Reveal: Investigating Proactive Location-Based Reminiscing with Personal Digital Photo Repositories


– So thank you, I’m Dave
McGookin, from Ferratum, but this work was done when I was at Aalto University in Finland. You don’t need me to tell you,
though I’m going to anyway, that smartphones have changed the nature of how we do photography. Rather than just taking images of perhaps when you’re on holiday to a nice place or you really need to,
you know you’re going to want to take pictures. Smartphones have been able to allow us to take pictures almost anywhere and that means that we’ve
increased the number of photos that we take of some
everyday and mundane things like in the supermarket. That’s led to an explosion
of the number of pictures that we actually have
and if you look through your smartphone I’m sure you can find you’ve got a lot of them. And that’s led to manufacturers
like Apple and Google, there are no new manufacturers these days, building software
technologies and toolkits to cloud back the photo library, so no more do you take
pictures on an SD card and you transfer them,
you take the picture and it automatically appears in the cloud and you just end up with
this larger and larger kind of digital repository of images. And the question is, of course, what do we actually do with those images? Well, obviously one of the main things is that we share them on social media, maybe on Instagram or Twitter or Facebook, but when you share things on social media we tend to try to present
an aspect of ourselves which is positive towards the audience that we are presenting. We’re kind of tailoring
content to the viewer we wish to project of ourselves. And this in many cases,
with Instagram can be taken really quite far, with people
going to specific places or getting professional
photographers to take their images so that they look good on Instagram. But it does mean that in
everyday life there are images that we probably don’t share on Instagram and therefore we end up with
these large amounts of images increasingly which never ever look at all, maybe you’ve got images on your
phone you’ve never looked at since you’ve taken them. So the question is then,
what can we do with that? And one core value of older
images that we’ve taken is an ability to reminisce over them, so to help stimulate and help
consider forgotten memories that maybe have fallen
into the back of our mind. Now, XCI has, over the years,
completed a significant amount of work into reminiscing. It’s found that reminiscing
is valuable at all ages and it’s something that people
wish they did more often. It’s also found that people
need to be prompted to reminisce you tend not to just do
it, you need actually some kind of external prompt. And when giving external prompts, prompts which link now to then in someway or to the period that you’re
going to reminisce over are actually stronger, so
they would be showing you what you were doing a year ago today or two years ago today
are examples of that. Much of this work has
been already embedded into mainstream applications,
so Facebook have on this day, which will pop up previous posts that you made in previous years. Apple, in Apple’s usual not
really sure what we’re doing way you’ll create these sort
of thematic photo galleries which might show stuff that
had been going on a while ago, but they tend not to link
it specifically to dates and then also things like Timehop as well. In all this work at particularly XCI, we really looked at
temporal links between now and the media, so a year ago today. But we’ve also moved from
diary studies by Van Eerde that what prompts people to
reminisce is often location, location is a very important
feature of reminiscing, but for images Whittaker found
that a large number of people couldn’t remember where
they’d taken the image just by looking at the image,
so location’s important, but you don’t really
remember location very well. In addition, because we’ve got these sort of everyday
photographs they happen in sort of everyday environments, therefore, multiple events may overlap. Maybe you went to the same
bar every three months with some other people. Maybe your in city center
many times in a month but there are different things going on. So we wanted to investigate
this role of location for reminiscing over personal photo images and also because we’ve now
got things like Google photos and Apple photos that have
these Cloud back repositories, we can do it over the entire
photo library we’ve created, rather than just those images that you’ve uploaded on social media, which, a lot of existing
academic research, largely because of when it was done, uses. So we don’t have that presentation view that you’re trying to give. In this part I want to look
at three research questions. First of all’s the impact
of a location-based relationship to support when reminiscing over personal digital photo collections. What’s the value of supporting
reminiscing beyond all those images that were previously
uploaded onto social media, so looking at the entire photo
library that was created. And then also how do users fit
this kind of location-based reminiscing into their day-to-day lives, where does it work, where
is it not an important addition to our day. So to do that, we created
a technology called Reveal. Reveal is an iPhone application, ’cause it’s actually really hard to do this on Android at the moment and it uses a standard
location-based system research core, so as the user walks
around in the environment the phone periodically searches
over their photo collection for images taken in the nearby distance. In this case we’re using 150 meter radius around the user’s current location. And I should say we’ll be using
geo-tagged photos in this, we’re not using anything particularly sophisticated over all of the images. If it finds images that
the user hasn’t seen before it sets one of these
to be the primary image and it will then fire a
notification to the user to tell them the image
has been found nearby and prompts them to then
open and look at them. The user can ignore this and walk away in which case the notification’s
cleared after a few meters or they can launch the notification and go in to be able to
see what they’ve found. When they do this is
sort of what they get. Out of all the images in the area, assuming there’s more than one, we set one of the images
to be called primary image and that’s displayed
immediately full screen. If there are more images
they get this small button in the button right where they
can then go to a grid view of all the images in
that area sorted by year. And they can also access a
large scale version of the image or they can flip around
and see any metadata that might be existing on that image. We carried out a study
with Reveal over 14 days, with 14 participants. The criteria for participants was that they had to have an iPhone,
they had to use iCloud photos and they had to have geo-tagging
of their images switched on if they didn’t have geo-tagging
of their images switched on, there wouldn’t be any data
for us to actually look at. And also because we think many
of these everyday photographs are taken in everyday situations, we also asked that all
participants had to have lived in their current
area for about two years and we basically put
this on the city level, so if they’ve lived in the same
city for the last two years. They used the application
as they wanted over 14 days and this is similar to
other reminiscing work but in our case we had
to have a lot more time between the creation of the media and when the reviewing took place. And we also did some meta-analysis
of their photo libraries, so the shorter bit is that
Reveal was largely accessing images that were taken over
the previous two years, given that many of the
images the participants had in their photo libraries
did not have a geo-tag. I’m not going to go into the
quantitative stuff very much, but this is just a little bit
on what actually happened. Over the two weeks, there
was a median number of 69 notifications presented,
that’s about five per day, there were slightly more at the beginning than there were towards the end. A mean media of eight were
responded to over the study, so again the number of
notification presented were quite a bit larger those that were actually responded to. We also had a higher ratio
towards the start of the study. But when people did interact with Reveal, they did so in a deep way. The more images where people
could access the other images that were seen were
accessed about the same number of times as the notifications, so when people did respond they wanted to look at all the photos in that area. I want to go through the results now, just to relatively check
the box of core issues as to why they’ve been
a bit picky with them because there’s quite a bit in here. But the first is a long list of practical implications of location. This is location-driven work, time-driven, so people had to be in the areas
where they got notification and respond to it to get
information from Reveal and what we found was that, obviously, people only respond if
the really have time, if they were going somewhere
or they were meeting somebody or they were in transit,
they tended not to respond to the notification
and those notifications were auto-cleared and
they didn’t see them. But what we did also find
from the notifications that they got is that people
began to develop a photographic landscape of their image
taking, that they became aware quite early on where
they had taken images, so the images they had taken at work and at home were two very
obvious cases that we found. And this acted as a driver
to respond to or not, so if people expected to
find images in an area you got a notification for then
they tended not to respond. On the other hand, if
they were in an area where notifications were not expected or they haven’t thought they’d been, they were actually much more motivated to respond when that happened. Where people got these
notifications was actually and important factor for whether they were responded to or not. When we look then at the
value of these images and the value of looking at the location roughly where they were taken, there were a number of
things that we found. Obviously, these provided
a new perspective to relive old memories,
so this was an example that we had from a
participant who had previously been in a different job and
happened to find himself on the same bus he used to travel on and he found that the photographs that he saw on the bus when
he got the notification actually helped take him
back there and reflect. Also, interestingly, and
probably expected with this kind of application is the way in which the environment changes all the time. So we obviously had
this in cases where the physical environment had changed, maybe a building had been
built or been knocked down and people could see the
difference between when the photograph was taken. But it happened in much
more smaller cases as well. The first example there is someone who had moved into their home,
they’d been taking pictures over the last couple of years of them getting their house turned into their home and they could really see the progress they’d made over time. The ability to use Reveal
meant that all the images were kind of cut down to the
small subset just about area and they could see the
progression over times how the environment they
had created had changed. Also, because we did this in Finland, there’s fairly significant
season variations and we also saw the same kind of thing, we got notifications showing
images in different seasons and people could begin to
see how the environment had changed over and in a sense, this also reflected back on their thoughts on how they took photographs. So this participant here
found that they taken lots of really nice images in the
summer but hadn’t really been doing so much in the winter
and maybe they should consider visiting a little bit more. As I said, previous thing
that’s happened a lot, is a lot of this works
been done on social media an existing core of
images, but we wanted to look fully into our photo library and what we found was
that these everyday images are actually valuable to participants. Images which are bad and
I put bad in quotes there because they’re photographically
not very interesting by themselves, by putting
them back into their location they’re often cited as helping
to provide reminiscing, so this is an example of
a person taking an image which was just for notes,
it was just a note, they had just taken this little
reference for the future, but it actually generated
this whole idea of a project they had done that they’d done when they were a freshman at university. Another thing that was valuable there was in how images might provoke more everyday event reminiscing,
so the ability to think of these very much smaller
events that had existed. A final thing that we found
to be really interesting was the way in which this wasn’t just reflections or reminiscing
about their lives but also reflecting on
their photo taking practice. So people, as I said
before, because when they have this photo map began to become aware of where they’d been taking images and patterns in their life
and how they took images. So, people began to become
aware that they’d always taken this route because there were
a larger number of images or there were a larger
number of notifications versus other things. And also people began to reflect on the kind of photos they’d taken, maybe their taking images of just places and not enough of people or
they weren’t taking images that they thought might be in 20 years were going to help reflect
on life as it was now. And then we also saw some
examples where people wanted to move beyond just reminiscing
and actually act on things. So, I mentioned at the beginning, we have these large repositories of images and we really don’t know that we’re going to actually manage them if they include things we don’t want. But because we were doing this
in a location-based sense, we cut down the number
of images people saw, we actually were reverting
to more energy use which was deleting lots of them. So, in conclusion, very quickly, working over the user’s whole photo library does provide benefits. We getting reminiscing
from images that people weren’t posting on social media. Obviously, images that were posted on social media cropped up during our study. The location-based part supported
contextual consideration. Then also, location acts as a filter to support viewing of related images, so images, as I said before,
which were related to a place could be pulled together as a scene . Location though also brings challenges, because of what we learned
through notifcations where they start to stop responding to them. This is not something we have
a solution of how you do this. And also, as I said
before, a lot of research has been done on temporal approaches, so a year ago from now versus location, so one of the things
that we really want to look at in the future is how we might combine those approaches,
temporal and location approaches. So, with that I will stop and listen to any questions you’ve got. (crowd applauds) – [Dan] Hey, Dan Cossly,
Cornell University. I have lots of thoughts about this project to talk about offline. I want to say that I think
you made an off-hand comment about change over time being expected, actually I think that’s really interesting to think about pushing on
how people all over time and as those photo collections get bigger it’s gonna be more useful. The other thing I wanted to say is you post it almost as context versus time and then you said both,
but really there’s other kinds of context too
that you might consider, associative memory, say,
the more context added, the more likely it is that
it’s gonna be a useful and retrievable memory and so
thinks like extracting people from places or objects or
all the other kinds of things that you might play that
you would context play might actually really push this forward. – Yeah. – [Dan] I guess that’s a
comment not a question, sorry. – Well, I’ll just use it to
debate where it’s on, yes. One of the issues,
obviously, with this is that we’re looking at this in a
location-based environment and we said people had to
have been their for two years, well, people have been living in other places previous to those two years and then they never see those images or if you do go on holiday to a nice place then you don’t see those
images with this approach. So, I don’t think I verse it
as a temporal location thing, I think we need to kind of explore the different approaches here and find out which ones are good in
different situations. But we’re looking at also more of a kind of semantic location in future, so semantically, I’m in
a bar, maybe I should see other images of being in a bar. Yeah, that’s hard to approach (mumbles) – We can do one more. – [Pamela] Hi, I’m Pamela Gibbs, I’m University of Baltimore. I apologize, I came in a little late, so if you’ve already
answered this I do apologize. Do you think culture
or diverse demographics impacts the way they view this project? Like the way they view the picture, like pictures or react
to the notifications or react to, you know. – I think the point being is
that they’re your pictures, so everybody’s going to react differently, so in the project it’s
your own photo album, not a generic set of images. I didn’t mention it but we
definitely saw people that they changed their
perception over timed images, so images that maybe weren’t
very interesting, obviously, images can become better if
you’ve pictures of somebody and that person sadly dies, or whatever, that image becomes better, but
we also saw it the other way, images we thought were very important for time became less
important over time in a way. It’s a really interesting talk, I don’t think we had a
massively diverse demographic to kind of answer that, we
were focused more on their technical ability to fit
the photos that we had, but that’s a really interesting thing to–

Leave a Reply

Your email address will not be published. Required fields are marked *