As I am only too aware of information overload, the choice of a weekly rather than daily paper made sense.
With time, I’ll try to see if it is possible to improve results by changing various settings. Right now, the content is heavily biased towards technology, although none of my criteria included them. One of the hypotheses that would allow explaining this phenomenon might be that one of my keyword was #future, and that future orientated tweets might tend to be dominated by technological innovations. Furthermore people using Twitter are most likely to be cognizant and interested in technology.
For optimal use for horizon scanning, the types of categories preset in Paper.Li (headlines, politics, business, art etc.) would need to be customizable to truly fit our needs (Update January 2015: this has been done by Paper.Li).