Optimize Dictionary

Optimize Dictionary


Dictionary optimization may need to be done from time to time to improve performance of the prediction system. If you notice long delays when selecting words from the prediction window, or if scrolling in the dictionary editor is slow, you should consider dictionary optimization. Dictionary optimization will only have a limited benefit if you are using a large dictionary with a small dictionary cache.

Note: For optimization to complete, you will need enough free disk space to hold the new dictionary. If you have not chosen to Make the Dictionary use less disk space but be slower, which activates compression, you will need roughly double the amount of disk space free than the dictionary used before optimization. The dictionaries distributed with Aurora for Windows are not compressed.

Note: When you plan to optimize your dictionary, you should consider doing this overnight or over a weekend as it can take a very long time for large dictionaries to be optimized.

Note: If you are using a laptop, it is a good idea to plug it into the wall to prevent its batteries from dying in the middle of optimization.

Repair Damaged Dictionary (will not Optimize)

If you receive error messages that there was Internal damage detected in dictionary or you find that the Prediction program is hanging while you are using it, or regular optimization fails with an error message other than a segment load failure and you are sure you have enough free disk space, you should check this checkbox to repair the dictionary rather than optimize it. Dictionary repair takes roughly the same time as optimization and should repair most damage or internal inconsistencies in the dictionary. We strongly recommend you make a backup copy of your dictionary before using this option.

Note: Words which were damaged or that you have deleted using the dictionary editor will be included in the Words in original dictionary count as well as the Deleted and damaged words removed count. These words will be removed from the repaired dictionary, as well as any prediction information that is invalid. If you find that the bulk of the words in the dictionary have been lost, or hundreds of fixes made, you might consider contacting Aurora Systems Technical support before continuing. It is possible that the dictionary will be irreparable, in which case, we recommend contacting Aurora Systems Technical Support or restoring from a backup.

Note: The repaired dictionary is not optimized so you might want to run dictionary optimization on the repaired dictionary.

How much Memory should optimization use?

This value should be set to about 2 megabytes less than the amount of memory you have in your computer, so if you have 8 megabytes of RAM (8x1024=8192K) you should set this to 6144K. The larger the value here, the faster optimization will proceed. If you plan to run other applications while optimization runs, you may need to leave more than 2 megabytes of RAM aside. This value may not be set less than twice the Cache Size target value. If you do not leave enough memory free or you have too many programs running, the program may stop with an PREDICT.EXE caused a segment load failure in module XXXXXXXX.XXX error message in which case you will need to restart Windows and the optimization routine.

Make the Dictionary use Less Disk Space but be a little Slower

This feature controls the use of compression on the dictionary files. Dictionary performance can be improved on slower computers by not checking this checkbox. Note that not checking this checkbox will roughly double the disk space needed for the dictionary (.DCT) file being optimized.

Save Backup Copy of Dictionary File

If you do not want to save a backup copy of your unoptimized dictionary, you may uncheck this option. It is generally a good idea to save a backup copy when optimizing. Note that you will need enough free disk space for the optimized dictionary while optimization is running, even though the old one is to be deleted.

What Cache Size should we Optimize For?

This should be the same as the Dictionary Cache setting under the Dictionary tab of the Prediction Options dialog. Dictionary Optimization attempts to build a dictionary that will work best with this amount of cache memory.

In general, the cache size of a dictionary should be about 25% of its size. So if you have a 2 megabyte compressed dictionary on disk, you should use 1024K of cache (2 megabytes, double it because it is compressed, and divide by 4). For a 2 megabyte uncompressed dictionary, you should select at least a 512K cache. If you have lots of RAM, you can only get better performance by choosing a larger cache. For an 8M RAM system, a minimum 1024K cache is recommended.

Delete unused Words

If checked, dictionary optimization will delete all words with a 0 priority value. Generally words with a 0 priority value are ones that the prediction system has only seen once, which may include junk words.

Delete unused Phrases

If checked, dictionary optimization will delete all phrases with a 0 priority value. This may significantly improve performance on dictionaries which have been used a great deal. Generally phrases with a 0 priority value are ones that the prediction system has only seen once, which may include junk phrases.

Delete unused Macros

If checked, dictionary optimization will delete all macros with a 0 priority value.

Delete Words Marked as Adult Content

If checked, dictionary optimization will delete all words that have their Adult Vocabulary checkbox checked.

Word Prediction Statistics Thinning Factor Caution!

This number determines how much the statistics used for word prediction are thinned out to reduce the size of your dictionary. Removing these statistics can severely reduce the effectiveness of Aurora Prediction's ability to predict next words. We recommend optimization with the Delete Unused Words and Delete Unused Phrases checkboxes checked before using this feature. A setting of one will remove prediction statistics for rarely used words and may significantly reduce dictionary size. We do not recommend using a value above 1 for this setting.

Click on the Start button or press Enter to start dictionary optimization. Click on Cancel to abort dictionary optimization.

Dictionary Optimization Status

Dictionary optimization shows a status bar to indicate the percentage complete. This bar is only an indicator that optimization is progressing, most often you cannot predict how long optimization will take based on it.

If Optimization Fails

If optimization fails with an error message other than a segment load failure and you are sure you didn't run out of disk space, it is possible that there is internal damage to the dictionary. To fix this, you should run optimization again with the Repair Damaged Dictionary, (will not optimize) checkbox checked to repair the dictionary. See above for more information on this option.