an example, the PIT mod-
ule can be exploited for sending messages (e.g., alerts, reminders, suggestions,
etc.) to the different SpONSOR users at proper time. The content of such mes-
sages might be context dependent, offering customized services that take into
account both aspects related to the curriculum vitae of the users as well as
their psychological status. To this purpose, it is worth highlighting the fact that
the SpONSOR platform is targeted to aged users which, although might have a
great experience in performing a particular task, resulting from a life spent to
do the same activities, might be tired of continuing to carry it. To this end, it is
important to keep into account psychological aspects, aiming at optimizing the
overall well-being of the SpONSOR users.
We have built PIT services making use of a particular type of automated plan-
ning called timeline-based (for further details, refer to [6]). This kind of planning
allows the modeling of complex domains which require the use of both temporal
reasoning and scheduling features. For this reason, timeline-based planning has
been an obvious choice.
In essence, timeline-based planning approaches planning by modeling the
problem by means of a set of relevant features of the domain which need to
be controlled to obtain a desired temporal behavior. Timelines model entities
whose properties may vary in time and which represent one or more physical
(or logical) subsystems which are relevant to a given planning context. The
planner/scheduler plays the role of the controller for these entities, and reasons
in terms of constraints that bound their internal evolutions and the desired
properties of the generated behaviors.
In timeline-based planning, the main data structure is the timeline which,
in generic terms, is a function of time over a finite domain. Events on timelines
are called tokens and are represented by temporally scoped first-order predicates
(i.e., predicates endowed with extra arguments belonging to the Time domain
T, either real or discrete). From a formal point of view, we have that
Definition 1. a token is an expression of the form:
n (x0 , . . . , xk ) @ [s, e, δ]
where n is a predicate name, x0 , . . . , xk are constants, numeric variables or object
variables, s and e are temporal variables belonging to T such that s ≤ e and δ is
a numeric variable such that δ = e − s. A token n (x0 , . . . , xk ) @ [s, e, δ] asserts
that ∀t such that s ≤ t ≤ e, the relation n (x0 , . . . , xk ) holds at the time t.
Similar to what has been done in [4], although the context is slightly dif-
ferent, we can use tokens for representing information that must be commu-
nicated to the users at proper time. For example, by introducing a token like
message (sender, recipients, content) @ [s, e, δ], we might represent a message
sent by sender to a collection of recipients, having a specific content at time s.
Since we do not need durations for representing messages, we can consider s = e
and, consequently, δ = 0.
It is worth to notice that, in general, tokens’ arguments are variables and,
as such, can be constrained so as to make them assume desired values. We can
use this expedient to place the tokens at desired times, generating a messaging
system capable of supporting the processes of the organizations.
Given the mutable nature of the user dynamic information in time, we have
addressed also the user modeling problem by making use of timelines for each
user representing both their psychological state as well as their dynamic skills
evolving in time. In this regard, we have adapted the work already done in [1,
3] in which timeline-based planning was adopted for modeling users in a crisis
training domain. It is worth noting that the profile of a user is dynamic and,
therefore, might change over time. Once changed, however, the profile of the user
is fully known. This means that although tokens are added dynamically into the
plan, their predicate’s arguments are, actually, constants.
Fig. 6. Temporally extended services provided by the SpONSOR solution.
As an example, consider the process depicted in Figure 6. This example de-
scribes a single subsidiary user, depicted at top, associated to an organization
having a single main user, depicted at the bottom. In this example, for sake
of space, we have packed all user profile timelines in a single timeline and we
have neglected the state of the main user in order to make the example more
accessible. A set of stimuli, represented through tokens, are planned in time,
depicted on the abscissas, so as to be sent to the different users involved. For
example, the content of the first message in time (i.e., “Bob, your volunteering
activities have been greatly appreciated”) constitutes a positive reinforcement
for the activity currently performed by the subsidiary user. Roughly speaking,
the aim of this message is to improve the user’s psychological state. At planned
time, the message is sent to the subsidiary user which might find it in his/her
mail in-box. Despite the positive reinforcement message, a second message is sent
after a while to the main user notifying him/her about the demotivated state of
the subsidiary user. In addition, the message contains a suggestion about some
possible ways to recover the situation. Following the intervention of the main
user, a cooking course is organized and attended by the subsidiary user which,
by means of the subsidiary user portal, provides feedback about the course to the
system. This feedback is interpreted by the system as an update to the profile
of the user and the overall process is adapted to meet the new state’s needs. As
a consequence, new stimuli are planned providing feedback to the users as, for
example, a message for the main user notifying him/her about the good skills of
the cooking course teacher.
The “rules of behaviour”, required by the planner to react to user stimuli, are
generalized in a concept which we call rule. More in general, rules represent the
tool for describing the causal knowledge in the timeline-based planning. From a
formal point of view,
Definition 2. a rule is a tuple c = (name (c) , R (c)), where:
– name (c) is the master (or reference) predicate and is an expression of the
form n (x0 . . . xk ), where n is a unique predicate symbol with respect to a
timeline (i.e., no two rules in a given timeline can have the same predicate
symbol), and x0 . . . xk are its associated variable symbols.
– R (c) is a requirement, i.e. either a slave (or target) predicate, a constraint
among predicates, a conjunction of requirements or a disjunction of require-
ments.
Rules define causal relations that tokens should comply to in order to be valid.
In other words, every token should be supported by a rule. It is worth underscor-
ing that these rules may often involve predicates defined on different timelines,
thus allowing to synchronize concurrent values on different domain components.
Through the use of disjunctions within such rules, it is possible to build different
temporal evolutions of the messages (involving, possibly, different content). By
exploiting the natural flexibility offered by the timeline-based approach, such
evolutions are adapted (or, sometime, “filtered out”) according to the current
psychological state of the involved users, giving the feeling to the end-users to
follow a custom behavior which takes into account the organization’s specific
process as well as the psychological state of the users. Going back to the pro-
posed example, indeed, it is worth highlighting that the planned messages are a
result of the combination of the current profiles of the users with the predefined
natural evolution of the associated organization specific processes. As an exam-
ple, the suggestion for organizing a cooking course is offered to the main user as
a consequence of the psychological state of the subsidiary user which, thanks to
the application of the rules by the planner, is recognized as interested in dealing
about cooking.
Since the same data-structure (i.e., the timelines) is used for maintaining
an internal representation of the different users involved in the system, in the
following we will provide an explanation with a greater detail of how the users
are modeled within SpONSOR and of how the timeline-based plan, represent-
ing the planned stimuli, is adapted to the dynamically evolving state of the
users. A first example of the ongoing personalization process is depicted in Fig-
ure 7. This example can be temporally located a few weeks after the example
of Figure 6 however, compared to the previous example, the profile of the user
is now further detailed, in order to better explain the idea underlying the dy-
namic personalization. In this example, indeed, the dynamic state of the users is
modeled by means of two timelines, representing respectively the cooking skills
of the subsidiary user and his/her abilities in playing the guitar. As a result of
having attended the cooking class, the cooking skills of the subsidiary user are
updated. This update is performed by adding new tokens to the related timelines
which in turn, by applying associated rules, result in an update of the planned
stimuli. As a consequence, a cooking activity is assigned to the subsidiary user
Fig. 7. A first example of ongoing process personalization.
and a feedback request is planned after some months, in order to monitor user
performance.
With the passing of time, the user continues to perform the cooking activity
till the planned request for feedback is reached. At this time, the feedback request
is sent to the subsidiary user which might find it in his/her mail in-box. When the
request is checked by the subsidiary user it is visualized as a five-level Likert scale.
In the example of Figure 8, the user answers with a value of two. The system
notices that the user did not perfectly adapted to the new ongoing activity. As
a consequence, a reinforcement message is planned for the subsidiary user, in
order to encourage him/her in continuing to perform the new activity. At the
same time, a new message is planned for the main user notifying him/her about
the poor performance of the cooking activity, with the objective of stimulating
an human intervention aimed at affecting positively the user status.
Fig. 8. A second example of ongoing process personalization.
5 Conclusions
Concluding, the paper has given an overview of the SpONSOR project. The
project activities have produced an early prototype during the second year of
activities that integrated several of the proposed platform modules. Currently
a more robust release, which will be used to test specific pilot cases so as to
perform validation with real users, is under development.
We have focused our attention on the general schema followed by the ar-
chitecture that reconciles organizations and senior workers/volunteers. From the
point of view of the organizations, the key issue is to serve customers (like people
in need) as better as possible, but is is also important to manage the workforce
according to their capabilities, so as that people is doing the right thing for them
in a specific time frame of their life (being late occupation or retirement). A spe-
cial attention is given to support the difficult and dynamic business of managing
volunteers, an area where maintaining people at work is particularly challenging.
It is worth underscoring how the specific aspect we have described here (i.e.,
the issue of personalized interaction) is motivated by a specific user requirement
that emerged from several of the explorations done in focus groups: the sense
of being abandoned that several people feel when looking for a job at an older
age. Hence, the direction we have pursued is strongly motivated by having the
workers “choose the right thing” and “never feel alone”, but rather “feel helped
in adapting to reality by pursuing new capabilities”. The PIT module is an
example of AI-based subsystem that contributes new system capabilities. Such
demonstration paves the way to a wider use of dialogue-based techniques to
strengthen the sense of engagement of and personalization of interaction with
different users.
References
1. Cesta, A., Cortellessa, G., Benedictis, R.D.: Training for Crisis Decision Making
- An Approach Based on Plan Adaptation. Knowledge-Based Systems 58, 98–112
(2014)
2. CompetenceSeniors: Competenceseniors (2016), http://www.competence-
seniors.com
3. Cortellessa, G., D’Amico, R., Pagani, M., Tiberio, L., De Benedictis, R., Bernardi,
G., Cesta, A.: Modeling Users of Crisis Training Environments by Integrating Psy-
chological and Physiological Data, pp. 79–88. Springer Berlin Heidelberg, Berlin,
Heidelberg (2011)
4. De Benedictis, R., Cesta, A., Coraci, L., Cortellessa, G., Orlandini, A.: Adaptive
Reminders in an Ambient Assisted Living Environment, pp. 219–230. Springer
International Publishing, Cham (2015)
5. Give&Take: Give&take (2016), http://givetake.eu/
6. Muscettola, N.: HSTS: Integrating Planning and Scheduling. In: Zweben, M. and
Fox, M.S. (ed.) Intelligent Scheduling. Morgan Kauffmann (1994)
7. Quintonic: Quintonic (2016), https://www.quintonic.fr/
8. Retiredbutable: Retiredbutable (2016), https://retiredbutable.com/
9. Rossel, P., Doyle, J., Turki, S., Nicolas, D., Cesta, A., Cortellessa, G., Benedictis,
R.D., Fracasso, F.: D2.1(b) - User requirements knowledge base inventory. Tech.
rep. (2015)
10. SeniorsAvotreService: Seniors a votre service (2016),
http://seniorsavotreservice.com/
11. TimeRepublik: Timerepublik (2016), https://timerepublik.com/
12. Vicker: Vicker (2016), https://www.vicker.org/