It's basic principle of paiwise alignment at all. You can find some information you need here:

http://psb.stanford.edu/psb-online/proceedings/psb02/chiaromonte.pdf

http://staffweb.ncnu.edu.tw/shieng/pairwise_alignment.pdf

http://www.incogen.com/bioinfo_tutorials/Bioinfo-Lecture_2-pairwise-align.html 

Also know as substitution matrices. Basically, they provide a score for any pair of matching residues which is related/obtained from the observed frequency in alignments of related sequences. You can also read more here: http://www.ncbi.nlm.nih.gov/BLAST/tutorial/ & http://www.ncbi.nlm.nih.gov/blast/html/sub_matrix.html 

Hello Isma

In my view a scoring Matrix is not 100% the same as a substitution matrix

A Substitution Matrix gives you the penalties of each possible Substitution in a given model (so gives you generalized relative scores).

A scoring Matrix gives you a score for a given alignment - so it is the application of a Substitution Matrix on an alignment. (usually you Count a positive score for fits in an alignment and substract misfits - either differences in bases or indels).

Greetings

Nikola

Sorry this is not true, just in case you already read it I did not want to delete it.

Scoring Matrix is used synonymously with Substitution Matrix

For Nucleotide sequences, there can be used a flat positive score for identical residues and for nonidentical there may be used a negative score or zero. The thing which is very important is gap penalty which is a negative score. Someone can use a flat gap score or someone can use different gap penalties for opening a gap and extending any gap.

For Amino acids, there are some available substitution matrices, which actually take care of identical residues as well as nonidentical but similar residues (having same properties). Those similar residues also get positive scores based on their substitution frequencies. Some use evolutionary concept or some matrices are built by direct observation of substitution frequencies among a large set of sequences.