Player is loading...
TRANSACT
Wednesday, Aug. 24, 2022 · 1:46 p.m. · 11m 19s
Targeted anticipation of customer expectations through a holistic and systemic approach to data.
Embed
Transcriptions
Note: this content has been automatically generated.
00:00:02
ah you ever
00:00:09
god had ever buy me thanks for still being here
00:00:13
for being awake after all of these uh different uh talks
00:00:17
so today i i'm not fond of class and i was presenting you pro x. and
00:00:21
i would like to present that together with my team members here so we have flowers
00:00:25
we have hunks one and we haven't found there and uh we together we built this
00:00:29
too so it's in a high listening to and will tell you like what that is
00:00:33
for pro active b. c. v. s. adviser so this is a challenge for b. b. c. v. s. that we were focusing on
00:00:39
so imagine the life of a financial adviser it's not easy i can tell you
00:00:44
i've talked to them they have a large number of clients and have very limited time
00:00:49
and there's increased competition everywhere so they have this big risk they might lose clients and also
00:00:55
they have this feeling of not knowing what the components in the minds of the clients what do they need
00:00:59
what a fierce and this is very important if you want to build a good relationship with your uh finance appliance
00:01:05
so yes we need to find it to to identify their needs in time and i asked them at the v. c. v. s.
00:01:13
well how do they currently do it because i wanted to understand what other resources and what
00:01:17
are the things that we should be looking out for with our tight back tool and they sat
00:01:21
well everybody has their own kitchen so i try to imagine what that looks like so i thought
00:01:26
so everybody has their all tools their own resources and then
00:01:29
use that to be able to uh advice declines well the thought
00:01:33
we can probably do something more better because this next one thing is you can
00:01:38
look at it it lie lacks the system yeah and it's everybody for their own sake
00:01:43
so we thought of this more modern looking page yeah i imagine it's a space where you can walk in
00:01:49
well in this case it would be a dashboard right and their
00:01:52
solution school proactive to systematic approach to prepare advisers for client needs
00:01:57
you going to this room you find a drawer and you see what
00:02:01
i people reading and feeling and what is thinking about going for that
00:02:05
maybe you want to know what competitors offering right now or you click on
00:02:09
another box and you see what i trending local search is in fine it's
00:02:13
who should i contact today and what products should i offer to whom
00:02:17
these are all questions that we would like to answer with our yard tool
00:02:22
now how are we going to do that you might wonder so the way we did that is by thinking of what can we have as input
00:02:29
what can we produces output as input we got from the c. v. s.
00:02:34
not even fourteen days ago i think it was really a just before the the jacket on we got internal data
00:02:39
from the bank and it were lots of excel sheets that we were looking at and trying to to not together
00:02:45
i don't suppose that we asked them what other competitors we did our own studies as well
00:02:50
and we thought what do they need to know they need to know what's happening in the lab
00:02:54
what's happening around here we need to look at local newspapers we need to look at social media
00:02:59
so that's what all our input and as a the tool
00:03:02
pro aqua doesn't do it uses natural language processing digital marketing and
00:03:08
data analysis a. i. data analysis to finally put choose a
00:03:12
different number of insights one it produces automatic summarisation of daily news
00:03:18
second it it looks at the sentiments of people how people feeling in the far left
00:03:22
today and what are what are they talking about and how they feel about these different things
00:03:26
what are the trending topics what people searching and found that what what i do but keep them busy
00:03:32
and we look better internal data and found some insight some interesting
00:03:35
patterns in their internal data who was contacted when what what will
00:03:40
this kind of information and how can we uh improve their uh strategies based on that
00:03:46
so why are we unique well we provide actionable
00:03:50
information so these pieces of information that we gave to
00:03:54
the bank all ready they were saying ah that's really good because we looked at the data and we sat
00:03:58
that's a wealth of data but you could do better you could get other
00:04:01
types of data with what with this data we could do much more sci specialist
00:04:06
um what we also did is we connected internal data sort of data from the
00:04:11
bank with external data the data that we find in the back panel social meeting
00:04:16
and uh what's also gonna by the fact that we are i don't be specialist we
00:04:19
can deal with lots of different types of data you don't need to have excel sheet
00:04:23
when you have species specific values redefines we can do a structured and unstructured data and we can
00:04:29
use an l. p. tools to to deal also with a free text that people might have put in your tax
00:04:35
um we'll stalemate state of the art an l. p. techniques and we have a focus on fun as you can see
00:04:41
and therefore we're giving the b. c. v. s. advisers and petted if advantage
00:04:47
so where stock team as i was already mentioning the different members of the team so
00:04:52
we have to an l. p. specialist is a most can't as you can see here
00:04:55
she has expertise in automatic summarisation which is very much needed for uh our task here
00:05:00
so myself i'm a a leader here at a at a at a
00:05:03
goodly there and i have a long time in working on multilingual text analysis
00:05:08
uh we have floor is uh who's uh our market here and he had to ten years of experience in financing digital marketing
00:05:15
and how much one who came from paris to join us and he has been working on the cross road of consulting engineering
00:05:22
so i'll what progress that we did did we make till now so we looked first when we got into
00:05:28
the challenges that let's look at their internal data so we need to data analysis and there was guy was very
00:05:33
busy saying all the time we met just uh excel sheet with this one and then we have contact with
00:05:38
the bank how do we actually map that and they gave us advice immediately so it was a very nice collaboration
00:05:43
so we got that done then we went onto scraping local news competition social media done
00:05:50
after that we took all those pieces of information and created
00:05:54
the state of the art transformer based models for the summarisation
00:05:58
keyword extraction sentiment analysis and finally and that was yesterday we put all that together uh to
00:06:06
create this modern kitchen as i said this dashboard right where people can see different things click on
00:06:11
different things and see what's coming out and we're still busy working on that so maybe now
00:06:15
is a good idea to show the dental in between oh sorry i should have the ability to
00:06:22
anyway
00:06:29
so the dental bubble really click quickly so proactive uh is that the tool right on top you
00:06:34
have different things you can click on so b. c. v. s. for example or you can click
00:06:38
only competitors and what that does is that it looks at uh the internet in this case or it looks at the specific
00:06:44
web sites that we have defined that are important and and then finds a in in order of twelve
00:06:49
relevance it finds a piece a different a website and
00:06:52
also gives you keywords and different uh summaries for that
00:06:57
so uh here for example have the competitors and we're looking at life
00:07:01
eyes and hear the click through on the links and see what's there
00:07:05
we look at civic segments because we learned from the internal data
00:07:08
but there was a difference between male and female uh and clients
00:07:12
so female segment for example here are clicking on that you feel that it changes
00:07:16
but this is still the mail now does the female segment uh well of unfortunately
00:07:21
we saw a lot of things with uh the force or uh_huh that's not so
00:07:24
nice but we also saw things like funny um yeah a lot of uh more
00:07:28
aspects of of of um the volumes and these kind of terms are more important for women
00:07:34
we look at the keyword analysis we have here us women health which is very
00:07:38
important in the competitors right is something that a lot of people are interested in
00:07:42
you can also see that in the trends information which we uh which we gathered for different she was that we found
00:07:49
searching to general news so yes there's a lot of different things you can do so that's why this demo is it gets that up but uh
00:07:56
here we're looking for phones an investment and you see that you can look
00:07:59
at some paycheck for example is on the b. c. v. s. but there's
00:08:03
many different pages they're all structures for the the advise uh so that they can look at those most relevant ones
00:08:10
the insides are blurred out because we didn't couldn't share those with you
00:08:14
but there was a quite some the uh like the things that we discussed already so she media analysis
00:08:20
and finally uh there is the recommendation system that things can also uh came up with a
00:08:26
as the latest the addition to our thing so here you fill in
00:08:29
some basic information uh age and uh uh important uh aspect for for banking
00:08:37
submitted and you get a lot of different investment recommendations right was
00:08:41
kind and then for all of these investment there's also information again
00:08:44
that's a or just the keywords there's the link and there's a
00:08:47
small snippet to tell you a witch or should be looking at so
00:08:52
oh well such information here i think one thing we didn't manage to add in this uh
00:08:56
video is the sentiment analysis but we also have it and it's on a separate slide but
00:09:00
i think we weren't in time to send that in before the deadline so but that's it
00:09:05
i think for the the damn also now i go back to the slight people liked it
00:09:16
right so that was how progress that was the dashboard so what are our next steps so that we were thinking of what
00:09:22
can we do for the b. c. v. s. and we also had some many many mentoring senses but we discussed as an exact
00:09:28
one thing that is i think right now the most important thing to do
00:09:31
together b. c. v. s. is think about the data strategy together with them
00:09:34
because they've already started this process and we found so many things we thought we need that writings can you
00:09:40
you talk to them as well said half we could have this kind of information we could link this up
00:09:44
we could do what we really wanted to do in this project based
00:09:48
on the on the letters happening outside predicting a little bit what clients want
00:09:53
so maybe we could have some consulting uh and uh like a collaboration
00:09:58
the next thing we thought of doing is a kind of trial where we um
00:10:02
go work on this for more than nine days right because mine is really short
00:10:06
come up with the tool box that can be tested with two adviser see what they think and we
00:10:11
would need like six months to come up with something that's really good right would have an extra functionalities
00:10:16
the file the the searching isn't a i listening full set up with multiple advisers
00:10:21
some uh some also some incense for management because we find many things we found
00:10:25
work maybe not just for advisers but also for the marketing team or for management
00:10:30
i will come in the house and the also with the safety of your data and all that's the kind of things
00:10:34
last thing isn't in a swiss research project would really like to do that with you because uh
00:10:39
it will be funded by the swiss government so that will be also advantages for the for the bank
00:10:44
we would have eighteen months to work with the post doc intensely on this and
00:10:49
have at the same time good results with a bank and also some research output for us
00:10:53
so i think that's it for me so the last things to say um if you
00:10:58
don't listen to your clients somebody else well we listen to applies using internal and external data
00:11:03
and then we present the highlights you so this way we think b.
00:11:06
c. v. s. advice can x. and anticipate declines ones that i think
Conference Program
LightAI4Comfort
Aug. 24, 2022 · 12:13 p.m.
119 views
LaternaMagica
Aug. 24, 2022 · 1:08 p.m.
118 views
SolveHR: Natural language inference for professional search
Aug. 24, 2022 · 1:26 p.m.
Recommended talks
Comment monter un dossier de sponsoring
Eric Héritier, Grand Raid BCVs
March 23, 2017 · 6:28 p.m.
346 views
