Is it a good idea to attend seminars and then try to generate new research problems? Once I attended a panel and a senior mathematician said, "That's like asking how a composer makes music." Not very helpful.
One tip I read was to ask for variations, e.g. modify hypotheses. The issue I've run into is that it sometimes leads to almost identical results with no new ideas or techniques. In other woeds, slight variations on known results. Those almost never get published in good journals.