Sammy Deprez and Wouter Baetens are managing partners at Arinti. As specialists in AI & Data Science, they have developed a new Cognitive Search Engine. But what is it? And what can you do with it?
We see that many companies have large amounts of unstructured data. On their sharepoint, there are often thousands of documents, audio and video clips mixed together and no one can find anything in them. To solve that problem, we use a service on the Microsoft platform called Azure Cognitive Search.
Specifically, we built a front-end on top of their environment so users can easily search through their own content. In addition, the Cognitive Search Engine (CSE) also creates additional information to optimize the search: for example, graph analytics can be used to visually show the connections between certain documents or keywords.
So the cognitive element is in effectively recognizing what is on a document, what is said on an audio clip, or what we can see on the image of a video clip. And then the CSE can therefore make connections.
We sometimes call the CSE Google or Bing 'on steroids': where classic search engines work with labels and tags, a CSE can really read comprehensively. Thanks to applications of NLP technology, the CSE understands and interprets what is in a document and thus also recognizes certain things such as persons, key words, etc.
The CSE brings all the knowledge together and presents itself to the searcher in a structured way. The main advantage of the CSE is therefore the time savings it provides. Information can no longer be secret.
The link to business is quickly made. For example, we have worked with Flanders Investement and Trade (FIT). FIT's mission is to get Flemish companies to export, and also to get international companies to invest in Flanders. They use their CSE, the 'FIT Enterprise Search', to easily find information and knowledge about those companies. Documents about trade missions, for example, or information reports on certain companies. There are currently about 15,000 documents in the FIT environment, so making them easily searchable is crucial.
Other sectors that make extensive use of CSE include knowledge-driven organizations such as the Social Secretariat or HR service providers. These organizations have to provide advice to other companies and therefore information must be retrieved easily and tailored to the customer's needs. The same applies to helpdesks or contact centers: a customer calls with a specific question, CSE ensures that the answer can be found quickly.
We also recently won a hackathon from the Federal Police. They see the application of our CSE especially in their investigative work. On the one hand to easily find information around a person, object or organization, but also because the graph analytics quickly shows visual connections between certain parts of the investigation, which is of course a great asset in police work.
A final example of a CSE application is online shops such as Bol.com or Amazon. Here, CSE is used to build a smarter recommending agent, which can quickly link the information you have searched for to certain products. News websites also use this tool to offer you interesting content so that you stay on their site as long as possible.