Jump to content


Photo

Bizarre narratives using all custom local sentences

narrative reports custom sentences

  • Please log in to reply
11 replies to this topic

#1 TomH

TomH

    Advanced Member

  • Members
  • PipPipPip
  • 6250 posts

Posted 29 January 2016 - 11:20 PM

I started exploring an idea for using narrative reports in a way that minimises computer generated fact sentences in favour of a report dominated by fact notes. A prior approach was to customise fact sentences to point form, i.e., no attempt at formulating grammatically correct sentences. Repetitious point form is not as discomfiting as proper but repeating sentence structures and words and it does save space. This new concept goes a step further and would suppress fact sentences, leaving point form presentation of facts to tabular reports such as Individual Summaries and Family Group Sheets. However, applying the concept stimulated bizarre results in the Narrative Reports which suggests some underlying problems that may be affecting to some degree more common usage.
 
Descendant_Narrative_Compare_No_vs_All_C

Click the thumbnail to see a side-by-side comparison of the first page of a Descendant Narrative created under two boundary conditions:

  1. Left side: using all default sentences
  2. Right side: using all custom local sentences whose content is simply a pair of square braces, "[]".

I have also tested with "ZZ" as the custom local sentence with the same result so the fact that square braces are the control codes used to denote a sentence variable is immaterial. My initial test was with an empty sentence switch, "<>"; same results but the square braces help to parse what is going on.

 

The red rectangle highlights the dislocation of the person's name and events when appearing as a parent or as a childless child.

 

For the same people, the blue rectangle points out that the default sentence for "people with no entered facts" is being outputted despite that the person has "entered birth facts". This is illogical and misleading in that it is the absence of any default event sentences for the person in the report that is the trigger. My inexhaustive testing suggests that if every event for the person has a custom local sentence, even ones containing the usual variables, the report process outputs the "factless" sentence. 

 

A child who is a parent gets a built-in "sentence" (effectively a point form) of vital facts on both sides (green rectangle) while a childless child gets the same narrative treatment as he would if appearing as a parent (a childless child does not reappear in the report, unlike a child who will be reported later as a parent). So the narrative generator for a parent and childless child is probably common to both; at least, if there are two generators, they share the same faults.

 

There is no difference between the two reports' endnotes and indexes of names and places. However, one report has no places outputted so the index works from the raw data, not what is actually output.


Tom user of RM7550 FTM2017 Ancestry.ca FamilySearch.org FindMyPast.com
SQLite_Tools_For_Roots_Magic_in_PR_Celti wiki, exploiting the database in special ways >>> RMtrix-tiny.png app, a bundle of RootsMagic utilities.


#2 kbens0n

kbens0n

    Advanced Member

  • Members
  • PipPipPip
  • 3455 posts

Posted 30 January 2016 - 02:38 AM

The output is definitely bizarre, but unsurprising really.
It appears that in order to generate sentences there needs to be, at minimum, one valid placeholder token as sentence subject (person, place, thing or idea). I am not near a copy of RM.
Everything else is programmatically-related; conditioned upon the quantity, types and order of tags along with the hard-coded words in order to artificially fill sentence output with data field values, type of noun or pronoun, prepositions, indicate gender, etc.
No subject (or tokens, for that matter), likely precludes the program's routines (designed to extract, assess grammar and construct) from being able to do any work.
I'm guessing the date problem maybe defaults to (date unknown) for the same reason (ie. its part of those routines that if there is no date tag, then the date field does not apply locally). Since the local Customize Sentence field is not blank (you've entered characters), the Fact-Type List defined defaults maybe cannot override either, except for the Birth fact which has a special bit of extra programming the other events do not get.
Can't imagine how hard it was just to implement Bill Biennia's token system as an afterthought, rather than at program design time. Looking forward to improvements Bruce & Co. surely hope to make in the future.

---
--- "GENEALOGY, n. An account of one's descent from an ancestor who did not particularly care to trace his own." - Ambrose Bierce
--- "The trouble ain't what people don't know, it's what they know that ain't so." - Josh Billings
---Ô¿Ô---
K e V i N


#3 TomH

TomH

    Advanced Member

  • Members
  • PipPipPip
  • 6250 posts

Posted 30 January 2016 - 10:24 AM

Thanks for your late night thoughts, Kevin. I have now discovered what actually triggers the default sentence for "people with no entered facts". It is the absence of the [person] field in the Individual facts/events. That is why this 'person is factless' sentence followed the list of facts. The first instance of the person's name is boldened, whichever is the first Individual fact to contain it and it receives the citation superscripts for the "General" sources. You could have a whole introductory paragraph of facts without the [person] in sentences and then introduce the name. 

 

So I have created a custom _Name fact type with the Principal sentence "[person]: ", added it with a SortDate of 1 so it is first on the Edit Person screen and first out in the Narrative. Now the other fact sentences may contain anything or nothing and the 'factless' sentence is not triggered. That will enable a 'Notes Only (almost)' Narrative or a point-form Narrative without boring name repetition dependent on the programmed name-cycling. (Jerry, you use point-form, I think, so is this old news to you?)

 

After some more play with SQLite on this, I will post some more results.


Edited by TomH, 30 January 2016 - 10:26 AM.

Tom user of RM7550 FTM2017 Ancestry.ca FamilySearch.org FindMyPast.com
SQLite_Tools_For_Roots_Magic_in_PR_Celti wiki, exploiting the database in special ways >>> RMtrix-tiny.png app, a bundle of RootsMagic utilities.


#4 TomH

TomH

    Advanced Member

  • Members
  • PipPipPip
  • 6250 posts

Posted 30 January 2016 - 11:17 AM

Here is a report from the same database. Everyone has the _Name fact type added (bad name, I should call it "_Heading" or something) and all other events have the "<>" custom sentence so only their Notes are output. Now it is much better behaved.
 
Turned off citations as that is another problem - my thought is that they would be included in the Individual Summaries which would be published alongside the Narrative.
Descendant_Narrative_All_Custom_Sentence
Have to look into the marriage and family sentences and the witness sentences for shared facts but this looks encouraging. It is as though the Edit Person screen is an outliner for a much freer narrative comprising one or more fact notes, including 'facts' which may be nothing more than notes.

And it would be wonderful if Notes could access the sentence template language variables. OMG! Why not use Note editor to draft the local 'Sentence'? Duplicate Note and Sentence so nothing lost in transfer. For export to other GEDCOM, RM would expand the variables in Note.


Tom user of RM7550 FTM2017 Ancestry.ca FamilySearch.org FindMyPast.com
SQLite_Tools_For_Roots_Magic_in_PR_Celti wiki, exploiting the database in special ways >>> RMtrix-tiny.png app, a bundle of RootsMagic utilities.


#5 Jerry Bryan

Jerry Bryan

    Advanced Member

  • Members
  • PipPipPip
  • 3566 posts

Posted 30 January 2016 - 07:12 PM

 (Jerry, you use point-form, I think, so is this old news to you?)

 

Yes. For example, I have played around with a Name fact that precedes the Birth fact and for which the sentence template consists simply of "[Person]" without the quotes. The purpose of such a fact would be as an attachment point for name sources, rather than using the general person as the attachment point for name sources. For now, I'm sticking with the sentence template for the Birth fact being the first appearance of the [Person] variable.

 

I find this whole thread to be very intriguing and very imaginative. I have also long wished that notes could access the sentence template variables, and of course I've expressed very strongly my interest in being able to associate citations with notes. That way, you could even dispense with the sentence templates themselves and sort of create whatever reports you wish with little difficulty. For example, with such things in place (and maybe a few other relatively minor tweaks), it might not even be too hard to produce quite reasonable reports in German or any other language. Of course, you have to deal with sentences for children without children or spouses, sentences for spouses, and other such things that RM gives you little or no control over.

 

Jerry



#6 Jerry Bryan

Jerry Bryan

    Advanced Member

  • Members
  • PipPipPip
  • 3566 posts

Posted 30 January 2016 - 07:22 PM

One more thought: I've even considered putting all my "notes" into individually customized sentence templates. That way, the template variables would be available. And I know that you can put carriage returns, bold, italics, etc. into sentence templates. But the trouble with putting text into the sentence templates instead of into the notes means that that text is virtually unmanageable (try searching  it, for example) and that the text is virtually un-exportable. So I really can't go that route. But all the various ideas in this thread are very suggestive of what is really needed to  produce good narrative reports and even good books. The existing sentence template language just isn't what is needed.

 

Well, and one more one more thought: what about images - for example a marriage photo (or photos!) associated with a marriage event in a narrative report, or a headstone photo associated with a burial event in a narrative report. I would love to know if these are the kinds of issues that the RootsMagician is planning to deal with in the rewrite of the report generator.

 

Jerry

 



#7 TomH

TomH

    Advanced Member

  • Members
  • PipPipPip
  • 6250 posts

Posted 31 January 2016 - 04:04 PM

I have evolved a rough script that sets custom sentences so that a point form narrative is output. Here is an example of the results. Alternate Name labels are expanded by their type and the role of a sharer in an event is included in the event heading.

Descendant_Narrative_Point_Form1.png Descendant_Narrative_Point_Form2.png

I might italicize or bold the fact headings which could help find them if multiple facts were in the same paragraph, thus saving some white space.

When the script is cleaned up and I can more completely describe it I will post it to you-know-where.

Tom user of RM7550 FTM2017 Ancestry.ca FamilySearch.org FindMyPast.com
SQLite_Tools_For_Roots_Magic_in_PR_Celti wiki, exploiting the database in special ways >>> RMtrix-tiny.png app, a bundle of RootsMagic utilities.


#8 zhangrau

zhangrau

    Advanced Member

  • Members
  • PipPipPip
  • 1518 posts

Posted 31 January 2016 - 06:45 PM

It's a terse and compact presentation, but I think it is seriously harmed by the repetitive occurrence of Birth Note. Death Note. Burial Note. when no actual note is presented. Maybe you can develop the script to clear those out?



#9 TomH

TomH

    Advanced Member

  • Members
  • PipPipPip
  • 6250 posts

Posted 31 January 2016 - 08:08 PM

Thanks, zhangrau. Those seemingly Note titles are actually just the text that I put in every note field to track where they come out. They are not part of the setup.

Tom user of RM7550 FTM2017 Ancestry.ca FamilySearch.org FindMyPast.com
SQLite_Tools_For_Roots_Magic_in_PR_Celti wiki, exploiting the database in special ways >>> RMtrix-tiny.png app, a bundle of RootsMagic utilities.


#10 zhangrau

zhangrau

    Advanced Member

  • Members
  • PipPipPip
  • 1518 posts

Posted 31 January 2016 - 10:14 PM

Ah, so just test notes ....



#11 Jerry Bryan

Jerry Bryan

    Advanced Member

  • Members
  • PipPipPip
  • 3566 posts

Posted 01 February 2016 - 08:56 AM

Ah, so just test notes ....

 

Yes, just test notes..

 

You can see the genesis of the ideas in this thread at http://forums.rootsm...-year/?hl=terse  where you can also see a link to a real report done somewhat in this style. Tom is taking the original idea even further, towards including more of the data in notes so as better to control the output from reports.  Also, I had to do all the sentence template configurations manually, and Tom is working on an SQLite script to automate the process.

 

No matter what is in your sentence templates and in your event notes, RM's reports provide a basic and useful structure or skeleton onto which dates, places, notes, etc. can be hung. Tom is stripping this structure or skeleton down to its bare essentials to see how it really works - hence, he has the test notes as place holders that don't say anything. But the test notes show where the real notes would be if they were there.

 

I think that Tom's analysis of this structure or skeleton is useful for any number of purposes, such as considering what would really be required to produce reports in languages other than English. In particular, those parts of RM's reports that are not under control of sentence templates would be a major problem for  producing reports in languages other than English.

 

Jerry



#12 genealogy4primm@earthlink.

genealogy4primm@earthlink.

    Advanced Member

  • Members
  • PipPipPip
  • 70 posts

Posted 10 March 2016 - 04:12 PM

Tom it looks good!

 

Jerry thanks to you also!

 

I would like these, but I do not post often enough to open that feature up.