Please help me to understand the clear functionality of Semantic groups in DTP

Hello SAP BW Gurus ,

I have read so many threads on the Semantic grouping on the SCN forums still i am not completely convinced with the answers given .. I must be poor in analyzing ... I have the below questions please clarify.

Scenario 1 : Source ...> DSO

We have Key fields and Data fields in DSO. In sematic group , do we need to give the same key fields which are part of DSO definition or based on any other condition we need to given the semantic group fields ?

For ex : Sales order item data

Order ID(k1) Item(k2) Sales Rep date price qty created date have become the key fields of the DSO . Do we need to give the same info objects in DTP Semantic key ?

And also please confirm me if my below understanding is correct

Source data : ORD1 P1 30

ORD1 P2 45 ( Assume this record is having bad data )

ORD1 P1 60

(Assuming semantic grouping is not maintained )When data loaded initially as per the standard key mechanism in DSO : ORD1 P1 60 WILL be loaded and after correcting the erroneous records from PSA ORD1 P2 45 Value will overwrite the existing record and which is wrong as per the functonality.

When Semantic grouping is enabled : All the records will be kept in a single package and after correcting the records in Error stack and finally the latest record gets updated in the DSO.

Scenario 2 : Source ...> Info cube

from source to info cube , in this combination we cannot use sematic group right ? We must be having DSO in between . So , it will be Source ...> DSO ...> Info cube ? or

We can load the data from source to info cube .

Thanks all in advance!!

Regards,

Teja