Jump to content

Why must nulls be searched for explicitly ?


Λύχνις Δαν

Recommended Posts

Hi ya,

 

  In searching the syntax one finds that nulls are not found unless one deliberately asks for them. I would think it better to include nulls by default so that cases are not missed and so that one does not have to do two searches in order to find all syntax results. Of course being able to find nulls on their own or exclude nulls makes sense but I wonder why they are not simply included in the results where a preference is not specified. Can someone explain why they are not ?

 

Thx

D

Link to comment
Share on other sites

Daniel, I am not sure I completely understand your point, but I would say that the syntax modules are built as add-ons to the Hebrew or to the Greek texts. By default, you would find words that are in those texts. If you want to find words that are not there, but that are understood as need for a clause to be well-formed, you will need to ask when you search.

 

If would be different if we rewrote the original text, adding substituting actual words for null elements.

Link to comment
Share on other sites

Hi Marco,

 

  Perhaps a fuller explanation of my question will help a little.

  Below are two searches - one for casus pendens, and the other for casus pendens which are null.

post-32023-0-32441000-1448374776_thumb.jpg

 

post-32023-0-35939200-1448374791_thumb.jpg

 

  Now these produce two non-intersecting sets of hits and I don't see a way to do the same search in just one construct - perhaps there is a way. So now I need to know that nulls would be present in order to search for them. Or at least I have to suspect so. I think it would be better if nulls were included in the first search and you could then weed them out if you were not interested.  As it stands while the first search expresses no preference for either null or not null it gets only a subset of the complete results back.

 

  Now, concerning implementation. Indeed the syntax module is an add-on and the text of the module does not include nulls where the syntax places them. I do not know how the syntax search for is actually implemented but combines searches of both the syntax tagging information and the text itself. I don't know how complex it would be to do or what the consequences might be on performance of searches. But the consequence of not doing it is that one must work out when multiple searches might be required and then do them.

 

  The difference I suspect is whether one primarily searches the syntax tree or the text. For a syntax search the former makes sense to me.

 

Thx

D

 

 

Link to comment
Share on other sites

Daniel,

 

I understand your point now. As you say, the nulls are represented in the syntax tree, that reflects the database.

 

I don't know what your suggested change would involve in terms of search algorithms.

 

I think that it's a healthy mental attitude to suspect that there also may be nulls.

 

Why does the search discriminate against them? I would still say because they is no hard evidence for them. They are inferred as syntactically required, and therefore understood. Now, one could be strict and understand fewer nulls, or one could understand more. It seems to be that it is good to distinguish between hard items and soft items, so to say.

Link to comment
Share on other sites

Actually, now that I understand the issue better (a helpful follow-up, Daniel), I'm not sure why nulls are excluded from a general search. I think I wondered this when they were first implemented in searching and I suspect that the answer is historical. In other words, the nulls were probably the most difficult issue to problem for the syntax searching and they were implemented a couple years after the basic syntax searching was added to the construct window. Then, when they were implemented, they were kept distinct. Whether it is a large programming issue to include them in the general search is beyond my ken. But I'll ask.

Link to comment
Share on other sites

better to include nulls by default 

 

Hi Daniel,

 

If they were included by default and you wanted to exclude them, you would have to negate them,

 

post-32543-0-08119700-1448472390_thumb.png

 

and of course, two negatives make a positive, and you would end up right where you started.  :)

 

Regards,

 

Michel

Link to comment
Share on other sites

 a programming issue to include them . . . I'll ask.

 

Hi Robert,

 

While you are on that topic, would you mind asking about searching for Nulls as Lexical items? As I mentioned elsewhere, one of the features of the database is the ability to plug in covert subjects for overt Nulls based on the diagram. If Nulls are empty records in the database, and they share the same constituent tag as their coverts, could Acc write an algorithm to replace empty record with their lexical coverts? I don't mean on the Syntax Diagrams, but as part of the Acc search engine. Then you wouldn't have to compromise the theoretical purity of the database. We can already drop Lexical items in syntax searches; it would add precision to be able to drop in implied Null Lexical items.

 

Of course, based on the complexity, an algorithm couldn't get it all right. It would require proofreading. I would be willing to proofread before it was offered to the public; I'm sure others would be too.

 

Regards,

 

Michel

Link to comment
Share on other sites

Michel,

 

I'll sort through some of the implications first and then see what they think. I can't promise it, though, since whenever I bring up "nulls," I see a bit of post-trauma in their facial responses.

 

Robert

  • Like 2
Link to comment
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...