 My session today is going to be about delivering intelligent content using Drupal. One of the questions that we've been trying to answer for ourselves at our organization has been, how do we bridge the new technologies that are available for us and give more value to our customers? So before I get started with the session, I'll just introduce myself. I'm Shyamala. I'm co-founder of Unimity Solutions, and I'm also on the board of the Drupal Association as the community elected representative. Our organization, Unimity Solutions, is in Chennai, down south in India. We've over 10 plus years of experience in Drupal. We've started using Drupal with one of our largest media sites down there in India. And since I've been delivering a lot of enterprise solutions using Drupal. So before I get into actually answering the question of delivering intelligence, I just want to start with a short story. And this is a story of a Chinese philosopher, and he was having a chat with his disciple. He was actually showing his disciple a set of pots. He showed his disciple a clay pot. He then showed him a brass pot, which had very intricate designs in it, and a golden pot with gems embedded in it. And the question that he asked his disciple was, which aspect of the pot do you think is most valuable? So I just want to pose that question back to you. Any thoughts on which aspect do you think of the pot is the most valuable? Any thoughts? So, I mean, is it the color? Is it the shape? Is it the size? Is it the design? What makes it, what aspect of the pot is valuable? And of course, the student answered, saying that the golden pot was most valuable. The philosopher said that the value of the pot was in its emptiness. So it isn't the size, shape, or the design, but the value of the pot itself is in its emptiness. Coming back to content, enterprises have large amount of content that's already existing in the enterprise. Large amount of PDFs, words, documents, PPTs could be testimonials, client testimonials, could be blogs that are written, maybe real-life case studies. There is a voluminous amount of content that is available in the enterprise. Which aspect of this content is most valuable to the enterprise? And how do we make this content valuable? The value of the content itself is only in its consumption. So we need to make sure that people are using it. Is this content getting consumed? So the usefulness of content is in its consumption and we need to build different layers of intelligence to the content so that we can make this content consumable. So just, I think, like the philosopher said, a pot is useful in its emptiness. We need to correlate it that for us, the usefulness of content is in its consumption. I wanna get started with first defining what is content intelligence. And here, I just wanna segregate content intelligence to two parts. One is assigned content intelligence and another is acquired content intelligence. So if you see assigned intelligence, this is typically, I would say, you could classify it into the metadata, the keywords, the various attributes that we add to make our content structured, maybe even the relationships that we bring in. If you look at a blog, you have the author for the blog and the tags. So it's the structure that we define when we create the content, becomes part of the assigned content intelligence. The other part, the acquired content intelligence, this is more related to what is the value that gets added to the content as the content gets consumed. As we use this content, as our users start using this content, what's the value that gets acquired to this content as acquired intelligence? Here, this is typically probably the statistics in the number of times the content was downloaded or viewed. It could also be pieces like comments and reviews where the user interacted with the content to post saying that the piece of information was very useful to them. Today, there are far greater avenues in capturing the interactions and the intelligence that are available with the content. I'm gonna just, in the next couple of slides, just get into a little bit more detailed, starting with the assigned content intelligence. Here again, with the new technologies that are available to us, we can go beyond just the metadata and content. We have, we can come up with processes where we automate some of this classification and meta tag that gets created with the content. There are tools that are available where your images and videos could be further classified using these AI and machine learning tools. We can also integrate with translation and transcription services at the time of content creation and have the content translated in different languages and the videos transcribed. In the acquired intelligence, here, we have an opportunity to, at every time, the user interacts with a piece of content and when he actually consumes it, maybe we could call it a session. Every time that a user interacts, we have an opportunity to identify who consumed it, further define the profile of the user who consumes it, the attributes, his social attributes as well, and when this content is getting consumed. What's the context in which it gets consumed? Was it in the morning or in the evening? So you have an opportunity to identify when the content was consumed. The other dimension is how, was it in a device? How long did he spend on it? Attributes around how he consumed could also include the interactions he had with the content, the purpose for which he reached out to the content. Another dimension could be the sequence, as in what was the content that he see, did he see before and after? And where is he actually physically, the location of the user itself? So if you see at the end of each interaction with the content, we have the capability to identify multiple dimensions of the user and capture these as analytics. And if we run machine learning on top of this, there is a capability for us to then come up with recommendations which are far more real. That we've kind of understood what is content intelligence and how content intelligence gets classified as assigned and acquired intelligence. We move into understanding how can we maximize content consumption using this content intelligence. So this is our goal, is to see how can we make content consumable and how are we going to utilize this content intelligence that we have built in to maximize consumption. So making content consumable would mean, am I speaking correctly? Okay, so making content consumable, the user could interact with the content in two ways. One is he comes seeking for the content where he's actually looking for the content, he's either searching or he's on a page which he came through search or through a link that he received on his email. The second piece is when we push the content to him, we stimulate him to come to our content. So these are broadly the two ways in which we can reach out to a user or have the user land up in a content. And in both scenarios, when the user lands up in a content, we have the capability to customize the content based on the user's profile. We could make it presented in a language that he can understand through up certain sections of the content that are related to his profile and his personal needs. And there could be certain sections which we customize based on his needs. The second case where we stimulate content, where we push content either as text messages or as mobile notifications. Here again, understanding the user's profile, the way he interacts with our system, we have then the capability to actually throw up to him contextual content that is most relevant to him and we can do this real time. Of course, providing a consistent experience across the different channels in which we interact with him will further help us maximize content consumption. So I think kind of the main idea that I wanted to bring is how by building this intelligence to the content, we can then get greater value to our content and make it consumable. So what's the key difference between content intelligence itself is not entirely new. We've been trying to do this in multiple ways. We started earlier with probably creating meta tags indices. But the key differential today and the key power that we have today with AI and machine learning is that this is no longer programmatic. There are no predefined rules that we set to build this intelligence. We today have the capability to actually learn and with learning what happens is that these recommendations closely mimic what the user would require. So we are able to real time capture these interactions and come up with recommendations that are far greater and far more pertaining to the user's needs. Just want to, this is definitely a topic that is much in renaissance. Here is a quote by the Amazon CEO where he says that some of these problems in machine learning and AI is almost like it was science fiction. And suddenly today it is real and we're actually trying to identify how people interact with the content and try to mimic user behavior and understand user behavior. So just at this juncture, just wanted to understand if any of you had any specific scenarios where you have used AI with content that you would like to share. Okay, excellent. Chatbots along with Drupal content, okay. And this was for a customer experience, customer, what was the use case that you were trying to solve? Okay, okay. And anybody else who've used AI with content or Drupal? So some of the examples that I just wanted to bring to the table. Of course, shopping is a large area where there is a lot of opportunity for us to throw up real time relevant recommendations. Here I just wanted to take a shopper who's actually looking for probably, he's just purchased an iPhone. Typically when the shopper buys a particular product in our traditional shopping experience a couple of years back, we used to throw up related products and these could be probably either other phones or other gadgets that would play with your iPhone and probably a case or a head, your piece or anything that is additional to the iPhone that would add value to the customer. But today what we have is that a capability to gather data which is far beyond and supposing there is a user from Washington DC and he is actually shopping for an iPhone, we can actually present to him waterproof essentials because there are suddenly a surge in the number of people from DC who are purchasing waterproof essentials because of a thunderstorm that's happening there. And this is not something we need to program into the system. The system can automatically pick this up based on the learning that it does. It identifies that there is a surge in a certain type, there is a demand for a certain type of product in a certain location and it automatically throws this up as a real time, very contextual recommendation to the user. So which is just an example of how we could, which illustrates the difference between how we traditionally did recommendations and how we have the power to do recommendations today. Today at Unimity we are working with one of our largest newspaper companies where they have about close to 35 publications online. We're using what's called a universal recommender, which runs on the analytics that is getting collected in the back and using an Oracle system to throw up recommendations. And this machine learning tool has a capability to ingest many number of actions, events, profile data, contextual information and then in just all this to come up with recommendations, the power that this brings is that for a organization with 35 plus newspaper is not just recommendations within a particular newspaper portal, but actually across their platforms, they are able to cross pollinate content between their portals. The second example that I wanted to showcase here as a good use case of how we can use content intelligence today, is in using natural language processing to classify content. Project Feels Initiative is initiated by NY Times where they classify, they parse content and classify them into different emotions and they call or use what's called as perspective targeting, where they are able to throw up ads to users based on the emotion the particular article is supposed to evoke. And this emotions is dynamically built using natural language processing. Another use case is translation, where we have all the large cloud players like Google, Microsoft, Amazon, all of them having services for us, which uses machine learning in the back end and deep learning and machine learning techniques, which help us to, without having the inbuilt expertise in our teams, we can just plug into some of these cloud services to bring these capabilities. One of the examples is DB Corp, which is a very large publishing company in India, which used the Google Translations APIs to translate their interviews and publications. And they had 95% they had a 95% good quality translation into the Indian language Hindi. Amazon also provides a translation service and hotel.com, they have close to about 34 different portals and I think 34 languages, 85 websites and close to 350,000 hotels that are registered. It gives them a clear hedge edge when they serve across their 20,000 locations when they are able to personalize the hotels into different languages. So in all these cases, if you note, I just want to bring back to say that whether it was the recommendation, whether it is targeting of ads based on emotions or in the translation AI tools that are available to us, we are adding a layer of intelligence to the content to make sure that this content is getting consumed by the user. There is a more likelihood of the user consuming our content. So I'm just actually going to conclude the session with just a short note on, now that we've understood how what is content intelligence and what are the layers and how it's some examples of how it's getting utilized. What is it that we need to do to make sure that we can then deliver this content intelligence jointly with our clients? So I think it's important for us to understand this is not something that we can solve just by plugging in an AI tool into our system. It's not like just plugging in an Apache solar and then you have search ready. It's very important that we understand the business that our customers serve, the different points at which our customers can interact with their customers. Nike is a very good example. They have a Nike Plus app that is for runners, which helps them to track running history. They give content that helps the runners to stay motivated, train them better. And this app is a good example of how gamification and digital can help better connect with not just your customer. And in this product that Nike brought in, they not only connected with their existing customers, but they had an opportunity to connect with also their new customers. And so what this means is while we try to deliver this, it's important that we also jointly identify with our businesses. What are these additional touch points at which we can better connect, gather data about our customers and then better serve them. So here in Nike's case, they were actually not just selling, they were trying to help people reach their goals better with the products that they purchased by creating this running app. So in the how do we achieve, it's important we jointly understand and connect better with our customer's customer. We have like a unified analytics. We have a unified content publishing system where we can create ones and publish everywhere. It's important that we work with our businesses to understand the specific use cases where we can learn and create intelligence and of course create a unified experience. So how does Drupal help us do this? And here is where we can leverage Drupal's innate capabilities in the content space, increased emphasis and increased features that we are having in the content authoring space where content creation is getting better. Workflow and auditing, which is very important piece for us as we serve the enterprises. Structured content that is possible for us to create and this structured content is what we can use in creating the assigned intelligence. The interoperability of Drupal, the capability for us to for it to interact with the many systems helps us to integrate with AI and machine learning tools and the API first initiative, where it allows us to deliver to the mobile and the other, not just the mobile and the web, but also the other devices or channels in which we wanna reach our customers. Last but not the least, it's important that our applications are secure, scalable and agile. So with all this, we're then all ready to deliver content that gets consumable by building not just intelligence while we create the content, but we constantly acquire and make the content contextual, real time and relevant for our consumers and make content more consumable. That's it, thank you. Thank you for listening in and just wanna finish my session to say please, those of you who've stayed back to Thursday, I know there's always a, the first day has the maximum number of attendees and then it becomes half and maybe, maybe even less than one fourth of the last day. But the most exciting part of Drupal is the contribution spins that happen tomorrow. So any of you can contribute, please join the contribution spins tomorrow. And just wanna conclude that please share your feedback on the DrupalCon, the link that I've shared here and happy to take questions. Thank you. Yeah, sure. Do you have an example of a way that content was delivered in a personal way across different devices? Like a company that's done this and what it's meant for them? So in our case, the example across devices, I don't have one right away, but in our case, it was only across their different properties, the 35 different newspaper publications, they were able to give a consistent experience. Yeah. I guess some example of the cross-device personalization is many of the big players, I think Amazon, BeatAmazon or Google. I use Google's assistant and the tools for news reading. They give you a personalized experience across the devices. So irrespective of what I'm shopping on Amazon, on my website or on my mobile phone, there's definitely a continuity of experience. And they're also able to personalize it for me based on where I am and some of my past transactions. Shopping is where I think a lot of AI has been implemented. Though financial sector, they are, I know of couple of projects that we personally wrote proposals for in the financial sector in India, where we wanted to bring in AI and experiences where it can be shared across devices. Do you wanna come up here and share? I don't know how relevant this will be, but so we are currently doing an Alexa integration with Drupal and there, this is also related to shopping. So the way Alexa is connected to Drupal commerce, you have it on your phone, you have it on the laptop and you just tell Alexa to do things and then she adds products to your cart and it's just consistent across the devices which have the same set of base as an integration-wise, which are all consistent. So experience-wise, it gives you an intelligent answer. When you, for example, I wanna add four red wine to the cart, it might just ask, or if I say 10,000 red wines to the cart, it might just remind me, are you really sure you wanna add 10,000 red wines to the cart? And that kind of an experience it gives across the devices, consistency-wise, plus it's smart enough to tell you that you might be making a mistake. Okay, thank you. Any other questions? So thank you very much for being in here. Once again, just encourage you to submit your feedback online. Thank you.