Smart Conversational Agents for Reminiscence

by   Svetlana Nikitina, et al.
Università di Trento

In this paper we describe the requirements and early system design for a smart conversational agent that can assist older adults in the reminiscence process. The practice of reminiscence has well documented benefits for the mental, social and emotional well-being of older adults. However, the technology support, valuable in many different ways, is still limited in terms of need of co-located human presence, data collection capabilities, and ability to support sustained engagement, thus missing key opportunities to improve care practices, facilitate social interactions, and bring the reminiscence practice closer to those with less opportunities to engage in co-located sessions with a (trained) companion. We discuss conversational agents and cognitive services as the platform for building the next generation of reminiscence applications, and introduce the concept application of a smart reminiscence agent.


page 3

page 4


Exploring Current User Web Search Behaviours in Analysis Tasks to be Supported in Conversational Search

Conversational search presents opportunities to support users in their s...

Is your chatbot GDPR compliant? Open issues in agent design

Conversational agents open the world to new opportunities for human inte...

Designing for Co-located and Virtual Social Interactions in Residential Care

In this paper we explore the feasibility and design challenges in suppor...

Migratable AI

Conversational AI agents are proliferating, embodying a range of devices...

Diplomat: A conversational agent framework for goal-oriented group discussion

Recent work in human-computer interaction has explored the use of conver...

Bringing Cognitive Augmentation to Web Browsing Accessibility

In this paper we explore the opportunities brought by cognitive augmenta...

Alexa as an Active Listener: How Backchanneling Can Elicit Self-Disclosure and Promote User Experience

Active listening is a well-known skill applied in human communication to...

1. Background and Motivation

In this paper we define the requirements and early design of a conversational agent that supports engaging and effective conversation about a person’s life and memories. Specifically, we discuss the case of personal and social reminiscence sessions by older adults.

Reminiscence is the process of collecting and recalling past memories through pictures, stories and other mementos (Webster and Gould, 2007). The practice of reminiscence, and more generally, reminiscence therapy and life review, have well documented benefits on emotional and mental wellbeing (Bender et al., 1998; Subramaniam and Woods, 2012), interpersonal relationships and interactions (Huldtgren et al., 2015; Allen, 2009), and preserving personal identity (Crete-Nishihata et al., 2012; Fivush et al., 2011). These benefits, along with the potential of reminiscence to transfer knowledge, stimulate conversations, reduce boredom, and maintain intimacy (Webster, 1993) makes it a very desirable practice, especially for older adults.

Research on technology-mediated reminiscence have taught us valuable lessons on how to facilitate usage by older adults (West et al., 2007; Piper et al., 2013), collect memories and support storytelling (Lee and Hsu, 2016; Lee et al., 2014), stimulate cognitive functions (Subramaniam and Woods, 2016; Lee et al., 2014), and support conversations (Kim and Zimmerman, 2006; Astell et al., 2009). However, these tools are very limited when it comes to creating opportunities for social interactions and for supporting social integration, something that ample and concordant research tells us it is a determinant of wellbeing at any age (Hartup and Stevens, 1999). In our previous work, we addressed some of these limitations in a reminiscence-based social interaction tool (Ibarra et al., 2017) that aimed at stimulating conversations and creating bonds among older adults in residential care. However, our previous efforts and those of the community have reached a limit in terms of what traditional reminiscence technology can do to support older adults.

As we will see, traditional reminiscence technology i) strongly relies on co-located human presence for assisting in the reminiscence process, engaging in conversations or joining social reminiscence sessions; this often requires dedicated personnel, the presence of family member and friends, which greatly limit the practice and potential benefits of reminiscence, especially for those with less access to social contacts; ii) do very little to actively guide reminiscence sessions in a way that is effective and engaging, requiring users (the participant or a guide) to recall important aspects of the participant’s life, to rely only on intentional triggers, or to follow predefined templates; thus missing the opportunity to reflect interests and previous stories to make sessions more fun and satisfying, and iii) memory collection and digital storytelling is mostly limited to archiving and browsing, missing the potential benefits of building a rich profile of users, e.g., to improve care practices (Thorgrimsdottir and Bjornsdottir, 2016) or facilitate social interactions (Sanchiz et al., 2016; Caforio et al., 2017), for example by identifying people with common values, or simply by making one’s own family, and especially grandchildren, aware of the rich and often interesting life their grandparents have lived and are living.

Conversational agents and the ever growing number of cognitive services offer a unique platform to engineer the next generation of smart reminiscence systems that are not only more personal and engaging, but that can make those who have less opportunities for co-located interactions enjoy the benefits of the reminiscence practice. Engineering such a system requires insights into the reminiscence process, the development conversational models for reminiscence, the underlying cognitive services and system design.

In this short paper we describe the requirements, models and a system design for such a conversational agent. From a technological standpoint we investigate the design of a chatbot that can flexibly drive a person (or a family) through a picture-driven storytelling session. The goal of the bot is to be able to sustain a conversation about a person’s life, while harvesting useful information and while keeping the session interesting and engaging.

2. Related work

Exploring the use of technology to support the reminiscence practice by older adults has been subject of plenty of research (see (Lazar et al., 2014) for a review on the subject). The work on this area can be summarised in efforts to facilitate usage by older adults (West et al., 2007; Piper et al., 2013), collect memories and support storytelling (Lee and Hsu, 2016; Lee et al., 2014), stimulate cognitive functions (Subramaniam and Woods, 2016; Lee et al., 2014), and support conversations (Kim and Zimmerman, 2006; Astell et al., 2009; Ibarra et al., 2017). Despite the valuable contributions, most of these works still rely either on the ability of participants to drive the personal reminiscence sessions or the presence of assistance (Astell et al., 2009; Ibarra et al., 2017). Support for social interactions is also mostly focused on co-located settings (Uriu et al., 2009; Astell et al., 2009) or offered in the form of online sharing (West et al., 2007). The efforts in improving data collection are also largely focused on collaboration (Lee and Hsu, 2016) and optimising the interaction design (Lee et al., 2014), with little to no intelligence. This represents not only a missed opportunity but also a constraint for those people with less opportunities for face to face interactions.

Another relevant type of technology is that of digital companions and relational agents. The feasibility of social companions for older adults, exploring a wide ranging set of activities was evaluated in (Vardoulakis et al., 2012), using the Wizard of Oz technique, with a researcher selecting the agent responses for a period of a week. A interesting finding from this work is identifying ”storytelling” as the type of interaction elders spent more time with the agent. In a follow up study (Ring et al., 2013), the authors explored the effects of proactive vs passive behavior in conversational agents using a scripted conversation system, identifying that agents capable of initiating interactions (proactive) were more effective in addressing loneliness.

Similarly, a commercial digital companion driven by a human operator was also evaluated in (Demiris et al., 2016). The companion, an always on digital pet with conversational skills, was positively assessed by the participants, especially in the ability to show pictures (sent by a caregiver) and ”remember” things about the older adults (e.g., a scripted reminder that included the birthday of the participant).

Another thread of work focused on identifying requirements as well as older adults attitudes toward autonomous virtual agents (not driven by a human operator). Tsiourti et al. (Tsiourti et al., 2014) explored the attitudes and perception of older adults towards different aspects of virtual companions during the design phase, pointing to the ”usefulness” and ”social intelligence” a determinant factors for the success. The authors later conducted a exploratory study in 20 homes (Tsiourti et al., 2016) with a virtual companion capable of assisting every day tasks, identifying several challenges, especially in the mismatch between expectations of the older adult in what regards the conversational capabilities of the virtual companion.

What the above tells us is that the use of conversational agents is feasible, and that reminiscence is an untapped and desirable application to explore. We also see that the success in previous studies was evident when agents displayed qualities attributed to humans and when proved useful to the task at hand – not surprisingly this was the case in agents operated by humans. All things considered, the use of cognitive services in reminiscence technology still remains in its infancy. This clearly motivates the need for research and development of intelligent conversational agents can assist the reminiscence process.

3. Requirements

To summarize the extensive theoretical work on the subject into a practical reference framework for technology-mediated reminiscence, we take the conceptual reminiscence model by Webster et al. (Webster et al., 2010) and instantiate it in a process that captures the most salient technology roles from the literature.

Figure 1. Technology-mediated reminiscence process

This summary, illustrated in Figure 1, shows how the process is initiated by a trigger, usually a memento in the form of pictures, videos or music, or specific questions. The trigger results in memories being primed, which can be then shared (virtually or face to face) via storytelling, or used for personal reflection. These two visibilities of memories define two distinctive modes known as interpersonal and intrapersonal. Memories and related stories resulting from this elicitation process can be digitally collected. If memories are used for interpersonal reminiscence in a storytelling session, then companions can engage in conversations, provide feedback, and further inquire about the aspects of the stories told by the person reminiscing. The role of the companion can vary depending on the degree of participation: companions can take a leading role in guiding the reminiscence session, co-participate in the case of social reminiscence, or take a more passive role as recipient of the storytelling. Regulating the entire process we have the context and moderators. These are intrinsic properties of the person (and reminiscence session) that can influence the reminiscence process and outcome, such as aspects of personality, age, gender, and relationship with the reminiscing companion. Finally, the practice of reminiscence is related to very specific functions, among which we highlight: i) boredom reduction, referring to the potential to create joyful experiences, ii) conversation, to the opportunity to create a dialog, iii) and teach / inform, to the opportunity to transfer stories and knowledge.

Figure 2. Example conversations illustrating chatbot skills

Informed by the above process, and our own experience designing and evaluating reminiscence technology for nursing home residents (on average, 80+ older adults that require assistance to perform activities of daily living) (Ibarra et al., 2017), we identified the following main requirements:

(Hide=100, Hang=true,Style*=• ) & Eliciting memories, in a way that engage participants in storytelling, in conversations that are sensitive to (changing) interests of the user, the topic, sentiment, and context.

& Archiving life stories and memories, to support personal reflection, knowledge transfer or even virtual storytelling.

& Building representations of the life of a person, not only to facilitate browsing, but to serve the growing need for historical information about older adults, e.g., to improve care practices (Thorgrimsdottir and Bjornsdottir, 2016), facilitate social interactions (Sanchiz et al., 2016; Caforio et al., 2017), and build better context and moderators to improve the reminiscence process.

& Navigating interests and passions, as to engage users in conversations where they can have fun by navigating their passions (e.g., music, video, movies). This also helps elicit stories, and more importantly collecting preferences.

& Facilitating (re-)connections, by identifying potential companions and creating opportunities for social reminiscence and interactions. Interactions are facilitated to avoid stress in initiating conversations.

In addressing the above requirement, technology support will have to take a more proactive role in serving the reminiscence functions and needs of older adults.

4. Concept of a Social Reminiscence Service

We propose the design of an intelligent reminiscence chatbot that can play the role of companion in reminiscence sessions with older adults. The chatbot acts as the guide of reminiscence session in a way that is engaging and fun, facilitating the tasks of collecting and organising memories and stories, while finding opportunities to connect people and stimulate social interactions. We exemplify these capabilities in Figure 2.

The above requires the bot to adopt qualities that go beyond traditional scripted dialog systems (Ring et al., 2013; Demiris et al., 2016), to qualities traditionally attributed to humans: i) understand triggers, as to formulate meaningful questions in the same way someone looking at a picture would use that information to formulate related questions, ii) learn about the user, and use this knowledge to elaborate more natural conversations and improve memory elicitation, iii) sense where the conversation is going and guide the sessions accordingly, balancing the objective of collecting specific information (e.g., details on what a picture represents) with the opportunity to follow the person’s flow of thinking and be directed by it, even if it takes us away from the picture we want to collect information on.

Note that while use chatbot and text-based interactions in our example, these are only to illustrate the capabilities of the system, and to drive the concept development. Designing interactive technologies for older adults is a challenging task that requires specific attention to their abilities and limitations (Nurgalieva et al., 2017; Hawthorn, 2000). Thus, as part of part of our ongoing efforts we are investigating designs of conversational interfaces that can support individuals of potentially different interaction abilities.

4.1. Conceptual model

The chatbot operates over a model comprising three families of concepts: the life model, the reminiscence model, and the conversation model.

The life model captures the information about the life of the person and the properties that describe them. This includes i) life events, which are well defined moments in the life of a person, such as birthdays, wedding, graduation, among others; ii) habits and skills, describing the person capabilities and routines, such as knitting or reading; iii) values, describing the person beliefs and ideals, such as religion, political views and social causes; iv) preferences, capturing passions and hobbies, such as music, sports, gardening; and v) relationships, referring to familial and non familial connections of the person. All this information is defined from a life course perspective, as they can refer to specific points in life. This temporal dimension cannot be measured in terms of specific dates, but in more flexible terms such periods of life.

Figure 3. Example of bot execution

Most of the concepts that are specific to the reminiscence practice were introduced in Figure 1. The reminiscence model defines: i) mementos, digital artefacts (private or public) that can be used in the reminiscence session to trigger memories, such as pictures, videos or songs; ii) tags, annotations that describe the mementos, such as people, places, dates related to a picture; iii) life stories, anecdotes and memories that result from the reminiscence and storytelling session, possibly in response to a memento. Complementing this model there is the concept of the participant of the reminiscence session, which is captured by the life model, and the companion, which we define as part of the conversation model.

The interactions between the user and the agent define conversations. We organise conversations in sessions, which give scope to the interactions during a reminiscence session. An interaction between the user and agent is usually referred to as turn, and characterised by an input and a response. The set of input and responses define a context, which provide implicit knowledge to the current and future turns. In the reminiscence bot, the memento under discussion, the questions recently asked, and the information provided by the user are key components of the immediate context. At a more global scale, the knowledge about the user (see life model) and the history of conversations define the macro context. The actions define the capabilities of our reminiscence bot, such as eliciting information, showing understanding (reactions), bringing up interesting multimedia content, and connecting people. Serving these actions we have a database of questions related to specific user knowledge entities (to elicit information), a database of mementos (public and private, to bring up content), and a continuously updated database of user similarity and user clusters (to find potential connections). Orchestrating the decisions of what action to take and how, there is a set of rules, and metrics that facilitate the self-assessment of the bot.

4.2. The chatbot at work

The conversation model together with the reminiscence and life model work with the cognitive services as shown Figure 3.

The picture used as trigger is first passed through a information extraction service such as Google Cloud Vision that automatically detects its salient features (e.g., color palette, people, artefacts, landscape, emotions). In the example, the service detected a city ”skyline” as the most important feature. Being the first interaction, and the context limited to the fact that a picture of a city was provided, the rule system decides to elicit basic information about the picture, embedding this information to elicit the related knowledge (place: Nice picture! It looks like a big city. Where was it taken?). The user reply is then processed by an NLP service such as DialogFlow or Google Cloud Natural Language to extract the related entities. Metrics are also run to compute engagement and effectiveness of the action and question. The relevant tag (place: Chicago) is then associated to the picture.

In deciding the next action the system consider again the context, which is now comprised by a picture of Chicago, and information about the last exchange. This context combined with knowledge about the user (user: born in Trento) and metrics assessing a positive engagement, makes the following information elicitation rule to kick in (”User was born in Trento, and the picture is from Chicago: the user might have lived or visited Chicago”), requiring the connection of the user with the place to be inquired. This rule combined with the entity in context is paraphrased in a question as (”That’s far away from Trento!. Were you visiting Chicago?). As before, the subsequent user responses are used to build the new knowledge that (The user had lived in Chicago in the 80s). Following the conversation, the interest of the user or the gain in eliciting information on the picture might decrease, motivating the bot to bring up new content related to the context to elicit new information, or suggest relevant connections.

It is important to note that the rules and context switching are not predefined, but can be trained using crowdsourcing and machine learning techniques, by crowds of experts and non experts depending on needs of the specific target population.

Mockups of the chatbot can be seen at

4.3. Metrics and measures of success

We briefly describe metrics that might shape the chatbot training and ability to self-assess:

(Hide=100, Hang=true,Style*=• )

& Engagement describes the level of involvement and interest of a person in the conversation. Capturing engagement is important as it has shown to be associated with longer conversations and higher sentiment score (Cervone et al., [n. d.]). Different engagement metrics have been proposed for conversational agents, such as: i) conversation length, in minutes and/or amount of dialog turns (Fitzpatrick et al., 2017; Cervone et al., [n. d.]); ii) total number of interactions with the bot over a session (Fitzpatrick et al., 2017); iii) subjective user engagement ratings (Cervone et al., [n. d.]); and iv) cumulative average sentiment score, to see the user sentiment towards a topic (Cervone et al., [n. d.]).

& Task completion captures the effectiveness achieved in the execution of a specific action. The metrics that have been recently proposed for the task-based evaluation are: i) task completion rate and ii) completion time, which can be measured as a number of dialog turns the user talks to bot to complete a task (Huang et al., 2015).

& Conversation quality describes how much consistent the conversation is and the ability of the bot to correctly remember things the user said on the previous turns of conversation. For evaluation of conversation quality several metrics has been proposed: i) conversational consistency, ii ) memory of past events, and iii) response speed (Lasecki et al., 2013).

& Human-like communication is proposed as a measure of the conversation quality and naturalness, where conversation logs are analysed to see if user had conversed with the bot as if with a human, e.g. using greetings or showing ”proactivity” (Kopp et al., 2005).

On the other hand, measuring the actual benefits of the such a tool on the mental and social well-being of individuals would require the design of RCTs and the use of validated instruments.

5. Conclusion and Future work

It is clear that conversational agents have the potential to overcome the limitations of traditional reminiscence technology, but further research is needed in order to go beyond scripted or human-operated systems. We did a step forward in this direction and proposed the use of well defined models (life, reminiscence and conversation models) in combination with cognitive services to provide agents with some human qualities.

The use of cognitive services is fundamental in providing bots with the required level of understanding to drive the reminiscence sessions. Recent developments in automatic caption and question generation (Mostafazadeh et al., 2016; Tran et al., 2016) show promising results, and inspire our future work in building more intelligent reminiscence applications.

As part of our ongoing and future work, we are investigating strategies for training the reminiscence bot with the help of the crowd.

This work has received funding from the EU Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement. It was also supported by the project ”Evaluation and enhancement of social, economic and emotional wellbeing of older adults” under the agreement No.14.Z50.31.0029, Tomsk Polytechnic University.


  • (1)
  • Allen (2009) Rebecca S Allen. 2009. The Legacy Project intervention to enhance meaningful family interactions: Case examples. Clinical Gerontologist 32, 2 (2009), 164–176.
  • Astell et al. (2009) Arlene Astell, Norman Alm, Gary Gowans, Maggie Ellis, Richard Dye, Jim Campbell, and Phil Vaughan. 2009. Working with people with dementia to develop technology: The CIRCA and Living in the Moment projects. PSIGE Newsletter 64 (2009).
  • Bender et al. (1998) Mike Bender, Paulette Bauckham, and Andrew Norris. 1998. The therapeutic purposes of reminiscence. Sage.
  • Caforio et al. (2017) Valentina Caforio, Marcos Baez, and Fabio Casati. 2017. Viability of Magazines for Stimulating Social Interactions in Nursing Homes. In IFIP Conference on Human-Computer Interaction. Springer, 72–81.
  • Cervone et al. ([n. d.]) A Cervone, G Tortoreto, S Mezza, E Gambi, and G Riccardi. [n. d.]. Roving Mind: a balancing act between open–domain and engaging dialogue systems. ([n. d.]).
  • Crete-Nishihata et al. (2012) Masashi Crete-Nishihata, Ronald M Baecker, Michael Massimi, Deborah Ptak, Rachelle Campigotto, Liam D Kaufman, Adam M Brickman, Gary R Turner, Joshua R Steinerman, and Sandra E Black. 2012. Reconstructing the past: personal memory technologies are not just personal and not just for memory. Human–Computer Interaction 27, 1-2 (2012), 92–123.
  • Demiris et al. (2016) George Demiris, Hilaire J Thompson, Amanda Lazar, and Shih-Yin Lin. 2016. Evaluation of a digital companion for older adults with mild cognitive impairment. In AMIA Annual Symposium Proceedings, Vol. 2016. American Medical Informatics Association, 496.
  • Fitzpatrick et al. (2017) Kathleen Kara Fitzpatrick, Alison Darcy, and Molly Vierhile. 2017. Delivering Cognitive Behavior Therapy to Young Adults With Symptoms of Depression and Anxiety Using a Fully Automated Conversational Agent (Woebot): A Randomized Controlled Trial. JMIR Mental Health 4, 2 (2017), e19.
  • Fivush et al. (2011) Robyn Fivush, Tilmann Habermas, Theodore EA Waters, and Widaad Zaman. 2011. The making of autobiographical memory: Intersections of culture, narratives and identity. International Journal of Psychology 46, 5 (2011), 321–345.
  • Hartup and Stevens (1999) Willard W Hartup and Nan Stevens. 1999. Friendships and adaptation across the life span. Current directions in psychological science 8, 3 (1999), 76–79.
  • Hawthorn (2000) Dan Hawthorn. 2000. Possible implications of aging for interface designers. Interacting with computers 12, 5 (2000), 507–528.
  • Huang et al. (2015) Ting-Hao Kenneth Huang, Walter S Lasecki, and Jeffrey P Bigham. 2015. Guardian: A crowd-powered spoken dialog system for web apis. In Third AAAI conference on human computation and crowdsourcing.
  • Huldtgren et al. (2015) Alina Huldtgren, Fabian Mertl, Anja Vormann, and Chris Geiger. 2015. Probing the potential of multimedia artefacts to support communication of people with dementia. In Human-Computer Interaction. Springer, 71–79.
  • Ibarra et al. (2017) Francisco Ibarra, Marcos Baez, Francesca Fiore, and Fabio Casati. 2017. Stimulating Conversations in Residential Care Through Technology-Mediated Reminiscence. In IFIP Conference on Human-Computer Interaction. Springer, 62–71.
  • Kim and Zimmerman (2006) Jeong Kim and John Zimmerman. 2006. Cherish: smart digital photo frames for sharing social narratives at home. In CHI’06 Extended Abstracts on Human Factors in Computing Systems. ACM, 953–958.
  • Kopp et al. (2005) Stefan Kopp, Lars Gesellensetter, Nicole C Krämer, and Ipke Wachsmuth. 2005. A conversational agent as museum guide–design and evaluation of a real-world application. In International Workshop on Intelligent Virtual Agents. Springer, 329–343.
  • Lasecki et al. (2013) Walter S Lasecki, Rachel Wesley, Jeffrey Nichols, Anand Kulkarni, James F Allen, and Jeffrey P Bigham. 2013. Chorus: a crowd-powered conversational assistant. In Proceedings of the 26th annual ACM symposium on User interface software and technology. ACM, 151–162.
  • Lazar et al. (2014) Amanda Lazar, Hilaire Thompson, and George Demiris. 2014. A systematic review of the use of technology for reminiscence therapy. Health education & behavior 41, 1_suppl (2014), 51S–61S.
  • Lee et al. (2014) Hung-Chi Lee, Ya Fang Cheng, Szu Yang Cho, Hsien-Hui Tang, Jane Hsu, and Chien-Hsiung Chen. 2014. Picgo: designing reminiscence and storytelling for the elderly with photo annotation. In Proceedings of the 2014 companion publication on Designing interactive systems. ACM, 9–12.
  • Lee and Hsu (2016) Hung-Chi Lee and Jane Yung-jen Hsu. 2016. PicMemory: Enriching Intergenerational Family Interaction and Memory Collection. In Proceedings of the 2016 CHI Conference Extended Abstracts on Human Factors in Computing Systems. ACM, 3715–3718.
  • Mostafazadeh et al. (2016) Nasrin Mostafazadeh, Ishan Misra, Jacob Devlin, Margaret Mitchell, Xiaodong He, and Lucy Vanderwende. 2016. Generating natural questions about an image. arXiv preprint arXiv:1603.06059 (2016).
  • Nurgalieva et al. (2017) Leysan Nurgalieva, Juan Jose Jara Laconich, Marcos Baez, Fabio Casati, and Maurizio Marchese. 2017. Designing for older adults: review of touchscreen design guidelines. arXiv preprint arXiv:1703.06317 (2017).
  • Piper et al. (2013) Anne Marie Piper, Nadir Weibel, and James Hollan. 2013. Audio-enhanced paper photos: encouraging social interaction at age 105. In Proceedings of the 2013 conference on Computer supported cooperative work. ACM, 215–224.
  • Ring et al. (2013) Lazlo Ring, Barbara Barry, Kathleen Totzke, and Timothy Bickmore. 2013. Addressing loneliness and isolation in older adults: Proactive affective agents provide better support. In Affective Computing and Intelligent Interaction (ACII), 2013 Humaine Association Conference on. IEEE, 61–66.
  • Sanchiz et al. (2016) Emanuel Sanchiz, Francisco Ibarra, Svetlana Nikitina, Marcos Báez, and Fabio Casati. 2016. What makes people bond?: A study on social interactions and common life points on Facebook. In Collaboration Technologies and Systems (CTS), 2016 International Conference on. IEEE, 26–30.
  • Subramaniam and Woods (2012) Ponnusamy Subramaniam and Bob Woods. 2012. The impact of individual reminiscence therapy for people with dementia: systematic review. Expert Review of Neurotherapeutics 12, 5 (2012), 545–555.
  • Subramaniam and Woods (2016) Ponnusamy Subramaniam and Bob Woods. 2016. Digital life storybooks for people with dementia living in care homes: an evaluation. Clinical interventions in aging 11 (2016), 1263.
  • Thorgrimsdottir and Bjornsdottir (2016) Sigrun Huld Thorgrimsdottir and Kristin Bjornsdottir. 2016. Reminiscence work with older people: the development of a historical reminiscence tool. International journal of older people nursing 11, 1 (2016), 70–79.
  • Tran et al. (2016) Kenneth Tran, Xiaodong He, Lei Zhang, and Jian Sun. 2016. Rich image captioning in the wild. In Computer Vision and Pattern Recognition Workshops (CVPRW), 2016 IEEE Conference on. IEEE, 434–441.
  • Tsiourti et al. (2014) Christiana Tsiourti, Emilie Joly, Cindy Wings, Maher Ben Moussa, and Katarzyna Wac. 2014. Virtual assistive companions for older adults: qualitative field study and design implications. In Proceedings of the 8th International Conference on Pervasive Computing Technologies for Healthcare. ICST (Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering), 57–64.
  • Tsiourti et al. (2016) Christiana Tsiourti, Maher Ben Moussa, João Quintas, Ben Loke, Inge Jochem, Joana Albuquerque Lopes, and Dimitri Konstantas. 2016. A virtual assistive companion for older adults: design implications for a real-world application. In Proceedings of SAI Intelligent Systems Conference. Springer, 1014–1033.
  • Uriu et al. (2009) Daisuke Uriu, Naruhiko Shiratori, Satoru Hashimoto, Shuichi Ishibashi, and Naohito Okude. 2009. CaraClock: an interactive photo viewer designed for family memories. In CHI’09 Extended Abstracts on Human Factors in Computing Systems. ACM, 3205–3210.
  • Vardoulakis et al. (2012) Laura Pfeifer Vardoulakis, Lazlo Ring, Barbara Barry, Candace L Sidner, and Timothy Bickmore. 2012. Designing relational agents as long term social companions for older adults. In International Conference on Intelligent Virtual Agents. Springer, 289–302.
  • Webster (1993) Jeffrey Dean Webster. 1993. Construction and validation of the Reminiscence Functions Scale. Journal of Gerontology 48, 5 (1993), P256–P262.
  • Webster et al. (2010) Jeffrey Dean Webster, Ernst T Bohlmeijer, and Gerben J Westerhof. 2010. Mapping the future of reminiscence: A conceptual guide for research and practice. Research on Aging 32, 4 (2010), 527–564.
  • Webster and Gould (2007) Jeffrey Dean Webster and Odette Gould. 2007. Reminiscence and vivid personal memories across adulthood. The International Journal of Aging and Human Development 64, 2 (2007), 149–170.
  • West et al. (2007) David West, Aaron Quigley, and Judy Kay. 2007. MEMENTO: a digital-physical scrapbook for memory sharing. Personal and Ubiquitous Computing 11, 4 (2007), 313–328.