hi
i'm adrian cooper i'm chief product
officer for um torrentia
uh hello again if you saw my um earlier
talk um
this in this talk which is called when
data becomes the asset
i'm talking talking about data
management
in the age of agile integration
give me a second
so i want to drill into um
into sort of below the systems level to
the data or the metadata
itself and the importance of data
management data management for for
digital transformation
so you could say that the great digital
innovators have recognized the necessity
uh to connect data and systems not just
for the front end or
customer facing elements but also to
join these together
with the middle or back office systems
they understand this this need to align
online and offline
so it's not really digital
transformation in my view if we're only
changing one part of it
and organizations that only consider
changing the public facing elements are
likely to struggle to deliver value over
time
and i think certainly in my own
experience
[Music]
you know a lot of museums have actually
largely focused on that customer facing
or public
uh uh public uh or audience driven part
um digital teams and services have been
established but mostly
you know ultimately another silo and the
rest
of the of the organization has been
business as usual of course at least
until the
the pandemic um so i think collectively
i mean there are obviously
great examples uh and but we need to
work towards what's
what's often referred to as the
connected enterprise or or in this case
the connected museum
in the last talk i introduced this
concept of services as a way to help
organizations
to change their mindset um services
as a macro level uh as a way to
kickstart some of the thinking about key
problems
and develop what i call connected
processes
um so a service that doesn't respect or
follow you know an internal organization
and so if you're thinking about
developing developing something
there's this then an increased need for
organizations to find smarter and easier
ways
to combine the data that's traditionally
held in in these separate
systems internally
modern cloud native applications are
generally designed with this
microservices approach
and microservices are a way to enable
organizations to
orchestrate and bring together and
process
and deliver efficient experiences and
transform workflows
so a microsoft sorry a micro services
approach means that although data and
information sources become more
distributed across the organization
the data itself is brought together or
connected
as needed to support the task the
workflow
um or the overall the overall service
and this is what we mean by um agile
integration
so we can we can we can orchestrate and
bring together data
from a range of different microservices
which each do something specific
in order to present to the user an
interface that allows them to do
what they need to do for the set of
tasks that's part of their service
nothing more nothing less
so if we recognize this need to make the
alignment between the internal
and the external uh if you like the
workplace and the digital space
my data is connected merged or enhanced
through a range of new digital services
what does this mean
to the way we think about and manage our
data or
metadata so really so really what is the
challenge
for for data in that context
so we could say that that data is the
currency
of digital transformation because you
can't do digital
digital transformation without
transforming your approach to that to
the data to metadata
so as we start to plan any migration to
new cloud
native platforms with smart connections
we need to think about how we can ensure
that the data becomes
or remains a key asset so it won't be
smart unless you can unless you can
connect it
so for anything to be an asset it needs
to have value
and and therefore data needs to have
value to be treated um
and seen as an and be an asset
and so i think there are some key
challenges or key things to think about
in terms of what
as a profile we might think about
uh you know in in terms of that data
in this decentralized digital ecosystem
so we need to be able to share data
clearly
we need to be able to trust it uh it
needs to have
meaning and and context um so obviously
that's where
taxonomies and and and controls uh come
in
um it needs to be consistent
comprehensive um
and accurate it needs to be
accessible um controlled uh
secure uh if you think about right
rights and licensing and
all of those things we only only the
people that need to have access
uh we'll get we'll get access
and of course last but not least it
needs to be
machine readable so we need to we need
to have uh
computers to be able to look at data
interpret it know what it means
and to be able to do something logical
with it to process it
in order to to make these make these
connections
so it's clear that we need a more
flexible and
adaptive data management strategy in
order to deliver this agile and trusted
data which then begs the questions you
know who's
who's responsible or accountable how do
we
improve or maintain the quality
and and how do we how do we make these
smart uh connections or automate them
and in term
in terms of what we call the the data
supply chain the joining together of the
data
as needed to support the different
processes that make up
uh an overall service
so i think the ability to manage data as
an asset
across the whole organization depends on
a number of things
technology organizational culture
governance and and sort of staff
employee accountability you know
everyone's got to do their bit and i
think um deliberately in this slide
i've made the organizational culture and
and governance
circles bigger than the technology and
uh because i think that you know
whilst a lot of other people assume that
by upgrading their technology then the
data is going to take care of itself
sadly that's not that's not the case and
and
really it's about organizational culture
and governance
and thinking about things across the
whole organization
that's really going to going to make the
difference
of course technology uh will and can
provide
provide some help in terms of of
cleaning up and sorting data to you know
assuming that there are that there are
rules to follow um
you know data transformation services
can be uh
established and scripts written to try
to clean up poor data and you can
certainly use ai and machine learning to
to to process
um you know data and and and try and um
you know using using machine learning
and ocr and things like that to actually
improve the the tags and and other
metadata
about about the about the assets um and
and we can also use machine learning to
to attempt to create links to
uh linked open data sources um and you
know wiki data
and things like that to it to improve
the uh
the accountability and and add to the
weight
of of that content
so where are we where are we now
i think there are kind of two uh two
types of barriers there's organizational
barriers and there's tech
there's technical barriers to this in
terms of organizational barriers
well really museums aren't any different
to most other sectors
and his data has been managed in roughly
the same way in most
most organizations across all sectors
for this in the same way for the last 20
or 30 years
and that tends to be uh that it's
managed
kind of independently within it within a
departmental focus
uh because in turn that that relates to
the kind of systems that are that are
put in place
and so if you imagine saying in a museum
context uh in typically museum data is
designed and managed in these large
scale
business area specific applications such
as a collections management system
but but even within the same sort of
core area as we say collections
there is you know a lot of cases and
certainly my own experience
as a consultant a strong tendency
towards separate management and control
even within that that area so for
example management of objects archives
books
all part of the overall collection and
and that they're often
handled in in independent uh systems and
i know there are those who say well
there are reasons why but i mean that's
that's more historic i think than than
relevant for
for for the future but but having
independent metadata structures having
separate controlled vocabularies or
taxonomies
uh to to to to describe the data
uh leads to issues where it's hard to
harmonize any of that data
um if if you you know right at the
beginning i said of course services
don't respect organizational culture
so users don't need to know that your
data is held as separately in in
in a library collections or an archive
system they don't care they just want to
find
uh what they're looking for by topic
subject
etc so we need to find ways in order to
bring this data
um together to aggregate it and to
harmonize it
and then on top of the kind of the
thesauri problem as it were with
multiple different
separate systems we then got uh
functions which which are perhaps
duplicated across the organization so
something like like loans might be
managed by you know a number of systems
independently in an organization
or it could be that loans is only
managed in one system and in order for
say an archive object to be loaned it
has to actually be added to the
collection as
a as a temporary record in order to then
have a loan transaction uh carried out
which so
so these things you know are all
examples that
i've i witnessed as as a consultant
and um this siloed approach
as it were is perpetuated by some of
course very basic
and natural tendencies for teams to be
possessive about their data and i think
uh
that's quite understandable and and
there can be a reluctance to share data
or to adapt cataloging practices to
support the organizational wide needs
and and it but in a lot of cases maybe
the organization itself isn't
isn't strong enough or clear enough to
actually articulate what it wants in
order for these
departmental systems to come together
which is then part of this
you know service orientated thinking
let's think more about the outcomes that
we want
rather than thinking in terms of systems
and features and functions because that
just really doesn't help going forward
so um poor integrations and other
technical limitations of course help to
enforce that kind of departmental
outlook so they be it becomes symbiotic
organizational barriers uh are to the
connected uh
museum enterprise are in turn
perpetuated by this
this nature of legacy systems um
some things legacy not not not because
it's old um
it um it it can be it can be legacy for
for a number of reasons um but
um you know they tend to offer this
single interface single database you
know single set of of of modules
um and and often have data entry screens
which are not terribly user friendly or
difficult to use
there's a lack of flexibility and so
this leads to what's sometimes called
field hijacking where users you know out
of out of frustration or you know
uh just just put data where they want to
put it rather than necessary in the
place that
that is designated for it um so that
incoming
in combination perhaps with poor or lack
of data validation tools
uh you know if you're allowed to put a
date in the text field or vice versa
then then of course uh results in the
data quality being uh significantly
reduced that combined with with
terminology control which may not be as
effective as it could be
all points towards lower quality and
dirty data
so you know how do we properly identify
you know people places objects materials
subjects events etc
if the values that we're entering are
are simply strings of characters
that have no in intrinsic meaning that
can be
interpreted by by by other systems and
where the you know the data's in in
fields that
means something different to what they
were intended and that only confuses the
situation
so data managed through a sort of
business specific application
may well also require some form of
interpretation or logic layer to have me
to have meaning if the data is exported
outside the system
it can't necessarily be uh understood um
and and and as you know it's been talked
about a lot today
you know if there are no apis that makes
data hard to integrate
and that results in of course
time-consuming projects
to build inflexible or one-off
integrations between specific systems as
as needed and
there's certainly been a lot of debate
about you know what's the right
level of integration between the cms and
the dems but
from my perspective uh these
point-to-point integrations are
hard to maintain and any change to
either point will result in the
integration breaking and so we should be
thinking more in terms of
of cert of a service oriented view
rather than this system to system
type view because that that really is is
an older way of thinking
so if if given all of that we we look to
compare kind of legacy data
with our data data value profile that i
that i
put up earlier we can see that the
legacy data
uh comes with a number of number of
issues often so
you know it's it's inflexible it's it's
possibly hard to share
it doesn't necessarily have meaning um
it could well be inconsistent because
data isn't in the in the right fields or
it's not being properly validated
so which means that we can't trust it
completely which means that its value is
lower
and and you know possibly you know again
more importantly it isn't machine
readable
because we can't use any logic to
interpret what it
what it means so how can we move towards
uh data being being seen as an asset
and i think as we said if you know
organizations have relied on this
uh business or departmental uh areas
creating and managing their own
applications and data models and tax on
taxonomies uh historically but i think
now there's a need
to change the approach to data
management so that it works more
effectively
across the whole organization and that's
that's going to be a big change for some
and so in the connected organization
data is the most significant and
tangible asset
it needs to become the fuel uh that that
powers multiple use cases initiatives
not not just designed for one particular
use case or project or thing uh new
digital services will only be as strong
as the underlying data
that that fuels fuels them and we need
data to be structured
and machine readable in order to have to
have use
and the value will come from the ability
to collect
and correlate data from different
systems without need for this manual
interpretation
logic and automation so data that's
locked inside systems that has no
intelligent intelligible structure or
meaning can't be connected
um into a digital supply chain which is
then where we have to fall back on
manual
integrations so as organizations shift
from these large monolithic projects to
more agile
and service based approaches the concept
of an organizational
data catalog will become more and more
important
it's going to be important to build and
maintain
this shared data catalog in order to
understand how
key data types key entities gathered
across
the range of systems can be connected to
support the overall data
supply chain and and and support the
needs of specific
uh services so you can see here some
examples of the kinds
of entities i'm thinking about that are
currently held across
uh different systems whether that's
objects or digital assets or
products or locations or customers or
rights you know you know again conscious
that rights are
handled separately in digital asset
managements and collect management and
collection management systems
whereas the reality is that they need to
be brought together need to be properly
harmonized
in in a in a museum context which makes
you know currently makes workflow
uh difficult for those who are trying to
do uh do that role so
we need to think about these key
entities and look where this where this
data is stored
agree how we're going to you know share
share on thing you know on things like
taxonomies across the organization
in order to be able to orchestrate the
data in a way that it can come together
using uh you know machines and services
to deliver the kinds of services that
you know uses
uh and uh expect
so in summary i i think there are
you know there are a few a few key thing
things to think about
um and and so first of all is this
rethinking the data supply chain making
sure that that's done at
an organizational level rather than
necessarily a departmental
or departmental by department level and
i think the service based thinking
where you're thinking about about
specific problems and outcomes
uh will will help with that there's
certainly a need to increase the data
modularity
and and you know to break things down
and move towards this micro services
and focus on on the development of of
apis
uh to to uh to to support this this
bringing together of the data
um and and i need to think more as an
organization about the data
the data governments data is an asset
that everyone has to manage well it's
not it's not
you know somebody over there's
department uh
you know to think to look after um you
know everybody
everybody has to have responsibility and
of course we can use more modern systems
to
to improve that that and keep that keep
that maintained
as we as we go along and last but not
least is the harmonization of metadata
metadata models this is
absolutely critical when we start to
think about shared taxonomies
not only between you know within an
organization but actually between organ
between organizations whilst you know
we've been developing apis some of the
bigger organizations have managed to do
that
there's certainly no uh um
there's no standard around the way that
they've done that so there are already
issues in trying to draw
draw data from different places together
if they're if that if that's not uh
it's self common so within the
institution and then institution to
institution
as we start to collaborate more on on
these on these projects
um so so i hope that's given you uh
some things to think about it's not
always the most exciting part but i
think if we don't get the data element
right
uh then a lot of the other things can't
uh can't follow on
um from that so thank you for for
listening
and good night
