I have been trying out SkimtoDevn script, and it seemed very useful until I realized that it appears to make the See Also function almost useless. I suspect that this is because it produces notes that have a similar form, and therefore the AI thinks that other notes with a similar form are the most closely related.
Does this make sense?
If so, is there any way to avoid it?
Or will it always consider notes produced by SkimtoDevn to be more closely related than other notes produced manually?
Thanks,
Can you explain in more detail what you mean when you say the See Also function is almost useless? Do you still get results for See Also & Classify, but they are not accurate? Or are there no suggestions for See Also & Classify at all? I ask because I am seeing an issue this morning that I have not seen before. I have a large number of various file kinds that now turn up no matches for See Also & Classify, and I’m wondering if it is similar/related to what you are seeing?
PS: I don’t use the SkimtoDevn script myself, so I cannot test this myself. Perhaps if @houthakker reads this, he may also have some input.
No, some results show up all right, but only from the articles that have been produced by the script.
I speculate that this is because of the format that the script produces, such that such documents have a similar form, similar words and so on. This effect would be diminished , I suppose, if every note had been produced in the same way, but for the moment all I’m getting is notes produced by that script.
Well, I have resolved my problem but it took a full database rebuild. Good luck with finding a resolution to your issue-it sure sounds curious that the format of the document limits the AI’s content search/comparison.
No, I suspect it is just the fact that increases the similarity of individual records because of the way it outputs the notes.
I don’t imagine there is much that anyone can do about it, unless there is some way to programme the AI to ignore or accept certain elements.
Anyway, I was interested to know if anyone had had similar experiences and found ways around it.