In short, when comparing personality, you take his/her profile and dissect it according to "kinds of activity classification". There are weights for different levels of creative processes and by number of such processes artist uses with taking weights into account you will get concrete aggregate number for each person you try to analyze.
Then you put it in bigger scale and use "personality structure" to include into consideration other two major factors: socio-cultural influence, which measured by numbers such as sales, ratings and so on, and lesser quantitative psychological/personal qualities branch. For simplicity, that latter branch can be excluded since it does not really related to phenomenon of artist, when you try to compare several of them between each other (contrary to creativity and socio-culturological branches, which are both necessary).
So with filling in all of information with weights and coefficients one can compare two artists without dealing with whatever personal likes/dislikes. And you will get result that, for example, Paul MacCartney is objectively bigger socio-creative phenomenon than Britney Spears, for example. You will get that not because you loke MacCartney, but because the aggregate figure you got for him would be the way bigger than for Spears.
The same comes to songs: if you dissect them as it shown in deeper layers of creativity (as shown in research "Personality structure") you can fill the fields:
5) Work "Earth Song" =
text/poem +
sound +
scenic performance +
mini-film.
6) Sound of "Earth Song" =
structure +
rhythm +
performance +
melody +
arrangement +
timbres.
7) Structure of "Earth Song" sound =
first part: the overture +
second part: the development +
third part: the culmination +
forth part: the relaxation, main theme in canonic performance
(classic 4-part symphony structure).
In items 6 and 7 there are detailed analyses of "Earth Song" sound parts (see concise one in "Alternative HIStory" article \\\reference\\\).
So you expand each branch of this structure and fill in everywhere if certain aspect of work/song exists or not, if it is new to this artist or not, and so on.
Thus you can compare "YANA" and "Morphine" or "Earth Song" by counting quantity of new/existing subparameters for each. Eventually there will be figure that will tell you that no matter how anyone can like it or not "Morphine" is work of greater self-expression degree and creativity than "YANA".
Please notice, of course, that there is no way to determine objectively "truly" good and just "good" songs unless there will be some conditional/arbitrary number to differentiate two.
Also, it is important to note that there is technique to determine relative strength of the melody -- part is analyzed on the matter of dispersion of tones with time intervals taken into account, as well as repetitiousness of tones and harmonic tonality. With that, it can be said that melody for the song "D.S." is weaker than "Earth Song" objectively. However, computer programmed rendition of that kind of analysis is subject of universities' research and is not accessible to common public yet.
Of course, this methodology is not absolute and in some cases one can cheat it with harmonic tonal parts that do not constitute melody, but all other ways are fit to be evaluated as stronger melody than actual average quality melody. But if this method used on actual songs, it works.