Basically, the first comment of https://community.rws.com/members/jerzyczopik description is correct and good one. (Thanks)

Here, I'm just adding some wordings for a little bit more clarification.

1. There are no big differencies between these three score 100 matches.  "100% Match", "ContextMatch" and "PerfectMatch".

2. The extra wording is.. I guess, for marketing purpose.
    "Context" means, it applies two more conditions.
    Does this segment have the same previous source content as TM's TU ? and
    Does this segment have the same previous target content as TM's TU ?

    For instance, when I have finished a simple translation job like this;

in the TM's TU has a "Context" information too on top of Source/Target contents.
so
for the TU of 2nd segment, it has its own Source/Target ("This is my my second...bulahBulah..") AND its previous segment information also - the 1st segment Source/Target ("This is my First..bulahBulah") at the time of TM creation of the same document.

This is it.
This is a bare meaning of "Context"

Therefore, if you make a second project with the same docx and the same TM through "Pre-translation" step.
Then you have result like this. All "CM" of course.

On the contrary, if you make a second project with the same docx and the same TM not through "Pre-translation" but manully apply TM from Bottom to Top (starts with seg#3 ends with seg#1)
Then you have result like this. All "100% Match" (alahs.. the first one has no choice but to "CM")

If you lose any one of two conditions of "CM" then it goes to "100% Match" like this. The previous source and the previous target, both of them.

and

"PM" ?
never mind, 
it is for the users who spent "USD 3,000.-"
obviously not me. 
unless you pay it, you can not reproduce "PM"
just think it as "High priced 100%"

Basically, the first comment of https://community.rws.com/members/jerzyczopik description is correct and good one. (Thanks)

Here, I'm just adding some wordings for a little bit more clarification.

1. There are no big differencies between these three score 100 matches.  "100% Match", "ContextMatch" and "PerfectMatch".

2. The extra wording is.. I guess, for marketing purpose.
    "Context" means, it applies two more conditions.
    Does this segment have the same previous source content as TM's TU ? and
    Does this segment have the same previous target content as TM's TU ?

    For instance, when I have finished a simple translation job like this;

in the TM's TU has a "Context" information too on top of Source/Target contents.
so
for the TU of 2nd segment, it has its own Source/Target ("This is my my second...bulahBulah..") AND its previous segment information also - the 1st segment Source/Target ("This is my First..bulahBulah") at the time of TM creation of the same document.

This is it.
This is a bare meaning of "Context"

Therefore, if you make a second project with the same docx and the same TM through "Pre-translation" step.
Then you have result like this. All "CM" of course.

On the contrary, if you make a second project with the same docx and the same TM not through "Pre-translation" but manully apply TM from Bottom to Top (starts with seg#3 ends with seg#1)
Then you have result like this. All "100% Match" (alahs.. the first one has no choice but to "CM")

If you lose any one of two conditions of "CM" then it goes to "100% Match" like this. The previous source and the previous target, both of them.

and

"PM" ?
never mind, 
it is for the users who spent "USD 3,000.-"
obviously not me. 
unless you pay it, you can not reproduce "PM"
just think it as "High priced 100%"

Kelly Edward said:

1. There are no big differencies between these three score 100 matches.  "100% Match", "ContextMatch" and "PerfectMatch".

2. The extra wording is.. I guess, for marketing purpose.

Hi Kelly Edward

A very bold statement considering you didn't what these were at all at the start of this thread!  Perhaps I can just add an interesting example for you that might give you a hint as to why these different functions are important to anyone who does a lot of translation work.

Apologies to the German speakers... I may have something not perfect here but I wanted to illustrate the point.

Segment #2 in each file is exactly the same source, but the context is a little different in each case.  So when I translate the files from scratch the first time I get what you see above with three different translations for each one depending on the context.  Now if I clear them and pre-translate you can see that the correct translation is selected automatically in each case which can make a significant difference in terms of confidence when pre-translating large files.

If the context isn't known so easily then Perfect Match is even more likely to be correct.