- il y a 3 mois
AI-Powered Social Listening Transforming Intelligent Data Capture
Catégorie
🤖
TechnologieTranscription
00:00Hi everyone, welcome to this new VivaTech session. We are happy to welcome you
00:05today about social listening and AI. My name is Romain Ponceau, I'm the vice
00:11president of Digimind, a French company specialised in social media listening
00:15and one of the leaders of the market. And I'm thrilled to introduce to you my
00:18dear friend JD. My name is Jonathan Daglish, I lead the intelligence service
00:25centre in Deloitte UK. My background is in government and military intelligence and
00:30my team have taken those principles in what we do and applied it to a business
00:33model. We've been working with Romain as a partner for five, six years now and they
00:38are a Digimind our main data and technology provider for a lot of what we
00:42do. So before really deep diving in, you know, sorry for that, before deep diving
00:53into our core topic, we wanted to remind you what is social listening. So
00:58basically the aim of a social listening tool is to collect insight, to track
01:05trends, to follow some conversation that are produced on social medias and that's
01:12we are doing it live and the purpose of that data collection is of course to
01:17influence your strategy, your brand reputation and communication but as well
01:21to track your risk and that's JD's expertise.
01:26Yeah, thank you Romain. And from our perspective social listening is a huge part of what we
01:31do and one of the biggest challenges we found with social listening is that the
01:35quantity data that's out there, there are millions and millions of data points. It's
01:39almost impossible to actually collect against that in a granular way and an
01:43informed way without technology helping you. So the influence of AI and other
01:48aspects of technology has really helped our team and really driving for our
01:51insights and it's been a great part of our business with Digimind.
01:57So the first way we are using AI at Digimind is for query enrichment. So you have to
02:04understand that you have to tell to the platform what you are looking at, what you
02:08want to collect on social medias and basically it's a list of words and
02:15boolean operators and that was the formal way of doing it. Now thanks to AI we have a
02:22bunch of suggestions of words that might appear of keywords of hashtags that are
02:29suggested to better cover your topic, your subject. We have as well the
02:35possibility to push you some themes that might be interesting for your business
02:40and that are automatically generated in more than 81 languages and I know how
02:46relevant is it for you JD and especially for your use case. Yeah absolutely so as
02:50Roman says really relevant for our business we are a multi-jurisdictional
02:55business in our own right but all of our clients operate across many many
02:59countries in many different languages and quite often with a lot of things
03:03around crisis and risk time is of the essence and therefore the ability to
03:07create accurate searches in multiple languages means we can provide the
03:10relative response to our clients in the most appropriate ways and feed into that
03:16is trends around risk trend is very important to trend analysis within the
03:20platform the machine learning identifies trends very very quickly what that
03:25allows to do is that we get to form an opinion on what might be happening in the
03:29first instance so the machine learning will tell us that a change in the risk
03:33profile or the risk understanding has been detected and then able to dive into
03:37that and understand exactly what is driving that is it an influencer is it a
03:41topic a geographical topic a regulatory topic and very quickly we can dive into
03:46that so of course it's really important and relevant to identify this peak and and
03:53this peak appearing but the next step was for us to understand the why behind that
04:00peak so thanks to the AI we've got right away the the why the the context of the of
04:08this peak and as well some metrics that helps you to analyze this situation so we
04:13will identify who was the person who took the first what is the reach what are the
04:19interaction so you can of course answer right away if it in term of crisis for
04:25for this specific use case but as well to identify a good buzz and make sure
04:30that your communication team will be able to relay that on the social medias and
04:36and to bring that to life from our perspective we do a lot of services with
04:40third-party ecosystem so it could be around emerging acquisition it could be about
04:45due diligence and existing third parties and therefore it's imperative that we
04:48understand what is driving risk against those businesses because quite often
04:52those risks flow back through to our clients as the parent entity in a
04:55corporation or as the entity on those relationships and so not only do we
05:00detect the trend changes with the metrics that Raman mentioned we can
05:04quantify the risk and the impact so if it's a brand issue how big is that brand
05:08issue very quickly understand that reach that volume but also drilling down by
05:12things like location demographics languages and platforms so identifying
05:21understanding was really important but the next step runs away is this new
05:26feature the AI sentinel so it's smart alerting that will send you right away an
05:32email on alert to let you know that something is popping before it really
05:37happens so the machine learning the AI is able to identify that something is
05:43happening just before it explodes and we are using exactly the same way of
05:48analyzing giving you the context why you got this alert and what are the
05:55important metrics so fighting and find the influencer and point it out who talked
06:00about that what is his audience so you can deep dive in that analyze it what are the
06:07hashtags that have been used and some key metrics like interaction that might be
06:13low at the beginning but it will push it to you that you have to focus your
06:17attention on that specific topic because it might be an issue in the in the
06:22coming in the coming weeks or seconds and this is actually really
06:28revolutionizing how we create more resiliency through our clients so quite often
06:32people talk about resiliency resiliency isn't the ability to mitigate
06:37risk in the sense that you can avoid them a lot of it's been responding rather
06:41than reacting and what I mean by that is by having the foresight the the
06:45understanding intelligence that something is changing you can
06:48appropriately prepare yourself for what is what's going to come along therefore
06:51you were responding what AI sentinel does it gives that early warning that time
06:57that that breathing space to be able to do that versus reacting and reacting is
07:01quite often done from an uninformed place therefore decisions that you make
07:05aren't prepared they are understood and the actions that you're taking therefore
07:09might not necessarily solve the problem at hand so Roman and I we talk great length
07:15about AI sensor what a great factor is and and for operation resiliency third
07:19party resiliency brand resiliency it's a real game changer particularly for many of
07:24our clients that will work with at this moment in time so sentiment analysis is
07:31something that has been a big issue for a long time in an industry and
07:35especially silence what we call silent mentions meaning some some mention that
07:41are including irony sarcasm that were really really hard for us to tag as a
07:47positive neutral or I mean negative mention that happens a lot as well and
07:54thanks to AI NLP machine learning we can teach to the platform by providing some
08:01example of that mention is negative for my business because it might change from
08:08one client to another the machine will learn from that and will be able to reduce to
08:14very low volume of unqualified mentions and it's key for your business as well yeah
08:23absolutely so I think one thing that AI's always struggles with with sentiment analysis is in the
08:29shorter form of social media reports and so by having this machine they learn loads of
08:34the top when are on this go in there and actually review the data and it picks up
08:38actually watch relevant beat from the sources romances whether it is positive is
08:42negative the accuracy is phenomenally increased which means we think
08:46particularly around brands where sentiment is so imperative understanding that
08:50sentiment how you're being conceived by potential buyers consumers then it's a real game
08:57changer for us as well and let's quickly talk about visual this insert we talk
09:03about a lot of data one the biggest change in data probably in social media and
09:08media over the last 10-15 years is visual social media so the likes of your
09:15Instagrams YouTube's there's been a real uptake in the data there and what that
09:20has created for many of our clients is a risk around the use of their brands are then
09:25being incorporated into other brands and not counterfeiting potentially as well so
09:30where people are taking their products their designs and actually highlighting
09:34and this is really really important to what we do we'll we'll train the AI with a
09:39number of images with did you mind that will then allow us to run that search
09:44against the open web and identify instances where that brand has been used so
09:49working with clients if it's not a client website if it's not a client-owned product
09:53very quickly we can identify potential counterfeiting brand and logo misuse and
09:58abuse and then partnering with our legal teams in Deloitte would then have the
10:02ability to take down some of those websites and prosecute those counterfeiters and
10:07so of the last year or so we've been working with some great brands haven't we
10:10collectively and we've had some real traction success there and it's it's
10:14certainly one of my favorite parts of the platform
10:18so what's next because we are working so hard on AI and as many of the companies
10:25that you you probably met during VivaTech so the next step for us is the
10:30panorama so the idea is that on DigiMind on our social listening platform you are
10:37running from 1 to 150 queries about different topics around your brand or
10:45around your products around your competitors but how do you do you get the
10:50the core of what's happening right now and this panorama will be live in the
10:56coming weeks beginning of June and it will help you to identify the key topics
11:02that are popping at the moment so as soon as you connect on the platform you get
11:08what's happening in your environment can be legal it can be new products new
11:14competitors whatever but it will catch your attention right away on what's
11:18important for you and I'm personally looking forward to this we're quite often
11:24talking the team about proximity so you have a brand an organization and you're
11:29aware of the risks that are in front of your business but quite often that
11:32proximity that analysis allows you to see where else you might be at risk but
11:37also where the opportunities might lie so what are others doing in your field
11:40your geographic location and it provides that it's going to provide us a great
11:44insight to support clients with that maximizing of opportunities and minimizing
11:48the risk and I had a question for you before we leave out of all these AI
11:58features that the public discovered today what are the core ones what will be the game changer for
12:05your industry and your business it's a great question and I think there are a couple of areas
12:11and firstly the use to generate AI within our business has been something revolutionary so we're
12:16always looking to scale and how do you scale well you can put more people on the job I can
12:22use
12:22technology so we collect intelligence in a great positive way with did you mind and then the use
12:27to generative AI to allow us to summarize those results is something we've rolled out recently so
12:32we can write reports and then review with humans that's reduced our cost on that by about 90% and
12:39so I'm
12:40looking forward to seeing what else the platform can do in that space when the query sentiment all that kind
12:46of
12:46is great but Jenny is I guess the next move for everybody we would like to thank you for your
12:52attention and feel free to ask us questions afterwards and thank you once again for being here thank you
Commentaires