Insights Portal – Deeper Industry and Market Knowledge to Improve Your Business
Our Industry Insights service gives you the insights you need to run the business effectively. The relevancy of the content is the key: available news stories are always tailored to meet the individual customer needs. Industry Insights service consists of daily newsletters and access to Insights Portal.
To improve the service further, we have added new functionalities in Insights Portal and introduced a feature called My Insights. Fresh portal layout improves the overall user experience, while the launch of the recommendation engine and My Insights bring our AI to the surface, giving users even more relevant information to power their decision-making.
We talked to two of our in-house experts to find out about the operating logic behind the recommendation engine, and the benefits that our customers can expect to get from the refurbished portal.
Recommendation Engine: Applying AI in Insights Portal
Mauno Joukamaa works as a data scientist at M-Brain, developing and researching AI-based solutions for implementation in M-Brain’s products.
Mauno, you have done a lot of development work related to the recommendation engine that has been introduced in the Insights Portal. How would you describe the overall operating logic of recommendation engines? How does the operating logic of our recommendation engine differ from this? What are the similarities?
On a general level, most recommendation systems, such as those employed by Spotify and Netflix work by searching for similarities between the preferences exhibited by users. Users that are interested in the same kind of content or products are clustered together and a user belonging to a certain cluster gets recommended items matching the interest profile of that cluster.
However, this approach necessitates very large amounts of pre-existing user data, often several years’ worth. As Insights Portal does not yet have such an abundance of user data available, its recommendation engine needs to be able to generate recommendations from a much sparser dataset.
This is achieved by looking at what makes the preferences exhibited by a singular user distinct from those of all other users considered en masse. While the theoretical basis of this approach is quite different from the clustering approach, in practice it works well to provide recommendations matching the interests of the user.
Interesting. Can the recommendation engine fetch summaries in different languages? How do we make sure that the engine is able to, for instance, recommend Russian language content?
To infer the interests of the users, the Insights Portal recommendation engine leverages our proprietary Industry Insights classification system that identifies the key features of that content, such as the relevant industry sector and theme.
This system is applied across the different languages that Industry Insights content is produced in. The utilization of the classification system also allows the recommendation engine to work language-independently – when providing the recommendations, the engine can simply filter the result set according to the user’s language preferences.
So that’s a yes then. Will the recommendation engine continuously learn, based on the user choices?
Yes, the Insights Portal recommendation engine stays constantly up to date on the users’ preferences as exhibited through their interaction with the content in the portal. Thus, if the interests of a user expand to new directions, the recommendation engine is able to take this into account in the recommendations that it provides.
Cool. Finally, what other AI-related projects keep you busy at M-Brain, Mauno?
Besides development work related to the recommendation engine, my job at M-Brain revolves mostly around natural language processing (NLP) via machine learning methods, such as text generation and topic classification with neural networks. The NLP subfield of data science is currently seeing rapid developments and I’m excited to see how they can be employed to enhance the usability and functioning of M-Brain’s products.
My Insights – Access to Personalized, AI-based Global Media Content
Erika Linna works as product manager for Industry Insights at M-Brain.
Erika, what would you say are the main improvements in the portal, in terms of the user interface?
I would say it’s the overall user experience and user-friendliness that are now better, also when using the portal from mobile devices. We have introduced new functionalities, as well as user interface improvements to make the portal more visually pleasing.
Then there’s of course the recommendation engine. Is that available to all users of Insights Portal?
Yes, all the new layout features and the recommendation engine on the home page of Insights Portal are available to all users of the portal. The recommendation engine provides five recommended insights and the user can update the list to get a new set of recommendations. The recommendation engine brings personalization to the portal, as the recommended content is based on user behavior.
You mentioned the home page of Insights Portal. So, that leads us to the My Insights page. Tell us about that.
My Insights is an add-on feature that brings a very modern, AI-based element to the portal. The users can navigate between these two views, the home page, and My Insights.
What are the main benefits of the latter?
My Insights takes personalization to the next level. With My Insights, users will get access to our extensive editorial and social media source portfolio. They are able to follow those industries and topics that they are interested in. Users will receive new recommended content based on the summaries that they have followed.
So, users can follow individual summaries and receive new recommendations based on the content they have chosen to follow. Where do the recommendations originate from?
From the mentioned global editorial and social media source portfolio that gets activated with My Insights. These media recommendations are based on artificial intelligence, which refines the content based on user interests. So, ultimately this results in accurate content that is most relevant to the user.
Receiving personalized, AI-based content from a global source portfolio, is that how you would summarise My Insights?
Pretty much so. But I would emphasize that users will end up getting deeper insight and understanding of the industries and topics in the media. And everything is based on their own choices. We need to remember that the portal content has already been tailored to meet individual user needs. Now, these users can go deeper into the industries and topics that are of most interest to them, by following that content to get those further insights.
Thanks for explaining. And thanks for shedding light on Insights Portal and your work at M-Brain, Erika and Mauno!