• Moved positions of Table 2 and Figure 5 to produce better layout.

• Various phrasing tweaks.

• Removed bold math in figure captions.

• Finished table of implied edges and rewrote description of S1 SIG construction to match.

• Shifted details of implied edges into a table.

• Rolled rewritten functional dependency constraints into the lists with the other rewritten constraints.

• Made consistent use of math mode for \LS and \NLS.

• Switched to newtxtext and newtxmath packages from txfonts. ...

• Removed no longer needed \!, as the spacing is now better.

• Switched to subfig package from subfigure. ...

• Switched to align environment instead of eqnarray.

• Fixed a couple of unconverted - to \setminus.

• Re-ordered and re-numbered rules. ...

• Added note about likelihood of S3 having different information capacity, due to not being able to map all the constraints from S0.

• Added macros to generate constraint numbers and constraint lists in standard format. ...

• Removed unnecessary \makeat* macros.
• Changed notation for propagated constraints.
• Minor tweaks to phrasing.

• Rephrased definition of key constraint edges.

• Rearranged position of some figures.

• Fixed horrible line breaking in math formula. ...

• Corrected annotations of copied projection edge between \TT{NLS} and \TSSNL.
• Added note about whether copied projection edges actually still count as projection edges.

• Switched to \setminus for set differences. ...

• Made proper use of \TrivialSelection and \TrivialProjection.
• Deleted sentence explaining interpretation of edge annotations with respect to reading direction.
• Deleted paragraph containing Batini et al's transformation terminology; not important.
• Clarified what isomorphism means.
• Rewrote transformation discussion to make it clear that the SIGs are being transformed rather than the schemas (even though we are effectively transforming the schemas).
• Clarified composition transformations.
• Various minor phrasing and grammatical changes.

• Removed confusing “clarification” of edge annotations. ...

• Removed footnote about nulls, as it didn’t seem to make sense and added nothing useful.
• Edited discussion of selection and projection edges.

• Various changes to phrasing up to Section 4.1.

• Corrected publication year of Date’s view updating book.

• Updated Section 7.

• Updated Section 6.

• Updated section 5.

• Various edits to Sections 1–4. ...

• Oriented focus more towards modelling.

• Updated to use ACM style.

• Added missing ACM copyright style file.

• Renamed ACM class file back to original name.

• Added initial version of paper.