When I search for how to do personalization in Azure Search I only find the "Personalization Service" that is marked as obsolete. Nowhere I found what is supposed to replace that service.
I want to create a new search index in what is now called "Azure AI Search". I then want to add personalization to the scoring of the search results. So that search results that are more relevant the this users profile will rank higher. Preferable a fairly advanced personalization, using something like implicit collaborate filtering.
I can implement the personalization outside the Search Service, with a custom post-processing step of the search results. But if I could do it within the search service, that would be best.
I want to make a recommendation engine. Give boost (higher search score) to items that are more relevant to the users profile. Not only what this user explicitly likes but also implicitly based on what other users with similar profiles likes. One way to achieve this is to use implicit collaborate filtering.
I would imagine this is a common use case. Does Azure have any way to achieve this?
Consider implementing Scoring profiles that would let you personalize based on multiple functions, including, for example, adding tags that you can link to groups of users. Depending on your use case, you would have to identify the strategy to boost specific terms, however, you can then adjust your logic to be implemented as part of the product configuration.
If you need user recommendations specifically, take a look at intelligent recommendations or you can build your own real-time recommendations API based on this pattern