MRT Conversion and Proofing

How are Machine Readable Tables (MRT) created?

To create MRTs from an author supplied table a series of Perl and C+ routines are used. The Perl scripts were created by AAS Journals staff especially for MRT processing while the C+ routines were developed by the CDS for formating tables in their VizieR service. More than 75% of all papers with MRTs have the original tables submitted as LaTeX tables. For standard LaTeX/AASTeX 5.0 deluxetables the conversion is straight forward and the main process of creating a MRT from a LaTeX submission is given below. Tables submitted in different formats are created in only a slightly similar manner.

  1. Take all the tables out of the manuscript.
  2. Fix the LaTeX table if it is not standard. This inludes user macros or unusual commands, a table that uses the TeX command \begin{table} instead of \begin{deluxetable}, or has embedded LaTeX comments (i.e. "%"s).
  3. Remove any standard, extraneous LaTeX commands and keywords (e.g. \phantom, \nodata, \protect, \tableline, etc.) and convert special LaTeX commands (e.g. Greek symbols, \le, \ge, \sqrt, etc.) to their ASCII equivalents.
  4. Format the data using CDS's acut program and create a preliminary metadata header based on that formatted data.
  5. Fix any discrepancies in the data columns by hand. Fixes to common problems includes giving superscripted flags (i.e. \tablenotemark) their own seperate column and placing any string information in a numeric column into a nearby flag column.
  6. Flesh out the metadata header's "Explanation" section and "Notes" (if necessary) with information gathered from the table or paper.
  7. Check the agreement between the format indicated in the metadata header against the data with the CDS's anafile program.
  8. Check the metadata header against the AAS Journal's standards and styles.
  9. Send MRT to the author to proof.
  10. Update the MRT based on authors comments.
  11. Send the final MRT off to be posted with the article.

Proofing tips for long MRTs

Even though MRTs are checked against their originals before being sent to the authors for proofing, you are encouraged to carefully proof both the meta-data header and the data in their MRTs. At a minimum you should at least spot check the MRT enough to see that the intent and data of the original table has been transferred to the MRT. If the data was already formatted for a computer when submitted the data was not modified during the creation of the MRT. Only the meta-data header needs to be proofed for errors. However in some case, extraneous blank columns may be removed to make wide tables more compact. When the conversion is from LaTeX to MRT, problems are much more likely to occur in long LaTeX table that require extra personal formatting. This occurs when authors submit tables where many step 5 type iterations (see above) are needed to create the MRT. Each "personal alteration" slightly increases the odds of introducing errors to the data. For these tables, authors should carefully check items such as: that all flags are present and associated with the correct column, that string information in numeric columns is still faithfully represented, and that all columns have the correct number of significant figures.

One final way to check the data in a long table is to plot some of the data columns and see if the results are correct. Python, IDL, or Perl programs are provided to plot or extract columns of data in MRTs. Authors who want to avoid any proofing of their electronic only tables may use the web-based MRT creator page to make and submit their own files.