Abstract
Web search heuristics assume we have the user profile in the form of particular attribute ordering and a fuzzy aggregation function representing the user combining function. Having this there are sufficient algorithms for searching top-k answers.
Although we have already obtained significant progress in mining user combining function, there is still a problem of sample creation for user evaluation. Problem of finding particular attribute user ordering still remains a problem, too.
We overview our former approach, lessons learned and describe analysis and proposal of an upgrade of our system. Our main contributions are the description of an iterative process of acquisition of user preferences and proposal of two methods for creating a sample set for user evaluation during each iteration.