You need to ask more specific questions for us to provide better answers.
In the nutshell, it's scoring a 9-mer (for 5' splice site) and a 23-mer (for 3' splice site) against the consensus (representative) sequence built from all the splice sites in the given species. For 5' splice site (ss), 3 exonic flanks and 6 intronic flanks around the 5'ss are used as those positions are generally well conserved in the context of splicing with basal canonical elements. Similarly for 3'ss, 3 exonic flanks are used in addition to 20 intronic flanks to account for variable polypyrimidine tract lengths upstream of the 3'ss. The lower the score, the more difference one would expect the input sequence is from the consensus, or can be interpreted as a "weaker" splice site.
You can easily obtain the minimum and maximum values by using the tool to score every splice site in the transcriptome. We converted such scores to percentile ranks and received favorable responses from scientists who have to interpret these scores.