People search the internet for information to help them decide on things from ‘who am I’ to ‘where should I eat dinner’. As info is the basis for decision making, better and more focused info is the basis for quicker and easier decision making. Personalizing content and search results is a powerful way to provide such focused quality info. Instead of ‘2845 results for Italian Restaurants in NYC’, a more effective approach would be ‘Check out these 3 restaurants’ where each one is just right for you in terms of location, time of day, personal preferences, and other parameters. But how can a machine ‘decide’ which results to serve for each user personally? The answer lies in understanding what is a decision and how we make decisions.
Simply put, people make decisions by weighing various options against various consequences in order to address some needs. These options and consequences are based on the meaning attached to the information and knowledge we have about the world around us. Created in our own image, computers also posses the capability to use information in the form of data to weigh ‘options’ against ‘consequences’ (if-then conditional statements) to enable a result (execute). Unlike us, however, they have no means to derive meaning from this process. For instance, the information that there is a raging storm outside prompts us to instantly derive a long list of other inferences (e.g. need to wear warm cloths; check the road conditions, etc.) based on the meaning of the information (e.g. a storm means possible strong winds and dangerous conditions). For a machine, however, the input [a storm is raging outside] is a stream of data devoid of any meaning. As meaning is the basis for deciding how to best address our needs, the simple conclusion is that in order to enable machines to serve us with focused information that addresses our needs, they need to be able to translate cues about our needs into structured streams of data.
This is exactly what the Datapersona project is all about. The technology behind our personalization engine is capable of dynamically tailoring content and/or search results to match the interests of each individual user across all devices. It does that using a plethora of content-side and user-side inputs all processed using our proprietary ‘real-world’ minded algorithms. What are these inputs, and how are they used to provide personalization will be the subject of my upcoming posts.
Eyal Engelhardt Ari, PhD