"Bug in SparseFormatExampleSource?"

Legacy User

I read in some examples from a DB and vectorized them with the text tools. The resulting data set looked like this:



I saved it to a file with an ExampleSetWriter. Then I read it back in using a SparseFormatExampleSource. The result was:



Note that the word 'weight' has been _inserted_ into the Name column, pushing the other values down. But its value seems to have _overwritten_ the Statistics column.

Bug? Could it have to do with 'weight' being a special weighting column? I couldn't reproduce the problem with the sample newsgroup data. Maybe it didn't contain the word 'weight'.

Legacy User

To clarify, I didn't add any weighting column... Just sourced the text, applied the StringTextImput operators, saved, then tried to read in the file. Using both write and read operators creates two datasets, the original and the newly read one. So it's easy to see the differences.