Appendix C

Start      Previous      Next  

The next topic concerns the problems in the source data.

EduRS.xml Fragment 1 - Figure 18

• [F18H0] The middle name is all in lower case. If name parts are all shifted to upper case or all shifted to lower case before comparison then this becomes unimportant.
• [F18H1] The  MAJOR  data field is empty. This is needed to properly label the  EducationalTraining  instance and so this row will be rejected.

EduRS.xml Fragment 2 - Figure 19

• [F19H0] The  INSTITUTION  field has mixed case characters which will cause a mismatch if character for character equality is required for organizational matching. However, if all characters are shifted to the same case before comparison this will not be a problem.
• [F19H1] The  NETID  field is missing. This causes uncertainty when comparing names. Since it is not uncommon for there to be several distinct people with exactly the same name parts, a token like netid is perfect for disambiguation of people. This is because netid is guaranteed to be unique even when a person’s name has changed due to choice, marriage or divorce. A missing netid weakens the association of a degree record and a person.

EduRS.xml Fragment 3 - Figure 20

• [F20H0] See Figure 18, highlight 0.
• [F20H1] Extra embedded whitespace. The XPATH function normalize-space deals effectively with this sort of issue.

EduRS.xml Fragment 4 - Figure 21

• [F21H0] Trailing whitespace.
• [F21H1] Missing netid. See Figure 19 Highlight 1.

EduRS.xml Fragment 5 - Figure 22

• [F22H0] Same as [F18H1]. This row will also be rejected.
• [F22H1] Same as [F19H1].

EduRS.xml Fragment 6 - Figure 23

Start      Previous      Next